PDA

View Full Version : Gear Hobber



Pages : [1] 2

jkopel
12-17-2010, 02:50 PM
Since gears are always a popular topic of polite conversation around here, I thought I might mention something I just stumbled across.
Some of you may keep better tabs on this then I do, but it looks like Martin Models is working on a the plans/castings for a gear hobber.

Check out the bottom of this page:
http://www.martinmodel.com/MMPtools-subfiles/MMPtools-sub.html

You can click through the picture for some more interesting images and tantalizing clues.

Josh

ehughes
12-17-2010, 05:28 PM
Somewhere in the basement is a set of the Helix castings, looks like I will soon be one version behind. Thanks for posting the link. Regards, Earl

lazlo
12-17-2010, 05:40 PM
Oh nice! That's the College Engineering Jacob's hobber castings.

By coincidence, Model Engineering Workshop has had a series of articles in the last three months about modifying the CES hobber to cut spiral gears.

The only issue I see with it is that it uses tiny little hobs, so you'd have to be able to cut your own. Not many hobby-class QCGB lathes can cut diametral-pitch threads...

oldtiffie
12-17-2010, 06:15 PM
I think that it should be said that if cutting a hob for straight (ie spur) gears the hob will be a spiral (ie like a thread or a worm).

The lead of the hobbing cutter will need to be the circular pitch of the required (DP?) gear adjusted for the helix angle of the hobbing cutter.

The lead of thread used to cut the hobbing cutter will therefore have a "pi" component in it.

Pi ~ 3.1416 or in gear ratio terms: 22⁄7, 333⁄106, 355⁄113.

macona
12-17-2010, 06:19 PM
That kit is made by Gary Martin who I used to work with at TechShop. All the kits have nice castings made locally, especially the dinky deere engine.

TexasTurnado
12-17-2010, 07:18 PM
The lead of thread used to cut the hobbing cutter will therefore have a "pi" component in it.

Pi ~ 3.1416 or in gear ratio terms: 22⁄7, 333⁄106, 355⁄113.

Or, more practically, a fraction of pi, and then taking this into account when positioning the QCGB. On my Colchester, the ratio is pi/4 and the H lever (vs the normal L) is engaged to add a x4 multiplier:

http://i288.photobucket.com/albums/ll168/TexasTurnado/PA280006.jpg

So, it should be possible to change the end gears on most any lathe with a QCGB to make it cut dp pitches....

oldtiffie
12-17-2010, 07:45 PM
Thanks TT.

My no-QCGB lathe (3mm lead-screw) has a "pi" (22/7) function.

For those that don't have it, it is easy enough to work our the required pitch to cut the worm on the lathe and the, seek out the nearest tpi/lead from the tpi (inch) and lead (mm) tables and set that ratio on the lathe. It is surprising just how accurate you can get.

There is another matter of the tooth form and the positioning/tilt (or not) of the lathe tool.

In an ideal case the teeth of the hob will need to be relieved - not an easy task as the greater the number of teeth on the hobbing cutter the closer the "faceted cuts" (one per tooth) will be to a smooth involute curve on the gear teeth.

uncle pete
12-17-2010, 07:56 PM
Nope, Sorry, Wrong info as that machine will use the smaller commercial hobbs. I have both the castings and the drawings from CES in the uk but am a few yrs. from building it. And to be truthfull that hobbing machine could be built using steel plate and some larger structual steel angle iron sections. With maybe some bolted on cast iron or bronze for the dovetail slides. It is a pretty complicated build tho. But commercial hobbs generly run about half the price of a full set of gear cutters to cut one D.P. plus for those that don't know, Hobbing is really fast and it's the most accurate way to cut gears. (using CNC? I don't know)

Pete

lazlo
12-17-2010, 07:58 PM
Nope, Sorry, Wrong info as that machine will use the smaller commercial hobbs.

Are you sure? The smallest commercial hob that I've ever seen is 2" long, with a 1 1/4" arbor.

Eyeballing the dimensions on the CES kit, that hob looks a lot smaller than that? Have you found a source for tiny hobs, or is that a 2" hob in the Gary Martin's picture?

macona
12-17-2010, 08:06 PM
The hobs on that machine are actually pretty small, maybe 3/4" OD. You can get them from Ash Gear.

The guy that actually built the patters is part of the Portland Model Engineer group. He has spent years working on it. I have a closer, bigger, picture of the unit at home. I will post it later.

uncle pete
12-17-2010, 08:09 PM
Lazlo,
Not 100% but 99%, I've been reading a few articals about this machine over the years and do recall some pictures in M.E. or M.E.W.? magazine showing a completed machine with some gears it had made along with both commercial and home built hobs. WHAT issue of that magazine it was in sure doesn't ring any bells, But maybe some of the uk readers here can either confirm or say I'm out too lunch on that 1%.

Pete

John Stevenson
12-17-2010, 08:10 PM
Arceuro does the smaller hobs but only in MOD form, they do the commercial ones in DP with a standard 1" bore.

uncle pete
12-17-2010, 08:16 PM
Thanks john, I was racking my brains trying to remember who in the UK had fairly reasonable priced hobs that would fit this equipment. Edited to add, I'm almost positive this hobber was designed to use Myford lathe gears to provide both the drive and the proper gear ratio for each different hob.

Pete

Robin R
12-17-2010, 09:57 PM
Here are the hobbs. http://www.arceurotrade.co.uk/Catalogue/Cutting-Tools/Gear-Hobs

lazlo
12-17-2010, 10:41 PM
Here are the hobbs. http://www.arceurotrade.co.uk/Catalogue/Cutting-Tools/Gear-Hobs

I'm really confused. Based on Robin's link, I'm guessing the Jacobs uses 8mm hobs?? So there are no DP (Imperial) hobs available for the Jacobs kit?
The DP hobs on that link are 22 mm (??).

I've got about a dozen commercial gear hobs of various sizes, and they all use a 1 1/4" arbor.

I dropped Gary an email asking him about availability of the hobs...

oldtiffie
12-18-2010, 12:06 AM
Originally Posted by Robin R
Here are the hobbs. http://www.arceurotrade.co.uk/Catalo...ools/Gear-Hobs


I'm really confused. Based on Robin's link, I'm guessing the Jacobs uses 8mm hobs?? So there are no DP (Imperial) hobs available for the Jacobs kit?
The DP hobs on that link are 22 mm (??).

I've got about a dozen commercial gear hobs of various sizes, and they all use a 1 1/4" arbor.

I dropped Gary an email asking him about availability of the hobs...

And this is where it can "interesting" or you can dig yourself into quite a hole.

If you make, buy or acquire the subject hobbing machine and don't have the hobs with the required 8mm bore (and diameter and length) you are going to have to seriously consider making them!!

And that is where "transititional screw-cutting" (with a "pi" or 22/7 element/component in it) needs to be addressed.

First of all, all gears with the same DP have the same circular pitch which will equal the axial pitch of the spiral of that DP hobbing cutter required to cut/hob those gears. They will and must also have the same pressure angle.

Example:
a gear with a Pitch Circle Diameter of 4" with 80 teeth has a diametral pitch (DP) of no. teeth/Pitch Circle Diameter (PCD) = n/PCD = 80/4 = 20DP.

The Pitch Circle Circumference (PCC) = PCD x "pi" = PCD x 22/7 = 4 x 22/7 = 88/7 = 12.57143".

As the 40 teeth are equi-spaced on the PCC the Circular Pitch (CP) of the teeth = PCC/80 = 12.57143/80 = 0.157143".

This will be the lead required to be screw-cut on the hobbing cutter.

A lead 0.157143" = 1/0.157143 = 6.3636 tpi (and is also lead of 0.157143 x 25.4 = 3.99mm).

First of all, a tpi of 6 may be beyond smaller inch lates with a say 12 tpi lead-screw - or a small lathe with a 1.5mm lead-screw, but most 8" or 10" lathes should be OK.

If you have a metric lead-screw and as a lead of 3.99mm is pretty close to 4mm you may think its near enough.

But a lead of 0.157143" (ie 6.3636 tpi) is not really an option.

It just so happens that pi/0.157143 = 3.1416/0.157143 = ((22/7)/(0.157143)) = 19.999982 ~ 20 which is a nice whole number to work into a gear train with 22/7 ratio gears for an "inch" lead-screw lathe.

I have not allowed for the "transitional" corrections for the cutting tool and the screw-cut pitch or lead for the hobbing cutter.

I have not allowed for the practical lathe requirements for generating/cutting the required "relief" (front clearance) of/on the gear/hobbing cutter either.

Making that hobbing machine may be quite a challenge but if you have to make the cutters (hobbing cutters) too, its going to be an even bigger challenge.

But as the hobbing cutter/s need to have an 8mm bore and only ArcEuroTrade in the UK has them, but not in DP format but only in metric module format, I guess that some are going have to learn to cope with "metric" and make (only) "module" years.

Irrespective of whether the gears are DP or module, providing that all meshing gears have the same DP or module the velocity ratios of the gear will remain the same as it is the ratios of the numbers of gears that count.

Here are a couple of pics that show all DP and Module relationships:
http://i200.photobucket.com/albums/aa294/oldtiffie/Black_book/BB_50-51.jpg

http://i200.photobucket.com/albums/aa294/oldtiffie/Black_book/BB_52.jpg

If you want to make it for show or self-satisfaction, that's fine, but if you want to make the hobbing cutters or use the 8mm bore metric module system using hobbing cutters from ArcEurotrade (UK) that's fine too, but buying DP hobbing cutters with 8mm bores seems to be quite a problem.

lakeside53
12-18-2010, 12:19 AM
.

Not many hobby-class QCGB lathes can cut diametral-pitch threads...


So... I've ignored the selectors on my QCGB that allow me to cut diametral-pitch threads, mainly becase I've never needed to, but worse, what the heck are they used for? I see a glimmer in this thread, but would someone care to explain the bigger picture?

oldtiffie
12-18-2010, 12:35 AM
Thanks Lakeside - good questions.

As I explained just a post or so ago, as the pitch of a hobbing cutter has to equal the circular pitch of the gears it is to cut, and as (pi = 22/7) that lead or Circular Pitch = ((Pitch Circle Diameter x pi)/(number of teeth in the gear)) there is a "pi" or 22/7 component in the lead of screw-cut to the hobbing cutter.

The settings you refer to are "transitional" threads and are as much related to DP gears as they are to Metric Modular gears.

But the gear-train or gear-box setting bring in a 22/7 gear ratio.

lakeside53
12-18-2010, 12:41 AM
That I read - one of the "glimmers" I refered to. :)

Is there any other purpose to all that table glued to my lathe than for making gear hobs? Clearly I'm hoping for more... lol

oldtiffie
12-18-2010, 01:02 AM
Perhaps.

Its never mentioned but if you are stuck for a particular thread or lead that is outside the normal scope of your QCGB you can get some interesting results if you use the "pi" setting either on its own or in conjunction with the "metric thread" 127 gear as well as the QCGB.

I did an example previously where the lead for the hob in inches was very close to 4.00mm (3.99 actually - an error of only 0.25% - which is not bad and would "work" pretty well without the "pi" gear even being used.

John Stevenson
12-18-2010, 05:30 AM
Lakeside,
The table on your lathe has been worked out to do DP and MOD ratios for doing work like worms that have to mesh with gear wheels.

The fact that they will do hobs is only because they conform to the same formula's

For anyone who doesn't have a QCGB or a table to do do DP ratios and a leadscrew pitch of 8 tpi, you can get away with having a 55 driving a 35 on the first banjo and then a second ratio to get the correct DP.

The 55/35 takes into account the Pi factor and the second ratio works out the pitch.

DP Ratio

8 - 60/30
10 - 40/25
12 - 40/30
14 - 40/35
16 - direct drive.
18 - 40/45
20 - 40/50
22 - 40/55
24 - 40/60
26 - 40/65
28 - 40/70
30 - 40/75
32 - 30/60
36 - 20/45
40 - 20/50
48 - 20/60
60 - 20/75

So to do say a 20 DP you need a 55 driving a 35 with a 40 on the same stud which drives a 50 on the leadscrew.

Similar to metric, you can't disengage the leadscrew.

For small hobs in imperial with an 8mm bore search on Ebay, not forgetting Ebay.co.uk and Ebay.de with gear hob in the search box.

http://cgi.ebay.co.uk/MINIATURE-GEAR-HOB-CUTTER-DP40-PA20-HELIX-1-40-MINS-/250618740933

This guy has loads.

JCHannum
12-18-2010, 07:49 AM
I'm really confused. Based on Robin's link, I'm guessing the Jacobs uses 8mm hobs?? So there are no DP (Imperial) hobs available for the Jacobs kit?
The DP hobs on that link are 22 mm (??).

I've got about a dozen commercial gear hobs of various sizes, and they all use a 1 1/4" arbor.

I dropped Gary an email asking him about availability of the hobs...

Ash Gear has DP hobs with 8mm bores also;

http://www.ashgear.com/pdfs/hm8-14.pdf

oldtiffie
12-18-2010, 09:14 AM
Thanks John S and Jim H for advising that DP hobs with 8mm bores are available.


Lakeside,
The table on your lathe has been worked out to do DP and MOD ratios for doing work like worms that have to mesh with gear wheels.

The fact that they will do hobs is only because they conform to the same formula's

For anyone who doesn't have a QCGB or a table to do do DP ratios and a leadscrew pitch of 8 tpi, you can get away with having a 55 driving a 35 on the first banjo and then a second ratio to get the correct DP.

The 55/35 takes into account the Pi factor and the second ratio works out the pitch.

DP Ratio

8 - 60/30
10 - 40/25
12 - 40/30
14 - 40/35
16 - direct drive.
18 - 40/45
20 - 40/50
22 - 40/55
24 - 40/60
26 - 40/65
28 - 40/70
30 - 40/75
32 - 30/60
36 - 20/45
40 - 20/50
48 - 20/60
60 - 20/75

So to do say a 20 DP you need a 55 driving a 35 with a 40 on the same stud which drives a 50 on the leadscrew.

Similar to metric, you can't disengage the leadscrew.

For small hobs in imperial with an 8mm bore search on Ebay, not forgetting Ebay.co.uk and Ebay.de with gear hob in the search box.

http://cgi.ebay.co.uk/MINIATURE-GEAR-HOB-CUTTER-DP40-PA20-HELIX-1-40-MINS-/250618740933

This guy has loads.

For those that missed it, 55/35 = 22/7 x 2.5/5 = 22/7 x 1/2

John Stevenson
12-18-2010, 09:36 AM
For those that missed it, 55/35 = 22/7 x 2.5/5 = 22/7 x 1/2

I didn't miss it, that's why I posted

"The 55/35 takes into account the Pi factor and the second ratio works out the pitch."

lazlo
12-18-2010, 10:26 AM
Ash Gear has DP hobs with 8mm bores also;

http://www.ashgear.com/pdfs/hm8-14.pdf

I was intentionally avoiding Ash. They're expensive as hell. Each involute cutter sells for ~ $140, so you're talking well over $200 for a hob.

lazlo
12-18-2010, 10:29 AM
The only issue I see with it is that it uses tiny little hobs, so you'd have to be able to cut your own. Not many hobby-class QCGB lathes can cut diametral-pitch threads...
...

For anyone who doesn't have a QCGB or a table to do do DP ratios and a leadscrew pitch of 8 tpi, you can get away with having a 55 driving a 35 on the first banjo and then a second ratio to get the correct DP

I know John got my point -- many QCGB lathes, including my Clausing, don't have banjos. So you can't add conversion gears to the gear train to cut a DP thread.

kc5ezc
12-18-2010, 02:05 PM
Guys: The original ME articles by T.D. Jacobs are in Jan thru about May 1976 issues. There are other small articles and letters throughout the year of 1976.
The Martin Models version is supposed to have some improvements over the original.
Tiff: One of the MEW recent articles indicated that 22/7 is not accurate enough for hobbing. The author recommended using 355/113.
I am reading and rereading al of this stuff trying to understand how one of Martin's kits will benefit my shop.
Ain't going to be cheap, even before you buy hobs.
Looks like a lot of learning and fun though.

uncle pete
12-18-2010, 02:15 PM
Lazlo,
Yeah your right, Well over $200 is expensive. But one hob will cut the full range of gears that a set of 8 blade type gear cutters will, Plus it's far faster and more accurate than setting up and cutting each tooth at a time.

Pete

oldtiffie
12-18-2010, 03:47 PM
Ash Gear has DP hobs with 8mm bores also;

http://www.ashgear.com/pdfs/hm8-14.pdf


Thanks Jim.

With a bore of 8mm (~ 0.32" ~ 5/16") and the largest OD being 1" and a DP range of 20>120 you are not going to cut any large gears.

The PA listed is 14 1/2* - are there any other PA's (ie 20* PA) or is that it?

But having said that, it has a maximum OD of 4" and a max DP of 20.

It is a very neat clever machine with all the features of a universal mill set up for hobbing included.

http://www.martinmodel.com/MMPtools-subfiles/GearHobber/gearhobber-02.jpg

http://www.martinmodel.com/MMPtools-subfiles/gearhobber-01.jpg

http://www.martinmodel.com/MMPtools-subfiles/MMPtools-sub.html

It should be quite a project and quite a challenge to take it on to make it.

John Stevenson
12-18-2010, 03:59 PM
http://www.stevenson-engineers.co.uk/files/martin%20gear%20hobber.jpg

Spot the deliberate mistake.

.

DannyW
12-18-2010, 04:38 PM
Spot the deliberate mistake.

.


Automatic powerfeed disengage?


Regards,

Danny

John Stevenson
12-18-2010, 04:49 PM
No, didn't noticed that, what I have spotted is a design flaw that will cut an incorrect gear.

Bob Farr
12-18-2010, 06:47 PM
I understand the need to position and articulate the cutter in relation to the gear blank, but once positioned why cant the hob helix simply drive the blank while cutting? Why the need for the banjo/driveshaft link between the hob and the blank?

uncle pete
12-18-2010, 07:12 PM
Bob,
That's called "free hobbing" and can be even done on a lathe with the gear blank able to freewheel. It's just not near as accurate as a gear rotated at the correct speed to match the hob and cut properly formed teeth. LOL, You now know everything I do about hobbing gears.

Pete

macona
12-18-2010, 09:03 PM
I was intentionally avoiding Ash. They're expensive as hell. Each involute cutter sells for ~ $140, so you're talking well over $200 for a hob.

They do have a "bargain basement" where they have used and other cheap cutters.

LES A W HARRIS
12-18-2010, 10:12 PM
Hob slide needs to swivel, not the part slide.

Cheers,

John Stevenson
12-19-2010, 06:52 AM
Hob slide needs to swivel, not the part slide.

Cheers,



Les has it, Choose anything off the bottom shelf Les. :D

ehughes
12-19-2010, 08:01 AM
Recall someone (the ledgendary D. H. Chaddock?) pointing out that there would be some error introduced because the the input & output of the universal joints were not parallel. Unless constant velocity u-joints were used, that is. Regards, Earl

ehughes
12-19-2010, 08:08 AM
Hob slide needs to swivel, not the part slide.

Cheers,



Does it really matter which one swivels, wouldn't the effect be the same? Or am I missing something? Regards, Earl

Spin Doctor
12-19-2010, 10:16 AM
Well, I was fooling around some time back and drew this up. Loosely based on a Barber-Colman

http://img.photobucket.com/albums/v19/markandannie/scan0001.jpg

One reason you need to have the hob to able to swivel is of course to adapt to the different lead angles for different pitches. Plus there is the aspect of helical gears. That is a whole different ball game

John Stevenson
12-19-2010, 10:42 AM
Does it really matter which one swivels, wouldn't the effect be the same? Or am I missing something? Regards, Earl

I'll leave Tiffie to explain it later.

lazlo
12-19-2010, 12:28 PM
Gary replied to my query -- the castings are finished, they're just debugging the machining/drawings. John: you might want to drop Gary an email about the hob slide! :)


"The castings are available now, just no drawings yet! We have a couple of guys machining these to discover any errors and finalize the drawings. So hopefully in a couple months all will be complete. "

I got a PM from someone who knows Gary who indicated that Ash gear was the source of the tiny 8mm DP hobs. I'll call tomorrow and see how much we're talking about.

oldtiffie
12-19-2010, 05:20 PM
I'll leave Tiffie to explain it later.

A good question John.

I will reply later as I need to look at it closer as you and Les seem to think there's an error or fault and the developer seems to think its OK.

I need to have time to go over it again.

I also have concerns about how the cutter is located and constrained on its 8mm arbor (key or friction) and how the job is constrained as well (friction? key? chuck? on its arbor) as the cutter and the job must not move relative to each other other than as driven/positioned by the hobbing process. If either the job or the cutter "slips" on its arbor the hobbing synchronisation will be lost.

It might not take much of a blow or force to bend the cutter 8mm arbor/spindle.

But never the less its a very nice concept and would be a nice project.

I'm surprised that we haven't heard from the "it needs ABEC 103 1/3 bearings and micron tolerances" brigade yet.

(A micron - 1 millionth of a metre or 1/1,000 of a millimeter is no big deal either as its only ~ 0.00004" or about "half a tenth" - or 40 millionths of an inch).

dp
12-19-2010, 05:55 PM
I thought the blank and hob were on the wrong shafts.

Bob Farr
12-19-2010, 06:08 PM
Well, I was fooling around some time back and drew this up. Loosely based on a Barber-Colman *** One reason you need to have the hob to able to swivel is of course to adapt to the different lead angles for different pitches. Plus there is the aspect of helical gears. That is a whole different ball game

Nice drawing Spin Doctor! Nicely done hand drawings (vs CAD) always seem more 'craftsmanlike' to me. Ok, back to the regular programming now.

Bob

TexasTurnado
12-19-2010, 07:47 PM
...


I know John got my point -- many QCGB lathes, including my Clausing, don't have banjos. So you can't add conversion gears to the gear train to cut a DP thread.

That's blasphemy! Who ever heard of a lathe without a banjo...? :) And how do you cut metric threads or anything not in the chart?

lazlo
12-19-2010, 07:50 PM
That's blasphemy! Who ever heard of a lathe without a banjo...? :) And how do you cut metric threads or anything not in the chart?

You don't.

TexasTurnado
12-19-2010, 08:02 PM
You don't.

But I thought I have seem metric gear kits for that lathe. Sounds like a opportunity for a project to add a banjo to me....:D

Is this not the correct one for your lathe:

http://cgi.ebay.com/Metric-change-gear-set-12-quick-change-Clausing-/300505364150?pt=LH_DefaultDomain_0&hash=item45f783f6b6

neonman
12-19-2010, 08:13 PM
I saw a lot of discussion about this really neat little machine, but didn't see
anything about all the change gears needed to make it versatile, or where to get them. Did I miss this? So far I have about 60 change gears for my Barber Coleman #3, and am still working toward a full set. Some can be made on the machine, but some need to be had before a duplicate can be made. Also, I didn't see a whole lot of discussion about making helical gears with this machine, can it be done?

neonman

lazlo
12-19-2010, 08:25 PM
But I thought I have seem metric gear kits for that lathe. Sounds like a opportunity for a project to add a banjo to me....:D

This is going to be a huge diversion for the thread, but the Clausing, like many other hobby-class lathes, have fixed rotation points for the quadrant gears.
Clausing takes that one step further, and doubles the number of feeds with a compound sliding gear engaging a double-wide leadscrew gear.

So not only are the quadrant gear points fixed, but if you made an adjustable banjo that allowed for adjustment of the gear centers, your leadscrew ratios would all be fubar because of the high/low gear ratio on the compound gear.

This was done to save an extra set of gears in the quick-change gearbox.


Is this not the correct one for your lathe:

http://cgi.ebay.com/Metric-change-gear-set-12-quick-change-Clausing-/300505364150?pt=LH_DefaultDomain_0&hash=item45f783f6b6

It is indeed. But as the Clausing instruction manual for the Metric conversion gearset explains, the quadrant rack (non-adjustable banjo) in that picture only works for Metric thread pitches. Once you convert the quadrant rack to metric, you can't thread imperial, and it's a major operation to change the quadrant rack, including removing the entire QC gear box!

TexasTurnado
12-19-2010, 08:37 PM
It is indeed. But as the Clausing instruction manual for the Metric conversion gearset explains, the quadrant rack (non-adjustable banjo) in that picture only works for Metric thread pitches. Once you convert the quadrant rack to metric, you can't thread imperial, and it's a major operation to change the quadrant rack, including removing the entire QC gear box!

I'll make you a really great deal on my cheapie Atlas 12 inch that not only has a banjo, but comes with a metric gear set too.... :D

lazlo
12-19-2010, 08:40 PM
I'll make you a really great deal on my cheapie Atlas 12 inch that not only has a banjo, but comes with a metric gear set too.... :D

That's the only design flaw on an otherwise superb lathe. Many others in that price/quality class have the same problem.

In any event, my HLV has an outboard quadrant -- that's one way to fix the problem :)

TexasTurnado
12-19-2010, 08:55 PM
That's the only design flaw on an otherwise superb lathe. Many others in that price/quality class have the same problem.

In any event, my HLV has an outboard quadrant -- that's one way to fix the problem :)

No only blasphemy,but asinine as well! Just to save one set of gears? Clausing should be publicly humiliated at the stake for doing such a thing in an otherwise good lathe. :D Do you have a manual for this lathe I can look at (or a link to such) that gives this sort of detail?

LES A W HARRIS
12-19-2010, 10:06 PM
eg. 20DP 20° PA hob 8mm bore. O D of hob 1.00", Pitch dia of hob about 0.876", 20 DP = pi/20 = 0.1571" axial pitch, or about 3.272° Lead angle.

If one swivels the hob the part passes straight along its own axis, producing a straight tooth, if one swivels the part, the part will cross the hob at the lead angle and do things, like widen tooth space, change the generated pressure angle and so on.

For the home shop, guess & by godding it, they should look goodish, will go round without thumping, [constant index], smoother than faceting cuts anyway. They may growl a bit, as the contact will probably be low.


Cheers,

John Stevenson
12-20-2010, 03:19 AM
I saw a lot of discussion about this really neat little machine, but didn't see
anything about all the change gears needed to make it versatile, or where to get them. Did I miss this? So far I have about 60 change gears for my Barber Coleman #3, and am still working toward a full set. Some can be made on the machine, but some need to be had before a duplicate can be made. Also, I didn't see a whole lot of discussion about making helical gears with this machine, can it be done?

neonman

It is designed to use Myford 20 DP change wheels which is OK for the original Jacob's / Brit design as Myford gears are cheap and easily available here but only in certain sizes , roughly working in multiples of 5's.

There is a series of articles running at the moment in both ME and MEW on converting one to do helicals.
Giles Parkes did the first one and he takes his machine to shows and demonstrates it cutting helicals.
Without studying it closely it requires a fair bit of rework.

Giles does use his machine a lot one of his recent jobs has been to cut the helical gears for a 1/4 scale model of a Ferguson tractor gearbox for another modeller. His machine can use larger hobs as he's borrowed 1" bore hobs off me in the past.

To be honest I feel that anyone who's interested in making a gear cutting machine such as this one is better off fabricating one up just using first principles and taking it from there.

The Martin one does had a design fault that Les has explained whilst Tiffie was still searching Tiffiepedia but couldn't find a page to scan.

Given the small size of the machine and it's corresponding lack of rigidity it also needs outboard support for both the hob and the work.
Giles has modified his so that it has provision for an overarm and support centre for the work.

uncle pete
12-20-2010, 04:28 AM
Thanks for the information John, I'll be sure to incorporate those ideas once I start building mine.

Pete

oldtiffie
12-20-2010, 05:34 AM
OK - here we go.

I think that the developer is correct and that John and Les are not.

http://www.martinmodel.com/MMPtools-subfiles/GearHobber/gearhobber-02.jpg

Let's sort out a couple of differences between hobbing plain (ie zero helix angle) spur gears - which this hobber is designed to do - and milling helical gears which this hobber is not designed to do - on a classic universal horizontal mill.

For gear hobbing the table is off-set by the helix angle of the hobbing cutter and the job spindle is indexed (ie is connected by a gear-train) to the mill cutter spindle/arbor. In short, the cutter spindle and the job spindle speeds are synchronised so that the job rotates at a reduced speed that is directly proportional to the number of teeth to be cut on the job and the hobbing cutter rotates one turn for each tooth on the gear.

In the case of spiral milling/hobbing the the mill table is off-set by the hobbing cutter helix angle +/- the job spiral/helix angle. In spiral milling using a single slot cutter (as per the usual set of DP and Modular gear cutters) the job spindle is gear-train coupled to and driven (via a dividing head or rotary table) the mill "X" lead-screw. The cutter spindle/arbor is not connected to the job spindle.

Cutting a spiral/helical gear with a hobbing cutter really does require a CNC-ed 4-axis set up as the non-CNC-ed machines that do this are well outside the scope of the usual HSM-er.

So, the gear hobber referred to in its current state and configuration is for hobbing plain (zero helix angle) gears.

It is not intended or able to mill or hob spiral gears.

But can it hob straight spur gears?

My answer is yes.

The milling cutter spindle/arbor stays static and the work spindle and the table it is on is moved/off-set left or right by and angle equal to the hobbing cutter helix angle so that the instantaneous line of cut of the cutter tooth is parallel to the job axis and the job axis is parallel to the mill table (ie "X").

The cutter and work spindles are correctly set up and connected physically be a pre-determined gear train.

Cutting feed is by hand - no power feed.

(The principal is the same as the gear-train connected lathe spindle and its QCGB/feed screw or lead screw).

The subject hobber meets all the normal criteria of a typical universal mill set up for plain gear hobbing.

The developer had done a fine job of his machine.

Given that it is on show regularly and is subjected to knowledgeable eyes as well as skeptics and "knockers" and still seems to more than adequately deal with all of that, I'd be loathe to pick holes in it - if there are any to pick.

John Stevenson
12-20-2010, 05:39 AM
OK - here we go.

I think that the developer is correct and that John and Les are not.



ROTFLMAO

Thanks Tiffie you have proved what I have suspected for some while

oldtiffie
12-20-2010, 05:44 AM
Yeah - we're like that in OZ.

DannyW
12-20-2010, 06:59 AM
Cutting feed is by hand - no power feed.



Ya-but, no-but, ya-but ...

What about the gear on the feed screw?
Looks like it can be slid in to engage power feed and will automatically
disengage before the end of travel is reached if the workpiece moves from right to left, as seen in the last picture. (the slide pushes it off.) Darn clever!

What say ye?

Regards,

Danny

willmac
12-20-2010, 07:07 AM
Tiffie-

We seem to have this discussion every few months.

You are lumping together hobbing with spiral milling. The two processes are fundamentally different although the result might look superficially similar in some special cases. The significant and fundamental difference is that with spiral milling you use a conventional milling cutter but hobbing requires a hob which generates the cut shape. The two cutters are entirely different and this results in different setup requirements.

I do agree that for a home workshop a CNC version built from first principles (as John has already suggested) makes more sense than building a desk-top hobber that has some design compromises (putting it mildly) or at the opposite extreme buying a full-sized hobber which would take up a significant chunk of floor space and using it once in a blue moon.

oldtiffie
12-20-2010, 07:13 AM
Bill.

I agree with you.

I thought I'd said that here:


In the case of spiral milling/hobbing the the mill table is off-set by the hobbing cutter helix angle +/- the job spiral/helix angle. In spiral milling using a single slot cutter (as per the usual set of DP and Modular gear cutters) the job spindle is gear-train coupled to and driven (via a dividing head or rotary table) the mill "X" lead-screw. The cutter spindle/arbor is not connected to the job spindle.

Cutting a spiral/helical gear with a hobbing cutter really does require a CNC-ed 4-axis set up as the non-CNC-ed machines that do this are well outside the scope of the usual HSM-er.

I hope that we agree as I certainly hope so.

willmac
12-20-2010, 07:26 AM
Tiffie -

No we don't agree. Your first sentence is the problem. You simply can't lump together spiral milling and hobbing. They are quite different processes.

oldtiffie
12-20-2010, 07:28 AM
Tiffie-

We seem to have this discussion every few months.

You are lumping together hobbing with spiral milling. The two processes are fundamentally different although the result might look superficially similar in some special cases. The significant and fundamental difference is that with spiral milling you use a conventional milling cutter but hobbing requires a hob which generates the cut shape. The two cutters are entirely different and this results in different setup requirements.

I do agree that for a home workshop a CNC version built from first principles (as John has already suggested) makes more sense than building a desk-top hobber that has some design compromises (putting it mildly) or at the opposite extreme buying a full-sized hobber which would take up a significant chunk of floor space and using it once in a blue moon.

Thanks Bill.

As I said, it would be a good challenge and a good project to make as a hobby job that would hob plain spur gears.

As for plain spur gears, the old-fashioned single spur gears in DP and module sets are hard to beat for the HSM-er in most cases.

Those cutters are quite accurate - especially so when clearance and back-lash are factored in.

I saw a table recently of comparisons of tooth width between hobbed and single-cut (DP sets etc) and it was remarkably small and consistent.

I'd like to see that table again.

Those cutters get a largely unjustified "bad rap".

They are my cutters of choice for spiral milling ('cept that I don't have a universal mill, dividing head (universal or compound) and gear-train etc. - yet.

They are pretty cheap from China - about US$20 for three or was it each but better than about US$200 in the USA - as I recall - recently and they worked very well - as I was informed by those that bought and used them.

John Stevenson
12-20-2010, 08:54 AM
Tiffie -

No we don't agree. Your first sentence is the problem. You simply can't lump together spiral milling and hobbing. They are quite different processes.

Bill, Forget it you are flogging a dead horse.

He can't find the Tiffiepedia link that explains the difference.

Perhaps he needs to SHOW us a gear he's cut......later. :D

lazlo
12-20-2010, 09:14 AM
I think that the developer is correct and that John and Les are not.
Thanks Tiffie you have proved what I have suspected for some while

LMAO! I have Tiffie on Ignore, but that's freaking hilarious!

A guy with a tenuous grasp (at best) on the concepts of gears and gear cutting, and he calls out the two most knowledgeable and experienced gearheads on the board :rolleyes:

lazlo
12-20-2010, 09:18 AM
Thanks for the information John, I'll be sure to incorporate those ideas once I start building mine.

Pete, I think John is saying that it's a design flaw with Gary Martin's prototype. If you have the College Engineering castings/drawings. Gary says the castings are complete, but they're debugging the drawings and build instructions.
Looks like they have a bug in the build, but if you have the College Engineering kit, you won't have that flaw.

By the way Pete, how did you get the castings in Canada? Did you buy them from the UK and have them shipped? How much was that?

djc
12-20-2010, 09:47 AM
If one swivels the hob the part passes straight along its own axis, producing a straight tooth, if one swivels the part, the part will cross the hob at the lead angle and do things, like widen tooth space, change the generated pressure angle and so on.

Sorry to labour the point, but I think you are incorrect. Yes, if one merely swivels the part, the tooth space will be widened but the photo displayed does not indicate this. The part always remains parallel/perpendicular to its axis of movement and the feed axis itself is swivelled to the lead angle of the hob.

The feed axis is attached to the vertical plane of the vertical slide and the pivot point appears to be on the horizontal plane (there are two unused spot-faced holes in the photo - these are misleading).

It is exactly analagous to the difference between a horizontal mill with a 'Universal' table (e.g. Schaublin 13, Deckel) where you can swivel the table but not the axis and a true/proper Universal mill where the table and axis swivel.

uncle pete
12-20-2010, 01:39 PM
Lazlo,
I wish Gary Martins hobber had been in development when I ordered mine. If the CES drawings are being debugged then I'll probably order a set of those too before I start on mine. I ordered my castings and drawings about 4 yrs. ago due to the fact that you never know when a casting supplier is going to fold up shop. I had my order shipped from the UK and I think ???? the cost was around $80. EVERYTHING from the Uk is expensive to ship. Heading back to work today but I'll be sure to check this thread when I get back in a couple of weeks.

Pete

oldtiffie
12-21-2010, 03:08 AM
Originally Posted by LES A W HARRIS
If one swivels the hob the part passes straight along its own axis, producing a straight tooth, if one swivels the part, the part will cross the hob at the lead angle and do things, like widen tooth space, change the generated pressure angle and so on.


Sorry to labour the point, but I think you are incorrect. Yes, if one merely swivels the part, the tooth space will be widened but the photo displayed does not indicate this. The part always remains parallel/perpendicular to its axis of movement and the feed axis itself is swivelled to the lead angle of the hob.

The feed axis is attached to the vertical plane of the vertical slide and the pivot point appears to be on the horizontal plane (there are two unused spot-faced holes in the photo - these are misleading).

It is exactly analagous to the difference between a horizontal mill with a 'Universal' table (e.g. Schaublin 13, Deckel) where you can swivel the table but not the axis and a true/proper Universal mill where the table and axis swivel.

Thanks djc.

I am still waiting for an answer to that.

If Les is not correct then John Stevenson isn't either - and neither is John's shadow and echo aka Lazlo.

John and Lazlo had lots to say and of course when John dried up so did Lazlo.

Roars of silence.

They are conspicuous by their absence.

But that aside, I am satisfied that hobber works as claimed by the developer - hobbing spur gears - nothing more and nothing less - no spiral/helical gears or bevel gears either.

Setting aside its value as a project, the question remains as to whether it will replace or be as good as or better than spur gears cut with a standard DP/Module gear cutter.

Possibly - but I doubt it.

J Tiers
12-21-2010, 08:53 AM
Youse guys have got me slightly confused........

1) if the relative angle between work and hob are the same in either case, independent of movements, then there can't be an argument about that, it drops out of the issue. Apparently that angle can be set, so presumably that isn't the issue if the angle is correctly set.

2) therefore the argument must be about the relative movement between hob and work. There are two basic parts.....

a) the rotational movement of work that generates the profile.

b) translational relative movement that extends the profile across the gear

Which movement are you saying is the issue? It appears that you are saying that the translational movement in this case will fail to properly extend the generated profile across the gear.

That would be true if the axis of movement were not in the correct relation with the angle between work and hob, in which case the movement would be at an angle with the "cut" the hob takes. The toothspace and form would be distorted.

Is that what you are saying?

Zahnrad Kopf
12-21-2010, 10:50 AM
A guy with a tenuous grasp (at best) on the concepts of gears and gear cutting, and he calls out the two most knowledgeable and experienced gearheads on the board :rolleyes:

I never understand why this happens, but it does. And it happens all the time, in a number of places (boards). Every time it does, I am tempted to join in the conversation to help, and maybe learn something during an exchange of ideas and information, when I am reminded of why I rarely do by someone exemplifying exactly what you observe... Then, I (and I know a lot of others in the same boat) sit back and view the comments for the comedic value... It should be mandatory that anyone commenting on the applications of gear manufacture actually have experience with it.

... going back to watching ...

djc
12-21-2010, 02:26 PM
Thanks djc.

<Other vacuous ranting cut>

Please don't try to assimilate me with or into your little feud.

One further useful point for discussion is if we look at the 'view from change gear end'

http://www.martinmodel.com/MMPtools-subfiles/GearHobber/gearhobber-02.jpg

What is the purpose of having the gear-blank-carrying-spindle able to swivel relative to its feed axis (evidenced by the curved slots)?

As far as I can see, if the spindle axis is anything other than parallel to the leadscrew axis, the depth of cut will vary as the blank is traversed across the hob. If the swivel point was _behind_ the leadscrew, it might be a useful feature as it would allow larger gears to be cut - see principle here:

http://www.lathes.co.uk/myford/img83.gif

ehughes
12-21-2010, 03:53 PM
http://www.martinmodel.com/MMPtools-subfiles/GearHobber/gearhobber-02.jpg

What is the purpose of having the gear-blank-carrying-spindle able to swivel relative to its feed axis (evidenced by the curved slots)?


It appears that the spindle does not swivel relative to the feed axis, but rather that they swivel together on the same assembly. Regards, Earl

oldtiffie
12-21-2010, 06:44 PM
I've gone over the gear hobber and reached the same conclusion.

In my opinion, it will not only work (design) and does work (the Developer has had it going successfully under critical eyes at public shows).

I have no problem with any who have opinions different to mine - as is their right - and mine.

http://www.martinmodel.com/MMPtools-subfiles/GearHobber/gearhobber-02.jpg

http://www.martinmodel.com/MMPtools-subfiles/gearhobber-01.jpg

I was intrigued to see that the gear train is a compound type able to cope with a wider range of tooth numbers on the gear as well as directional reversal and that cutters with left and right handed helixes can be catered for.

I don't know if the motor is reversible nor do I know if the motor has variable speed or if stepped pulleys are an option. Perhaps the motor is DC running under a variable speed control.

I can see no reason to think that the gears cut/hobbed will be much if any less accurate that those hobbed on a universal mill with a universal dividing head.

The gears should be able to be measured with gear calipers, shop calipers and disc micrometers etc. The data tables are in Machinery's Handbook. Clearance and back-lash can be incorporated and measured as well.

It would be a good project as it will require a lot of skills. It can be scraped and otherwise embellished or finished to suit the maker of it.

I was intrigued to see the Quorn Tool and Cutter Grinder as a kit on the same site as both work and they would be nice working show-pieces. The Quorn should be well able to (re?)sharpen the hobber cutters so they should work well together.

http://www.martinmodel.com/MMPtools-subfiles/quorn-01.jpg

http://www.martinmodel.com/MMPtools-subfiles/MMPtools-sub.html

I have checked the differences between the chordal thickness of "basic" (hobbed) gears and that of the standard DP/Module sets of gear cutters and while the first in each list is the same as the "basic" gear, there is very little difference really at the last cutter in the range. Given that clearance and back-lash have to be allowed for, gears cut with the normal gear cutters will be very close to hobbed gears.

I think that for most work in a HSM shop that the gears cut with standard cutters will do very well. Those cutters, while less accurate than a hobbing cutter, do very well, are cheap, and only need a dividing-head or a rotary table on a vertical mill to cut pretty satisfactory spur gears.

But back to the hobber.

It will hob gears with a 4" maximum outside diameter.

As to whether it works or not and given that seeing is believing (or not), I suggest that some contact the developer and ask where and when the next public demonstration is and go and see it for themselves or get someone else to do it for them.

I think it works as stated.

TexasTurnado
12-21-2010, 07:42 PM
EV = entertainment value :D

I'm still waiting for someone to explain why the hob and blank are interchaged between the website photo and the three view one posted here....;)

JCHannum
12-21-2010, 08:07 PM
All the pics are from the website. Click on the photo of the hobber on the brochure to see the in process photos. All show the hob on the fixed spindle.

Not only do John and Les feel the hob should be angled, Colvin and Stanley agree. I would tend to agree with those in the know.

Perhaps the operation is different south of the equator. Since the bathwater goes down the drain other way around does gear cutting follows suit?

Gary Martin is a good guy and an excellent pattern maker. He is doing a great service to those of us in the colonies by making some of the Brit kits available. I don't know all of the background of the design of the hobber his castings are derived from, but it appears that it would be a simple matter to put the hobbing gear on a subplate that could be swiveled to the appropriate angle. This would possibly open up the ability to do bevel gears.

oldtiffie
12-21-2010, 08:07 PM
Good question TT.

I wondered about that too.

The hob should rotate one turn for each tooth on the gear so there should be a speed/rpm reduction between the cutter (fastest) and the job (slowest).

Even if the gear train ration was 1:1 the hob spindle is (should be) driven by the motor and the job spindle is driven by the "wheel" in the worm/wheel reduction drive.

If the set-up in the web site is correct - as it seems to have had the cutter doing some cutting (see the swarf) - I guess I don't have an answer.

If the previous three pics are correct then the fourth pic (from the web) site may be wrong.

If the pic from the web site is correct then the previous/first three pics are wrong - as would I be.

I really don't have an answer that will reconcile all of the pics.

I will think about it later.

lazlo
12-21-2010, 08:07 PM
http://www.stevenson-engineers.co.uk/files/martin%20gear%20hobber.jpg

Spot the deliberate mistake.

For comparison, here's the British Jacobs/Helix Gear Hobber (from last month's MEW):

http://i164.photobucket.com/albums/u15/rtgeorge_album/jacobs.jpg

oldtiffie
12-21-2010, 08:18 PM
All the pics are from the website. Click on the photo of the hobber on the brochure to see the in process photos. All show the hob on the fixed spindle.

Not only do John and Les feel the hob should be angled, Colvin and Stanley agree. I would tend to agree with those in the know.

Perhaps the operation is different south of the equator. Since the bathwater goes down the drain other way around does gear cutting follows suit?

Gary Martin is a good guy and an excellent pattern maker. He is doing a great service to those of us in the colonies by making some of the Brit kits available. I don't know all of the background of the design of the hobber his castings are derived from, but it appears that it would be a simple matter to put the hobbing gear on a subplate that could be swiveled to the appropriate angle. This would possibly open up the ability to do bevel gears.

Good reply Jim.

Here is the link to the pics on the web site that you refer to:
http://www.martinmodel.com/MMPtools-subfiles/GearHobber/gearhobber-02.jpg

This is the link to the front page of the web site:
http://www.martinmodel.com/MMPtools-subfiles/MMPtools-sub.html

The cutter and the job are transposed.

As all the pics are on the web site and if only one set up is correct - which is it?

And if all are correct - how is it explained?

Despite what some may think, I have quite an open mind on this and just want to see the matter resolved.

If I am wrong - then so be it.

But at this stage I (still) consider the three pics of the same set-up with the cutter driven by the motor belt-drive and the work driven by the gear-train and the worm reduction drive.

lazlo
12-21-2010, 08:23 PM
Gary Martin is a good guy and an excellent pattern maker. He is doing a great service to those of us in the colonies by making some of the Brit kits available. I don't know all of the background of the design of the hobber his castings are derived from, but it appears that it would be a simple matter to put the hobbing gear on a subplate that could be swiveled to the appropriate angle.

Jim, I mentioned earlier in the thread that Gary replied and indicated that they're still debugging the prototype castings, which are shown in his pictures. I replied with John/Les' comments.

It looks like Gary's prototype has a couple of degrees of helix offset, but radically less than the British Jacobs/Helix. Gary's prototype also doesn't have the translation offset (the top dovetail slide), and as Danny pointed out, it doesn't have the powerfeed stops (which surely would be one of the last things you'd add anyway).

oldtiffie
12-21-2010, 08:29 PM
http://www.stevenson-engineers.co.uk/files/martin%20gear%20hobber.jpg


For comparison, here's the British Jacobs/Helix Gear Hobber (from last month's MEW):

http://i164.photobucket.com/albums/u15/rtgeorge_album/jacobs.jpg

Thanks Lazlo.

The principles of both are the same.

The cutter spindle is fixed and the job head/spindle is rotated an amount equal to the hobbing cutter helix angle.

In this pic from the web site, the cutter and job are transposed - why?

http://www.martinmodel.com/MMPtools-subfiles/gearhobber-01.jpg

JCHannum
12-21-2010, 08:52 PM
In this pic from the web site, the cutter and job are transposed - why?

[img]http://www.martinmodel.com/MMPtools-subfiles/gearhobber-01.jpg

They are not, it is a smaller blank, coated with red Dykem and smaller hob. Both are in the same positions as the other photos.

Robin R
12-21-2010, 08:55 PM
I think if you look closely at the photo of the painted gear hobber, you will see that the tool spindle does in fact have a hob on it and the work spindle has nothing on it.

I have yet to see an explanation as to why the hobber won't work as built. The work spindle slide can obviously be swivelled to a limited degree, though I'm sure it's enough to account for the helix of the hob. It isn't designed to cut helical gears, as the modified one on the cover of MEW obviously is. It would be helpful if any of the posters who say it won't work, could explain why that is the case.

I am quite willing to accept that JCHannum is correct about the work spindle having a small blank on it, a larger image would make that clearer.

oldtiffie
12-21-2010, 08:58 PM
Many thanks Jim - really.

It is so obvious when you pointed it out.

I had a real "head-banger" and "oh $hit yeah" moment when I read your post.

In that case all pics have the same positions - cutters on the fixed and faster spindle and the job on the slower left-right adjustable spindle.

All the pics are singing from the same hymn book.

lazlo
12-21-2010, 09:26 PM
The work spindle slide can obviously be swivelled to a limited degree, though I'm sure it's enough to account for the helix of the hob. It isn't designed to cut helical gears, as the modified one on the cover of MEW obviously is.

The only modifications for cutting helical gears was an additional bevel geartrain and adding more longitudinal travel to the hob slide.

In other words, that dovetail and 30° helix angle swivel mount is the basic Helix/Jacobs design, as you can see in this unmodified version from Harrogate. Gary's prototype looks like it only swivels ~ 3 - 5° degrees, and doesn't translate at all.

http://i164.photobucket.com/albums/u15/rtgeorge_album/Jacobs2.jpg
http://www.stevenson-engineers.co.uk/files/martin%20gear%20hobber.jpg


Edit: Scratch the lack of translation -- I see what Gary did: he put the translation dovetail on the back-side of the angle block.

oldtiffie
12-21-2010, 10:26 PM
I think if you look closely at the photo of the painted gear hobber, you will see that the tool spindle does in fact have a hob on it and the work spindle has nothing on it.

I have yet to see an explanation as to why the hobber won't work as built. The work spindle slide can obviously be swivelled to a limited degree, though I'm sure it's enough to account for the helix of the hob. It isn't designed to cut helical gears, as the modified one on the cover of MEW obviously is. It would be helpful if any of the posters who say it won't work, could explain why that is the case.

I am quite willing to accept that JCHannum is correct about the work spindle having a small blank on it, a larger image would make that clearer.

Thanks for the further clarification Robin. Makes sense.

LES A W HARRIS
12-22-2010, 01:10 PM
Getting ready to depart for Xmas, Seasons best everyone.

The Hob/Part swivel thingy, I see that the Part & it's Slide is swiveling, thus the part will travel along its own axis, not skewed across it's axis as I 1st viewed it, which would result in a transverse generated form.
although at these small pitches & face widths it would be a small mutilation & possibly overlooked.

So, I DO NOT NOW SEE A FLAW IN THE DESIGN, sorry.

.RC.
12-22-2010, 03:23 PM
That is what had me confused on what the fuss was about, that horizontal slide is angled as well, so you are dragging the work across the hob at the angle you set it at, which you have to do anyway to produce a spur gear due to the helix angle of the hob...

larger commercial machines impart all movement to the hob probably because it is far easier to design and build a machine that way..

oldtiffie
12-22-2010, 04:06 PM
Getting ready to depart for Xmas, Seasons best everyone.

The Hob/Part swivel thingy, I see that the Part & it's Slide is swiveling, thus the part will travel along its own axis, not skewed across it's axis as I 1st viewed it, which would result in a transverse generated form.
although at these small pitches & face widths it would be a small mutilation & possibly overlooked.

So, I DO NOT NOW SEE A FLAW IN THE DESIGN, sorry.




Thanks Les - appreciated.


That is what had me confused on what the fuss was about, that horizontal slide is angled as well, so you are dragging the work across the hob at the angle you set it at, which you have to do anyway to produce a spur gear due to the helix angle of the hob...

larger commercial machines impart all movement to the hob probably because it is far easier to design and build a machine that way..

.RC. aka Ringer:
perhaps so, but the machine/hobber very closely imitates/emulates a common universal mill where the job axis is aligned to the table, the table off-set by the cutter helix angle and the job driven by a universal or compound dividing head which is driven by a gear train that in turn is driven by the hobbing cutter arbor.

That is to say, the the cutter and the job are directly linked by gearing and as the gearing is driven by the cutter (arbor) and the dividing head is driven by the gearing, the job is indexed to/by the cutter.

The longitudinal ("X") table feed may be either powered or manual and theoretically can stop or start (but not reverse) at any point.

Assuming a hobbing cutter with a single-start spiral ("thread"), the job will rotate one "tooth" (ie a part of a revolution) for each revolution of the hobbing cutter.

The gear reduction - including the normally 40:1 internal worm drive reduction in the dividing head - ie the job - between the cutter (arbor) and the job (dividing -head) is equal to the number of teeth on the job.

For example, for a 120 tooth gear the cutter will make one turn for each gear tooth = 360/120 = 3 degrees of dividing head rotation.

There will be as many "facets" on the flanks/faces of the "involute" curve of the gear as there are numbers of teeth on one pitch or lead of the cutter.

I think that the hobber - including the one in the magazine cover pics as posted by Lazlo - are very clever - and they will work.

I will post some info on the the chordal thickness of comparable "basic" (ie hobbed) gears and the cutter chordal thicknesses for each cutter in a set later.

It may surprise some as to just how close they are and just how good the "cutter sets" are for general work in the HSM shop. They are not as "rough" or as inaccurate as some may have you believe.

So, a hobbing machine may be neither needed or necessary for most or many gears in a a HSM shop.

It certainly is not needed in my shop.

Others may vary.

.RC.
12-22-2010, 04:35 PM
So, a hobbing machine may be neither needed or necessary for most or many gears in a a HSM shop.



But in a HSM situation when has need ever come into the equation?

Some people like building model engines, some model planes, some like building HSM sized machine tools.. Simply because the building is as enjoyable as using the finished product..

I personally look at some tool I have made from raw materials and get a warm feeling inside of me knowing I made it myself...

Just like a chef would do when creating some culinary masterpiece..

Spin Doctor
12-22-2010, 04:43 PM
Yes involute cutters will work in a HSMers shop quite well. After all we are not trying to build super quite running gear trains.* Heck even the approximate hobbing method will work. There are a lot of ways to skin this cat. But if you have never run a hob it seems almost magical as the swarf falls away from the cutter and the tooth form swings into view as the work rotates. Having cut more gears and splines than I really want to think about (Spur, internal and external, Helical, external only doing my own index and feed calculations, and Splines, parallel key and involute, internal and external) it still gave a great deal of satifaction cut toothed forms in the Hob or Gear Shaper. Plus there is one aspect of hobbing/shaping that should not be overlooked. The tooth form is never shifted off axis simply because of the operating function of the machines involved. The only thing I can see I don't like about the hobber shown in the photos is the lack of capacity to cut splines on longer shafts.

*When I worked for AMC before they were absorbed by Chrysler for a while they had a contract tfrom Ford to supply rear end sets for the 8 inch rear end IIRC. When the first gear sets came into the final inspection area for checking contact patterns and such they were set-up on special inspection machines that would run the gear set at the proper lash adjustment for a period of time. These Ford sets were really noisy(they specified that the gear sets would not be lapped). The engineer from Ford thought they sounded really good. The inspector started the machine next to that one with one of the gear sets we made and his remark was they don't need to be that quiet (the machines drive made more noise than the gear set being run and the set was running in the open). All you need to do is add more sound deadner. And that is just one reason I will never buy a Ford.

oldtiffie
12-22-2010, 05:22 PM
Originally Posted by oldtiffie
So, a hobbing machine may be neither needed or necessary for most or many gears in a a HSM shop.


But in a HSM situation when has need ever come into the equation?

Some people like building model engines, some model planes, some like building HSM sized machine tools.. Simply because the building is as enjoyable as using the finished product..

I personally look at some tool I have made from raw materials and get a warm feeling inside of me knowing I made it myself...

Just like a chef would do when creating some culinary masterpiece..


But in a HSM situation when has need ever come into the equation?

In my shop - every time.

I go back to the basic requirement as regards jobs, tools and methods and work upward from there - and then I stop.

Once I've got to that stage, I know the job can be done.

Any "extra" is more in the "what I want" category than it is in the "what I need" category.

My jobs get only what is needed but they get all that is needed.

I am satisfied if and when the job does what it is intended for or to do. No more and no less.

I enjoy my work.

Quite often when I have got to the stage where I can see whether a concept will work or not I stop. The job may be kept but it will probably be consigned to stock or scrap.

I have no personal or emotional attachment to jobs, tools or the shop.

I have no issue with any who want to "go the extra nine yards".

.RC.
12-22-2010, 05:34 PM
In my shop - every time.



Yes but you are commercial, not a HSM... Your shop is primarily to make money, not a hobby...

oldtiffie
12-22-2010, 05:45 PM
Nope.

You've got that wrong RC.

Could not be further from the facts or the truth.

When I retired and really got into setting up my shop, I promised myself that I would never "go commercial" or do any "charged work". Its been that way for over 13 years and I can see no reason to change.

I also promised myself that if I didn't want or need my shop or if I could not afford it - it would all go straight to scrap - and I would never miss or regret it.

My shop is only a means to an end. It is not an end in itself. As soon as it ceases to be that means to that end - it goes - finish.

oldtiffie
12-22-2010, 05:49 PM
Here is a Cincinnati horizontal universal mill set up for gear hobbing:

http://www.practicalmachinist.com/vb/attachments/f38/3424d1208843859-converting-horizontal-mill-gear-hobber-gear_hobbing_013.jpg

from:
http://www.practicalmachinist.com/vb/879522-post8.html

.RC.
12-22-2010, 05:55 PM
Nope.

You've got that wrong RC.

Could not be further from the facts or the truth.



Oh OK, I got the impression from your posts that you did commercial work still... Did not realise that your shop is purely hobby..

dp
12-22-2010, 06:05 PM
I am quite willing to accept that JCHannum is correct about the work spindle having a small blank on it, a larger image would make that clearer.

It appears to me on the work spindle it is a spline on a shaft being cut and from which a gear will be parted off.

willmac
12-22-2010, 06:14 PM
Building a tool represents an investment in money and time, and for many of us time is often the most precious factor. If I decide to build a tool, I need to have real confidence in the design; otherwise it feels like a waste of time that can't be recovered. I am most reluctant to spend months building a tool that seems dubious or compromised in design. In the case of the hobber we are considering I don't doubt that gears can be produced, but I am not sure that the quality of the resulting simple spur gears would be significantly better than can be produced by other methods, so to that extent, I do agree with Tiffie, although maybe for different reasons.
The functional reasons for building a hobber such as this might be :
1) To minimise costs of multiple cutters needed for making many gears with varying numbers of teeth;
2) To produce higher quality gears;
3) To produce gears faster;
4) To produce a wider range of gear types.
I would suggest that this design will meet at most one or maybe two of these requirements.
Of course you might want to build it because it will be an interesting project and this is a good non-functional reason to do so.
I think the design is compromised because:
1) The support for the hob and blank is single ended. This makes the design simple, but results in a very large loss in rigidity.
2)The hob and work arbors are small in relation to the maximum gear diameter which is apparently 4 inches. This again will lead to poor rigidity. I would be interested to see a 4” steel gear being hobbed on this machine.
3)The double U joint drive is not a good choice for a mechanism that requires precise angular synchronisation of hob and blank. I would expect this to introduce cyclical errors in the gear being hobbed.
4)The overall design looks ‘wobbly’ for want of a better description. The brackets supporting the gear train are a case in point.
So after being negative about the design what would I do if I wanted to hob gears in my workshop?
I think (as has already been pointed out) that a much less compromised design could be built based on direct drives and digital controls. This would eliminate most of the gear trains and the U joints and enable much more solid arbour support. This type of design could possibly be built more quickly and easily than the machine we are reviewing. We have seen some examples of these in this forum in the past. So in terms of a better investment of time available I think that this would be the better choice if you value your time and satisfaction.

oldtiffie
12-22-2010, 06:32 PM
Thanks Bill.

You make your case and your points very well.

oldtiffie
12-22-2010, 06:34 PM
Most or many here have a dividing head or a rotary table with or without chucks and tail-stock centres and with an adaptor in the milling head spindle can make pretty passable gears with standard gear cutters.

Those cutters are NOT expensive as there are some very good HSS ones available from China.

You don't always need the exact cutter either as there is not a lot of difference between cutters in the same set. Making the cut deeper or shallower and/or taking a bit more left and right will get you very close and will do for most applications - cast iron and some steel lathe gear train gears included.

Its quite amazing as to just how much wear and abuse a gear couple or train will take and still seem to work well enough.

I will try to remember to cover some of this later when I get into the gear cutter tables.

This is not meant to be taken or read as saying that people should not make the hobber at all.

macona
12-22-2010, 06:54 PM
http://www.martinmodel.com/MMPtools-subfiles/gearhobber-01.jpg

This is the newest pic of the gear hobber. I replaced the thermocouple connector shown in the other pics with a 2 pin molex.

oldtiffie
12-25-2010, 02:13 AM
This is by way of a "bump" so that I see it and don't forget it.

I will go over the accuracy of the standard 8-cutter involute gear cutter sets shortly.

Some might be quite surprised at just how close they are to the hobbed gears.

John Stevenson
12-25-2010, 05:41 AM
Oh OK, I got the impression from your posts that you did commercial work still... Did not realise that your shop is purely hobby..

It's not even hobby.
Tiffie has NEVER posted a picture of one part he's made in this shop.
All the pictures posted have been posed shot with unworn and unmarked equipment.

lakeside53
12-25-2010, 12:14 PM
It's Xmas... I'm wishing for "unworn and unmarked equipment".

Wait... I've been really good this year; I'm sure that if I wander down to my shop, it's all like new now. :rolleyes:

oldtiffie
12-29-2010, 01:03 AM
This is another "bump"

I'm not finished with this thread yet.

http://www.martinmodel.com/MMPtools-subfiles/gearhobber-01.jpg

I will deal next with the limits of the smaller hobbing cutters required for this hobber. The hobs are typically 1" diameter. 3/4"+ long and with an 8mm (~0.315") bore. The hobber job blank maximum diameter is 4" but I will limit myself to 3".

The hobber, not surprisingly, is limited to about 25DP (Metric Module 1 = 25.4DP).

I will move on later to shop-cut gears using the standard "DP" gear cutter sets. It surprising just how close they are to a "basic" gear (tooth) as might be cut with a hobbing cutter set up on a universal mill. They are plenty good enough for most HSM shop work.

I'd opt for the normal DP cutter sets every time unless there was compelling reasons why I should not.

oldtiffie
12-29-2010, 03:36 AM
http://www.martinmodel.com/MMPtools-subfiles/gearhobber-01.jpg

These are the hobbing cutters to suit the 8mm spindle on the hobbing machine:
http://www.arceurotrade.co.uk/Catalogue/Cutting-Tools/Gear-Hobs

Note that the hobs are metric (ie Modular = 25.4/DP), 25mm (~1") diameter, 15mm (~0.60") or 20mm (~0.80") long have an 8mm (~0.32") bore and are limited to Module 0.3>1.0 Module in 0.1 Module steps and 1.25 Module.

These are quite small as DP = 25.4/Mod so Mod 1.0 = 25.4DP , Mod 0.5 = 50.8DP, Mod 0.1 = 254DP

So for a Mod1 75mm PCD metric gear the gear will have 75 teeth.

Or to approximate it in the usual DP gear format, that 3" gear with 75 teeth will have a DP of 75/3 = 25DP.

The depth of cut = 2.25/DP = 2.25/25 = 0.090"

The tooth chordal thickness (straight line between the pitch points on either/both the gear tooth and the milled slot is:
Tc = PCD Sin (90/No teeth) = 3 sin (90/75) = 3 Sin 1.2 = 25 x 0.0209 = 0.0628".

So those largest gears in terms of diameter and numbers and width of teeth are pretty small compared to those that are normally made in a HSM shop.

That being so, there are only two options left for larger gears:
1). hob the gears with a hobbing cutter on a universal mill set up with the work spindle geared and indexed to the cutter spindle; and

2). use the standard set of DP milling cutters (8 cutters for each DP).

Option 2 is the more likely in a HSM shop.

I will have a look at the inch/DP 8.0mm bore hobbing cutters for the hobbing cuter (see above) later.

Those small hobbing cutters (14 1/2* PA) with a maximum DP of 24 were posted by Jim Hannum:

Ash Gear has DP hobs with 8mm bores also;

http://www.ashgear.com/pdfs/hm8-14.pdf

I will then have a look at the standard hobbing cutters compared to the standard 8-cutter DP milling sets.

Later.

DICKEYBIRD
12-29-2010, 08:00 AM
Gear 'Obs....David 'Obs's (the reknowned Formula 1 commentator) machinist savant half-wit brother innit??:D

J Tiers
12-29-2010, 08:39 AM
The Hob/Part swivel thingy, I see that the Part & it's Slide is swiveling, thus the part will travel along its own axis, not skewed across it's axis as I 1st viewed it, which would result in a transverse generated form.
although at these small pitches & face widths it would be a small mutilation & possibly overlooked.

So, I DO NOT NOW SEE A FLAW IN THE DESIGN, sorry.


Ah, I just saw this.... had this thread on "ignore" for a while.

That brings the expert opinion in line with what I saw & thought. I was starting to think I had missed a basic issue.

lazlo
12-29-2010, 11:36 AM
Spot the deliberate mistake.


The Hob/Part swivel thingy, I see that the Part & it's Slide is swiveling, thus the part will travel along its own axis, not skewed across it's axis as I 1st viewed it, which would result in a transverse generated form.
although at these small pitches & face widths it would be a small mutilation & possibly overlooked.

So, I DO NOT NOW SEE A FLAW IN THE DESIGN, sorry.

You're sorry that there's no design flaw? :p Just teasing Les!

I think Sir John noticed that the dovetail slide and swivel plate on the British Helix/Jacobs version is missing from Gary's castings, and made that comment.

I mentioned earlier that Gary moved the translation dovetail to the back of the angle bracket, but I'm still puzzled by the extremely small helix angle adjustment?
Gary's castings look like they only provide a couple of degrees of helix angle, while the Helix/Jacobs provides 30°.

By the way, Gary replied back that the castings are complete and available for sale at $495 plus shipping. The builder (not Gary) is still debugging the (Imperial) drawings/build instructions.

oldtiffie
12-29-2010, 03:16 PM
I will press on with the 'merican??? DP (ie "inch") 8mm bore hobbing cutters for the hobbing machine - pic posted previously.

I will use the file and image in the link posted by Jim Hannum.


Originally Posted by JCHannum
Ash Gear has DP hobs with 8mm bores also;

http://www.ashgear.com/pdfs/hm8-14.pdf

http://i200.photobucket.com/albums/aa294/oldtiffie/Gear_theory/8mm-DP-hob1.jpg

Note that the depths of cut given - ie DOC = 2.2/DP + 0.002" - are slightly different to the usual 2.157/DP which in the case of the 24DP cutter on the 24DP cutter (top of page) would be: DOC = 2.157/24 = 0.090" which is 0.0037" less than the 0.0937" listed for the 24DP cutter at the top of the page. This may apply to all the hobbing cutters.

Back to the 3" PCD blank used previously.

If that blank had a 24DP the number of teeth will be 3 x 24 = 72 teeth.

The chordal thickness of the teeth will be: PCD x Sin (90/number of teeth) = 3 x sin (90/72) = 3 x sin 1.25 = 3 x 0.0218 = 0.0654"

This is a pretty small gear compared to the normal 3" PCD gear cut in the HSM shop with standard DP cutter sets.

The further down the DP column the smaller the gear teeth will be in terms of depth of cut and chordal thickness.

Some may have a real use for hobbed gears that are that small but anything larger will need to be hobbed on a universal mill set up for hobbing (least likely) or by using the standard DP gear-cutter sets (more/most likely).

I will cover the DP cutter sets next.

Later.

Zahnrad Kopf
12-29-2010, 03:21 PM
Note that the depths of cut given - ie DOC = 2.2/DP + 0.002" - are slightly different to the usual 2.157/DP which in the case of the 24DP cutter on the 24DP cutter (top of page) would be: DOC = 2.157/24 = 0.090" which is 0.0037" less than the 0.0937" listed for the 24DP cutter at the top of the page. This may apply to all the hobbing cutters.

Nope. That's wrong. Do you/Have you hobbed much? Thanks.

oldtiffie
12-29-2010, 03:27 PM
Yes - quite a lot quite some years ago.

These hobbing cutters are 14 1/2* PA

Normally:
Depth of cut:
20* PA = 2.25/DP

14 1/2* PA = 2.157/DP

Edit:

Machinery's Handbook 27 Page 2039 Table 4 shows "Fine Pitch" involute gears (14 1/2*, 20* and 25* PA) having a minimum depth of cut = 2.2/P + 0.002" (minimum).

I prefer:
http://i200.photobucket.com/albums/aa294/oldtiffie/Black_book/BB_50-51.jpg

http://i200.photobucket.com/albums/aa294/oldtiffie/Black_book/BB_52.jpg

End edit.

oldtiffie
12-29-2010, 04:18 PM
As a prelimary to the standard DP cutter discussion that I will address later regarding the standard 8-cutter sets to/for each DP, I thought I'd post these tables:


Cutter No.1 will cut gears from 135 teeth to a rack.

Cutter No.2 will cut gears of 55 to 134 teeth.

Cutter No.3 will cut gears of 35 to 54 teeth.

Cutter No.4 will cut gears of 26 to 34 teeth.

Cutter No.5 will cut gears of 21 to 25 teeth.

Cutter No.6 will cut gears of 17 to 20 teeth.

Cutter No.7 will cut gears of 14 to 16 teeth.

Cutter No.8 will cut gears of 12 and 13 teeth.

http://i200.photobucket.com/albums/aa294/oldtiffie/Machinery_HB27/MHB27_P2047.jpg

John Stevenson
12-29-2010, 05:19 PM
This is another "bump"

I'm not finished with this thread yet.

Hang on folks, go make a coffee Tiffie still has a few pages to scan in, Machinery Handbook, 28th Ed has 3,455 pages ;)

.RC.
12-29-2010, 06:21 PM
While we are waiting, John can go out and find some ***** for us all to share..
.
.
.
.
.

.
.
.
.
..
.
http://www.*****drinks.com/home.html

Optics Curmudgeon
12-29-2010, 06:27 PM
Somehow the discussion of the gear hobber is transforming into one about using involute cutters. Use caution, they are very different things. Perhaps some alien mind control going on.

jkopel
12-29-2010, 06:29 PM
Since gears are always a popular topic of polite conversation around here

I knew it would get to this sooner or later...

lazlo
12-29-2010, 07:25 PM
Since gears are always a popular topic of polite conversation around here.
I knew it would get to this sooner or later...

Sadly, it's always the same person who kills the gear threads :(

In any event, Gary's gear hobber castings look really interesting.

TexasTurnado
12-29-2010, 08:40 PM
In any event, Gary's gear hobber castings look really interesting.

Even though I doubt I will build it, I agree Robert. It would be nice if we could get the discussion going again on the original topic....

Robin R
12-29-2010, 09:10 PM
Perhaps it would be more appropriate for oldtiffie to start a new thread, to discuss the accuracy of gears made from sets of gear cutters.

lazlo
12-29-2010, 09:15 PM
Even though I doubt I will build it, I agree Robert. It would be nice if we could get the discussion going again on the original topic....

John, I'm doing the math in my head, and with Durabar at ~ $1.55/lb, it seems like it would be a lot cheaper to mill the Jacobs/Helix components out of billet like Terry Saxton did in the 1990's Home Shop Machinist magazine.

I sent an email to College Engineering (in the UK) asking if they sell the plans separately from the castings. Judging by the age of the project, I'm hoping the CES plans are Imperial.

http://www.collegeengineering.co.uk/Castings/582.htm

Optics Curmudgeon
12-29-2010, 09:28 PM
As a comment that pertains (mostly) to the original discussion (gear hobber), I've made gears using most of the conventional means (involute cutters, shaper and hobbing) and found that I ran into a particular problem with hobbing. Most of the time, the HSM is trying to provide a gear they don't have, for whatever reason. Sometimes as a replacement part, sometimes for a particular use. Hobbers rely on a reduction train, and as often as not you don't have the particular gear(s) you need to get the reduction right. Needing a gear to make a gear, the dog has itself firmly by the tail. An electronic reduction system would provide one workable answer, but is a project in itself. A working hobber is very handy for those that will make lots of gears (I keep considering it myself), but the best solution for me has remained the involute cutter.

lazlo
12-29-2010, 09:34 PM
Most of the time, the HSM is trying to provide a gear they don't have, for whatever reason. Sometimes as a replacement part, sometimes for a particular use. Hobbers rely on a reduction train, and as often as not you don't have the particular gear(s) you need to get the reduction right. Needing a gear to make a gear, the dog has itself firmly by the tail.

That's a good point. The Jacobs/Helix (I honestly don't understand the difference) kit advertises that you can get started with the standard Myford (or presumably South Bend, Logan, ...) change gear set, and then, with a single hob, generate all the gears you need to complete the hobber gear train set. In other words, you're hobbing the gears you need for the hobber.

That's a neat bootstrap sequence that appeals to my EE background :)

Optics Curmudgeon
12-29-2010, 09:42 PM
The rub comes with prime numbers, that was why machines like the Fellows had huge gears sets, I still see the electronic reduction as the way forward on these. Easily adapted to the design discussed here.

Robin R
12-29-2010, 09:59 PM
Unless all you are only going to produce gears for models, then the Jacobs hobber is rather small. I think hobbing with a universal horizontal with an electronic rotary table/ dividing head is the way to go. Then you aren't so limited in the size of gears and you can also make helical gears up to 45°.

TexasTurnado
12-29-2010, 10:17 PM
John, I'm doing the math in my head, and with Durabar at ~ $1.55/lb, it seems like it would be a lot cheaper to mill the Jacobs/Helix components out of billet like Terry Saxton did in the 1990's Home Shop Machinist magazine.

I sent an email to College Engineering (in the UK) asking if they sell the plans separately from the castings. Judging by the age of the project, I'm hoping the CES plans are Imperial.

http://www.collegeengineering.co.uk/Castings/582.htm

If I were to build it, that would be my approach - it looks too flimsy to me as is so I would do a x2 to x4 upsizing of everything. :D

But the prime number gear train is still a problem, which is why I am leaning towards an electronic setup like Sir John described. One could always buy or cut the special gears with an involute cutter, but the electronic division approach seems more expeditious.

oldtiffie
12-29-2010, 11:02 PM
I have no doubt that the hobber will work, but how well it works will be limited to or by the design, how well it is made and how sharp the cutters are - and are kept - as well as the care used in its use.

As I pointed out earlier that those 8mm bore(d) hobbing cutters only - bases on the cutters posted here - have a maximum DP of 24 or a Metric Module 1.0 or there abouts.

Machinery's Handbook only goes from 1DP/20DP for the purposes of getting the settings for your gear caliper (which is only calibrated to 0.001"/0.02mm anyway and measuring off those caliper points is a PITA at the best of times. Even a disc micrometer has its problems.

So some may not have tables to rely on and may have to do their own calculations.

These pics are Metric Module1 (~ 25.4DP) which can be handled but it sure gets harder as the DP's rise and the gears get smaller. Note that they are measuring over several gear teeth. Its lot harder measuring over one toothof that size (or smaller?) with a gear caliper.

http://i200.photobucket.com/albums/aa294/oldtiffie/measuring/Gear-measure1.jpg

Calipers are no better:
http://i200.photobucket.com/albums/aa294/oldtiffie/measuring/Gear-measure2.jpg

http://i200.photobucket.com/albums/aa294/oldtiffie/measuring/Gear-measure3.jpg

http://i200.photobucket.com/albums/aa294/oldtiffie/measuring/Gear-measure4.jpg

These tools are necessary as all the tables refer to "basic gear (tooth)" and there is no allowance for back-lash.

It is not a lot of use relying only on the outside diameter of the gear blank and the dialed-in depth of cut as there could be cutter or work "spring" etc. The accuracy is determined by the width across a single or multiple teeth.

If you have to remove the job from its spindle/arbor to measure or test (run, mesh) it and if it needs more off, re-setting it to the hobbing cutter might be quite a job.

If the job is not well enough supported and or "flutters" under a cut the cutting edge will "go" and will need a good tool & cutter grinder (with an 8.0mm arbor/mandrel) and some fancy wheel-shaping and grinding to re-sharpen it.

I don't know if all or any of the gears for the gear-trains are supplied as castings in the kit or not. But as most will not be able to hob/shape/generate the gears, I guess they will have to mill them with standard gear cutters.

If some-one has the interest and a lazy US$400 (plus) for the casting kit as well as any other requirements as regards parts, tools, machines and skills and wants to make a small hobber that works - this could be what your are looking (forward?) to as it may well fill many needs.

I am quite sure it will work and satisfy its intended purpose.

So far as I can see or recall, there are only two who have directly or indirectly doubted that it will work. In the absence of them not having retracted those views I can only assume that they still have those views.

In that event that I will assume that those who have not said it won't work seem to be the majority here and that they think it will work as intended.

I will get back to other related stuff regarding gear-cutting.

Later.

lazlo
12-29-2010, 11:11 PM
The rub comes with prime numbers, that was why machines like the Fellows had huge gears sets, I still see the electronic reduction as the way forward on these. Easily adapted to the design discussed here.

Ironically, an electronic divider like Sir John's electronic hobber would be vastly simpler for me to build, because of my profession. But as Jerry (Macona) mentioned in a private message (I'm sure he won't mind), "it's fascinating to watch it [the Jacob's hobber] run."

John E.: agree, it looks small and fragile. The max hobbed gear diameter on the UK version is ~ 4.5".

TexasTurnado
12-30-2010, 12:04 AM
Ironically, an electronic divider like Sir John's electronic hobber would be vastly simpler for me to build, because of my profession. But as Jerry (Macona) mentioned in a private message (I'm sure he won't mind), "it's fascinating to watch it [the Jacob's hobber] run."

John E.: agree, it looks small and fragile. The max hobbed gear diameter on the UK version is ~ 4.5".

What I am curious about concerning the electronic version, is whether the stepper imparts any noticeable distortion to the hobbed gear due to the limited number of pulses per gear tooth space. Low numbers of teeth should be less affected, but in a couple of pics, Sir John appears to be making 127 tooth gears where the pulses available are smaller. John, would you (or anyone else) like to comment on this? Does the inertia of the stepper smooth this out?

lazlo
12-30-2010, 12:26 AM
What I am curious about concerning the electronic version, is whether the stepper imparts any noticeable distortion to the hobbed gear due to the limited number of pulses per gear tooth space.

In the Model Engineer article, they mention that they had rasterization issues (faceting) with lower pulse counts, and ended up going to 4,000 pulses per revolution (if I remember correctly?)

I'll send you the article...

Black_Moons
12-30-2010, 03:23 AM
I would assume one would use a stepper to drive a standard 40:1 rotary table, And that would greatly improve resolution.
Throwing a 5:1 or reduction ontop of that would'nt hurt either, I don't think you need to worry about backlash either so much since you should only be rotating in one direction.

oldtiffie
12-30-2010, 03:52 AM
A 90:1 rotab ratio is 90/40 = 2.25 times as accurate as a 40:1 rotary table for the same step from the stepper.

That is to say that one turn of a direct-coupled stepper will drive the rotary table 360/90 = 4 degrees for a 90:1 reduction or 360/40 = 9 degrees for a 40:1 reduction.

Black_Moons
12-30-2010, 04:20 AM
Oh, Yes, a higher turn ratio rotary table would clearly improve resolution, I was just saying you would use a rotary table insted of just directly connecting the stepper.

John Stevenson
12-30-2010, 08:26 AM
Long post, go get a coffee.

I'll try to cover the Jacobs one first and from a practical point of view.
Firstly I don't have one and have never built one but at the shows Giles Parkes takes his and does demo's on gear hobbing, I'm very friendly with Giles and we bounce idea off each other.

At the last show were were cutting a Delrin gear 15 teeth 2.5 mod, that's about 10DP helical in about 34 minutes using just a 3mm end mill on the little KX1 CNC mill as part of a rolling demo.

Giles had his machine there cutting a Delrin helical gear about 24DP, 30 tooth and his machine took about 1 -1/2 hours. In Delrin he had to do it in tow passes to get a decent quality gear.
If he'd been using steel he would have had to have about 5 or 6 passes to get to full depth because of the machines limitations.

During cutting this machine is very noisy and you can see the various gears, including the one being cut flexing. Cut quality isn't bad a but I would not like to guess how accurate the generated profile is.

I don't have any pictures of Giles hobber but it has been extensively modified to do helicals and also spur gears.
The work head is different for a start and can carry an overarm a la Myford dividing head style to support the blank being cut.
Giles also has spare hob arbors which are changed to suit the hob in use as he uses anything from 8mm to 1"
Giles does have a full set of gears up to 100 including all the primes that have been cut on this machine by using temporary gears made by drilling a blank and then free hobbing it to form a passable gear which is then used to make the correct one.
He used to cut these at shows for something to do as part of the demonstrations.

Personally I thing this machine should be scaled up X 2 at least.

Electronic dividing, firstly you need it to work thru a worm and wheel drive like a dividing head or rotary table because of the cutting forces generated and the reduction needed.
just sticking a blank on the end of a stepper or belt driven reduction won't work as the cutter rotates the work, don't ask, been there, done that and wasted 1/2 a week making a nice D1-3 mounted head that couldn't even put a starigh keyway in without it spiralling.

I used an off the shelf reduction gearbox that are supplied to bolt onto metric framed electric motor, very popular here and quite cheap.

http://homepage.ntlworld.com/stevenson.engineers/lsteve/files/hob%20indexer1.jpg

I wanted a 20:1 but could only get a 40:1 at short notice so it was geared up 2:1. Because these do have a small amount of backlash in the drive although they are very well made, hardened worm, bronze wheel and bearing and seals all round, I bored the work bearing housings out and fitted eccentric sleeves so I could get the backlash down to a minimum
Initially the blank was carried on 22mm keyed arbors that went directly into the gearbox, this had now been modified to carry a face mounted ER32 lathe chuck so I can use any size stub arbor.

The encoder is driven from the hob arbor and is setup to give 4,000 pulses per rev.

http://homepage.ntlworld.com/stevenson.engineers/lsteve/files/hob%20indexer7.jpg

In this case I used a 1024 count encoder that came off a CNC lathe and it was ideal as it carries it's own bearings. The gearing is 125 to 32 to convert the 1024 to 4,000

This figure was arrived at by trial and error, if you have much less then you get facetting of the gear. If you go higher then you run out of pulses and the stepper drive stalls.
Even at this it's still a compromise, do a low count gear and the gear runs fast to keep up, if you try to run the hob too fast then it stalls.
4,000 is a happy medium that allows you to keep the hob speed within the speed range recommended for hobs.

Helicals can be done by simply bolting the stepper head to the bed at the required angle, no taking into account the helix as that is set on the table, no drive train needed as regards feed, you can hand feed as fast or as slow as you need.
http://homepage.ntlworld.com/stevenson.engineers/lsteve/files/hob%20indexer11.jpg

And matched pairs can be done all at the same setting by simply going below the hob for one and above for it's mate.

http://homepage.ntlworld.com/stevenson.engineers/lsteve/files/hob%20indexer17.jpg

Most of the initial work on this electronic hobbing was done by Brian Thompson, the later design that I use was a combined effort with Brian, myself and Don Foreman off the old RCM newsgroup who designed a better and simpler circuit.

Now I need a coffee......

Weston Bye
12-30-2010, 09:02 AM
John,
After your cup of coffee and a pee break, could you elaborate on the electronics between the encoder and the stepper? The principles appear to be similar to what I did with my electronic threading scheme. I would enjoy seeing other different ways of skinning the cat.

lazlo
12-30-2010, 10:30 AM
The principles appear to be similar to what I did with my electronic threading scheme.

Agreed -- I've noted several times before that John's electronic gear chain is nearly identical to an electronic leadscrew.

John Stevenson
12-30-2010, 10:48 AM
Weston,
Email sent.

Weston Bye
12-30-2010, 11:04 AM
Weston,
Email sent.

Thanks John. Well, imagine my astonishment!

John forwarded the details covered in an article carried in MEW 108. The principles were remarkably similar to the method I arrived at independently for electronic threading - including similar encoder pulse rates and selector switches. I suppose that the job dictates the means.

I am pleased that my own efforts in the electronic threading articles have been validated, or perhaps better to say, validate the work John describes.

Spin Doctor
12-30-2010, 12:22 PM
John, I am curious about one thing. When swinging the work on the table to compensate for the lead angle of the hob much less the helix angle in helical gears how are you compensating for the helical errors that in my mind will creep in given that the part of the hob in relation to its total cutting length that is forming the gear tooth is changing as the work traverses. Maybe I am just stuck in a gear driven hobber/shaper world but that is where my gear cutting experience is. When I would have to set-up our Barber-Coleman the Hob Swivel not only had to be locked down on the correct lead angle but the Hob Spindle also had to be lcoked in place to maintain the correct tooth profile and lead angle. :confused:

John Stevenson
12-30-2010, 12:48 PM
John, I am curious about one thing. When swinging the work on the table to compensate for the lead angle of the hob much less the helix angle in helical gears how are you compensating for the helical errors that in my mind will creep in given that the part of the hob in relation to its total cutting length that is forming the gear tooth is changing as the work traverses. Maybe I am just stuck in a gear driven hobber/shaper world but that is where my gear cutting experience is. When I would have to set-up our Barber-Coleman the Hob Swivel not only had to be locked down on the correct lead angle but the Hob Spindle also had to be locked in place to maintain the correct tooth profile and lead angle. :confused:

SD. the work is only swung on the table when doing helicals.
The table on this machine can swivel and that is where the lead angle applies.

When doing spur gears there is only the hob lead angle to take into account so the blank passes under the hob parallel to the teeth.

I don't follow what you say above in that you say the hob swivel has to be locked at the correct lead angle and the hob spindle has to be locked to the lead angle.

Are these not one and the same, Can't check as Giles has my Barber Coleman book.

TexasTurnado
12-30-2010, 03:24 PM
Long post, go get a coffee.

I'll try to cover the Jacobs one first and from a practical point of view.
Firstly I don't have one and have never built one but at the shows Giles Parkes ....
.
.
.
Now I need a coffee......

Thanks for your long post, John. I had run across your earlier files on ntworld during my searches for hobbers. One key parameter I have not seen mentioned is the actual rpm of the hob - would you comment on that?

The problem I am having trouble wrapping my brain around is this: to cut a theoretically correct gear with a hob requires a constant angular velocity of the blank (assuming the hob is being rotated at a constant rate). This is normally achieved with a gear train, but with the electronic version using a stepper this does not appear to necessarily be the case.

Perhaps I do not fully understand how a stepper motor works, but it would seem to make discreet steps of a certain size at a time dictated by the pulsed input. When an input pulse to the motor occurs, I would expect there to be an initial acceleration time, followed by rotation, then a deceleration time while to motor settled to its new position. As the pulse repetition rate increased, these motions would blur together, but there would still seem to be some angular velocity error.

When hobbing a gear, this motion would seem to produce errors in the profile of the teeth because the blank is alternately behind where it should be and then catches up when the pulse occurs. On succeeding revolutions, the step will occur at a different time in the profile (except perhaps for the case where the division does not have a remainder), thus imparting a jitter in angular velocity of the blank.

As a minumum, I would expect this jitter would cause the resulting tooth space to be wider than expected. For example, on a 16DP 127 tooth gear and using 4000 pulses, jitter should be about .006 in at the tooth space pitch line (4000/127 = 31.49 pulses per tooth space, pi/16 = .196 in per tooth space, .196/31.49 = .0062 in).

Am I lost here in my thinking?

John Stevenson
12-30-2010, 05:18 PM
Don't forget the 20:1 reduction in my case.

Spin Doctor
12-30-2010, 05:57 PM
OK, Spur Gears are clear. I had forgotten that you have an Elliot Universal Mill with a swiveling table IIRC. My bad. As to the Hob Spindle being locked on most B-Cs I have seen the Spindle Assembly that carries the Hob is adjustable in and out towards the front of the machine. This comes in handly especially in production work where let's say a Timing Gear/Sprocket being cut for a IC engine has to be in an exact relationship to the keyway in order to maintain valve timing. It also comes in handy when doing repair work on gears that have had one or more teeth broken and you have to pick the tooth profile back up. Although you can also do this with the lock screws on the Work Drive that lock the drive dog. Of course B-C did build Gear Hobbers that featured a Hob Slide that fed both Tangentially and Obliquely to produce Taper Root Splines. Figuring out Helicals was bad enough. I'm just glad we didn't have to do any of those.

willmac
12-30-2010, 06:09 PM
John -

Is there any reason why you could not use a beafy universal dividing head in place of the motor gearbox in your setup? Obviously would need to motorise it, but that might even be easier and would make a more versatile tool for other purposes.

rwf71
12-30-2010, 06:23 PM
Spin Doctor , I think the movement of the hob spindle fore & aft on a Barber Colman is what is called the "hob shifter" . When cutting spur gears only a small portion of the hob is in contact with the part & when that section of the hob gets dull, ( set by a part counter ) a small motor on the hob slide leadscrew shifts the hob to a "new/sharp" section of the hob , thus dulling the entire hob before it has to be sharpened in a production run. I'm no gear cutting expert but I work maintenance in a shop with a few old BC's. Rick

John Stevenson
12-30-2010, 06:31 PM
John -

Is there any reason why you could not use a beafy universal dividing head in place of the motor gearbox in your setup? Obviously would need to motorise it, but that might even be easier and would make a more versatile tool for other purposes.

Bill,
None at all.
In fact I have one.

http://www.stevenson-engineers.co.uk/files/hoffman.jpg

However this is in constant use and the Victoria mill has been setup for hobbing for about 5 years and I like to leave it setup.

.

djc
12-30-2010, 06:32 PM
...The principles were remarkably similar to the method I arrived at independently for electronic threading - including similar encoder pulse rates and selector switches.

One thing that has progressed since the article on gear hobbing was first written, which make a 'today solution' much easier, is the availability, and ease-of-use of microcontrollers. There are lots of flavours from which to choose: PIC, PICAXE, Basic Stamp, Arduino as well as others.

The circuit in the original hobber is 'hard-wired' for a specific encoder, worm gear and stepper combination and it is quite difficult to make it work if the parts you have at hand don't suit this.

With a microcontroller, all these numbers are simple to change. You don't even need a large count encoder as you can count each edge of each phase of, say, a 500ppr one and achieve 2000ppr for free. Your worm gearbox is 27:1 ratio (well, it was cheap)? Doesn't matter. You only have 13 and a 17 tooth timing gears to drive the gearbox? No problem. You want 6x microstepping? Can do.

[Edit: added "...on gear hobbing" after "since the article..." as original apears to reference Weston's article, which was not intended.]

TexasTurnado
12-30-2010, 06:45 PM
Don't forget the 20:1 reduction in my case.

That is how you got 4000 pulses per blank revolution from a 200 pulse stepper, is it not? So it seems it is included....

snowman
12-30-2010, 06:47 PM
I question whether a lot of the available mcu's would be capable of driving the stepper motor fast enough.

John has already pointed out that at higher hob RPM's, it is going faster than the electronics are capable of handling....and at lower RPM's you do not get fluid movement of the gear blank.

For what it is, it's pretty straightforward. The gears do not have to produce exactly 4000 pulses, but in order to change it, you have to have a firm understanding of what is actually happening.

TexasTurnado
12-30-2010, 06:48 PM
Bill,
None at all.
In fact I have one.

http://www.stevenson-engineers.co.uk/files/hoffman.jpg

However this is in constant use and the Victoria mill has been setup for hobbing for about 5 years and I like to leave it setup.

.

Damn, now that is HEFTY!

snowman
12-30-2010, 06:57 PM
With a microcontroller, all these numbers are simple to change. You don't even need a large count encoder as you can count each edge of each phase of, say, a 500ppr one and achieve 2000ppr for free. Your worm gearbox is 27:1 ratio (well, it was cheap)? Doesn't matter. You only have 13 and a 17 tooth timing gears to drive the gearbox? No problem. You want 6x microstepping? Can do.

The other thing though is the 500 cpr encoder is just as easy to turn in to 2000 cpr with nothing more than the schmitt trigger and a flip flop, but the magic number of pulses was 4000. Total cost $.72, need for programming...nodda...and you won't have to worry about bugs.

lazlo
12-30-2010, 07:02 PM
Can't check as Giles has my Barber Coleman book.

Doesn't Giles have your Barber Coleman too? :D

John Stevenson
12-30-2010, 07:24 PM
Doesn't Giles have your Barber Coleman too? :D

Sort off :D

Giles has this Barber Coleman hobber, nice machine, well equipped. I keep dropping hints that he ought to leave it to me in his will.

Giles is quite hard of hearing and every time I ring him up he shouts down the phone

"I'm not dead yet "

Weston Bye
12-30-2010, 08:24 PM
One thing that has progressed since the article was first written, which make a 'today solution' much easier, is the availability, and ease-of-use of microcontrollers. There are lots of flavours from which to choose: PIC, PICAXE, Basic Stamp, Arduino as well as others.

The circuit in the original hobber is 'hard-wired' for a specific encoder, worm gear and stepper combination and it is quite difficult to make it work if the parts you have at hand don't suit this.

With a microcontroller, all these numbers are simple to change. You don't even need a large count encoder as you can count each edge of each phase of, say, a 500ppr one and achieve 2000ppr for free. Your worm gearbox is 27:1 ratio (well, it was cheap)? Doesn't matter. You only have 13 and a 17 tooth timing gears to drive the gearbox? No problem. You want 6x microstepping? Can do.

As Snowman pointed out, many (most) microcontrollers will not be able to keep up. Maybe for my threading application, but probably not for hobbing. That was one of the reasons I resorted to TTL logic for my threading system. It can fly along at 17 Mhz if necessary. True, the hardwired electronic gearbox dictates much of the rest of the hardware, but once that is in place, the advantage of the programmable device is negated.

oldtiffie
12-30-2010, 08:49 PM
John, I am curious about one thing. When swinging the work on the table to compensate for the lead angle of the hob much less the helix angle in helical gears how are you compensating for the helical errors that in my mind will creep in given that the part of the hob in relation to its total cutting length that is forming the gear tooth is changing as the work traverses. Maybe I am just stuck in a gear driven hobber/shaper world but that is where my gear cutting experience is. When I would have to set-up our Barber-Coleman the Hob Swivel not only had to be locked down on the correct lead angle but the Hob Spindle also had to be lcoked in place to maintain the correct tooth profile and lead angle. :confused:

The true form of a hob, worm or worm-wheel is normal to the helix.

A screw thread true form is horizontal - parallel to the lathe spindle axis and the lathe bed.

There is no significant change in shape of the true shape of hob or worm if it is turned as (for) a screw thread provided that the helix angle is not greater than about 5*, otherwise the true shape cutting tool has to be tilted by the helix angle.

As the hob true form is normal to the grooves cut in a hob (which is a spiral - usually single start), the mill table has to be tilted left or right by the hob helix angle and the work axis is set parallel to the mill table "X" axis so that the work is fed directly into the hob with no lateral or side-ways movement relative to the hob.

With helical and spiral milling the feed is via the "X" slide/screw.

With worm-wheels the "X" and "Y" slides are locked and the feed is via the knee "Z" (ie a sort of "plunge" cut).

The lineal/axial pitch of the hobbing cutter normal to the helix angle is equal to the circular pitch for the DP of any gear it is to cut.

Hence a hob will cut any gear with these same pressure angle (slopes of the flanks/sides of the hob) and the same DP as the hob.

oldtiffie
12-30-2010, 08:57 PM
As Snowman pointed out, many (most) microcontrollers will not be able to keep up. Maybe for my threading application, but probably not for hobbing. That was one of the reasons I resorted to TTL logic for my threading system. It can fly along at 17 Mhz if necessary. True, the hardwired electronic gearbox dictates much of the rest of the hardware, but once that is in place, the advantage of the programmable device is negated.

It will not help either if the torque required of the stepper is insufficient as it will "miss steps" (too).

As I understand it, there is no feed-back from the driven drive (lead-screw or rotary table) as the step pulse producer just keeps pulsing and stepping on regardless.

It is different with servos where a positional back-feed can be compared with the computed position and any difference (error) rectified by an increase or decrease in the pulses and the steps.

lazlo
12-30-2010, 09:08 PM
As Snowman pointed out, many (most) microcontrollers will not be able to keep up.

Um, even a lowly 8-bit AVR runs at 20Mhz :)

AVR's are RISC, so Mhz = MIPS:

AVR mega/tiny family goes up to 20 MHz
AVR Xmega family goes up to 32 MHz
AVR UC3 (32-bit) family goes up to 72 MHz (1.46 DMIPS / MHz)

But why do you even need that? Assuming the spindle is running at 300 RPM (just pulling a number out of my as$), that's 5 revolutions per second. 4,000 pulses per revolution is 20,000 pulses per second. You're dividing the spindle RPM by the hob ratio.

So you only need to do 20,000 fixed-point divides per second. Even the most primitive microcontroller, like an Old-School Z-80, can do that.

I think Snowman's point is that it's dirt-simple to interpolate more pulses, to reduce the rasterization effects.

Weston Bye
12-30-2010, 09:17 PM
oldtiffie rightly points out that with open loop systems neither of the methods , hardwire logic or microcontroller, will "know" if the stepper misses a step due to overload. As such, the system needs to be designed and motors sized with this in mind.

An advantage of a system that uses an encoder with a reasonably high pulse rate coupled to the spindle and divide from there is that spindle speed variation in spindle speeds are more accurately tracked.

Some systems for threading use a single pulse per revolution and use the microcontroller to assume a constant speed through the whole revolution and calculate the step rate during that turn. I presume that this works well enough on a large lathe with plenty of inertial mass and power, but with a puny little Sherline or similar lathe, constant spindle speed can't be counted upon. I know this from experience.

However, with a sufficient PPR encoder I am able to stop and start the spindle in the middle of a thread without loosing position.

lazlo
12-30-2010, 09:37 PM
oldtiffie rightly points out that with open loop systems neither of the methods , hardwire logic or microcontroller, will "know" if the stepper misses a step due to overload.

Yes, but John's answer to that is brute force: he uses a 20:1 worm divider, so he has immense leverage, and doesn't lose steps.

The issue about moving up to 4,000 pulses was to reduce faceting, not because of skipped steps.

I don't see how increasing the PPR improves any issues with loss of steps?

TexasTurnado
12-30-2010, 09:59 PM
I question whether a lot of the available mcu's would be capable of driving the stepper motor fast enough.

John has already pointed out that at higher hob RPM's, it is going faster than the electronics are capable of handling....and at lower RPM's you do not get fluid movement of the gear blank.

For what it is, it's pretty straightforward. The gears do not have to produce exactly 4000 pulses, but in order to change it, you have to have a firm understanding of what is actually happening.

The speed of the mcu is only one of the the limitations - apparently the encoders are being run at their upper speed limit also. In this regard, the small hobs are at a disadvantage (other than rigidity) because they need to turn faster to achieve the same cutting speed in fpm at their teeth as a large diameter hob.

lazlo
12-30-2010, 10:08 PM
Ah, I missed that quote John/Wes/Snowman - sorry.

I don't think the pulse division is particularly taxing, for the reasons I described: ~20K fixed-point divisions per second is super simple.

As far as the encoders -- the 1,000 PPR encoder that Brian Thomson used in the MEW article is pretty old. I have some newer Mitsubishi servos/controller that have 100,000 a 1,000,000 pulses per revolution (seriously!) In one of John's posts he mentions a Heidenhain encoder, which would be several thousand pulses per revolution.

But I think the point of Brian's article is that you need enough divided pulses to cut a smooth involute. So the original 1,000 pulses per revolution on the spindle would be divided by 30 for a 30 tooth gear, and the gear blank would only be getting 33 steps per revolution. Not very smooth.

So they mechanically and electronically interpolated by a factor of 4x, which would get you 133 steps per revolution on the gear blank.

Weston Bye
12-30-2010, 10:12 PM
I don't see how increasing the PPR improves any issues with loss of steps?

True enough, it doesn't.


Um, even a lowly 8-bit AVR runs at 20Mhz

AVR's are RISC, so Mhz = MIPS:

AVR mega/tiny family goes up to 20 MHz
AVR Xmega family goes up to 32 MHz
AVR UC3 (32-bit) family goes up to 72 MHz (1.46 DMIPS / MHz)

But why do you even need that? Assuming the spindle is running at 300 RPM (just pulling a number out of my as$), that's 5 revolutions per second. 4,000 pulses per revolution is 20,000 pulses per second. You're dividing the spindle RPM by the hob ratio.

So you only need to do 20,000 fixed-point divides per second. Even the most primitive microcontroller, like an Old-School Z-80, can do that.

Also true, but sadly I haven't been able to get such speed yet. Much of my stepper control applications has been with controllers that use compiler software to generate the machine code that runs on the microchip. Therein lies some time-eating overhead. I am tempted to dust off an old Z-80 and program that in tight machine language, but to what purpose? Besides, time is limited, life goes on, and articles have to go out on schedule.

I have just completed a functional microcontroller system (20mhz clock) for my X-Y table for an upcoming article, but the rapid moves, as fast the the controller can generate, are rather slower than I wished.

oldtiffie
12-30-2010, 10:23 PM
TT.

I have a DivisionMaster which drives a 400 1/2-step stepper drive which is direct-coupled to my 6" Vertex 90:1 rotary table.

360/400 = 0.09 degrees rotation of the stepper and my rotary table worm drive. The 90:1 worm and wheel drive converts that 0.09 degrees to 0.09/90 = 0.001 degrees table movement.

0.001 degrees x 60 = 0.06 minutes x 60 = 3.6 arc seconds the tan of which is 0.0000175 or ~ 1:57,300

Read the "good oil" here:
http://medw.co.uk/storage/attachments/343/343/DivisionmasterManualV2-03-1.pdf

While it won't "hob", it can do some very fast and very accurate positioning for gear-cutting with the standard involute gear cutters or holes on a PCD and no need for indexing arms or plates either.

I was that pleased with it that I asked the developer if he had a kit for my 8"Vertex" 90:1 rotary table but he said that the rotab was OK but it would not handle the rotab with its 3-jaw chuck on.

If you read the pdf link you will see an example of what Weston Bye is saying.

oldtiffie
12-30-2010, 10:43 PM
Tooth profile and "depth of cut" are items often mentioned here.

It applies to all (for this purpose) spur gears with DP's of 25 and below (the small hobber - the subject of this thread) as well DP20 and above as cut with spur gears cutters and/or hobbing cutters on universal mills - in a HSM shop.

The depths of cut for involute gear cutters and hobs is for the "basic" gear where the chordal thickness and circular thickness of the tooth and the space between gears are the same.

This is not ideal as there is no functional back-lash.

This additional cutting is usually obtained by making the depth of cut deeper and/or taking side-cuts with the cutter.

If you are using a gear caliper to measure your gear/s (a single tooth) you will have to allow for the reduced chordal thickness and perhaps the great depth of cut (or side-cutting). Same applies to measuring over multiple teeth with a caliper of a disc micrometer.

Its going to be hard to get these measurements with small hobbers and larger numbered DP, both because its difficult and Machinery's Handbook is not much help as it is written for larger gears and cutters.

I will cover that in a later post.

A hob with the correct pressure angle and DP will be fine as it will cut all gears correctly.

With the 8-cutter sets of involute milling cutters only the following cutters will cut a tooth that is the same as one cut with a hob. All others are very good compromises and are certainly good enough for most gear-cutting work in a HSM or Jobbing shop:

Cutter No.1: 135 teeth

Cutter No.2: 55 teeth

Cutter No.3: 35 teeth

Cutter No.4: 26 teeth

Cutter No.5: 21 teeth

Cutter No.6: 17 teeth

Cutter No.7: 14 teeth

Cutter No.8: 12 teeth

Here is a partial copy of page 2047 of MHB27 which shows the chordal thickness and chordal addenda for a basic gear cut to nominal/basic depth of cut.

Chordal addenda and chordal thickness are required for setting or checking with a gear micrometer:

http://i200.photobucket.com/albums/aa294/oldtiffie/Gear_theory/MHB27_P2047A.jpg

http://i200.photobucket.com/albums/aa294/oldtiffie/Gear_theory/RMIT-Gear-measure1.jpg

If you have a look the 4DP line in the MHB table I posted above, you will see that for 4DP the chordal thickness for a basic gear tooth varies very little - from 0.3926" for a No.1 cutter at 135 teeth to 0.3915" for a number 8 cutter at 12 teeth. A difference of only (0.3926 - 0.3915 = 0.0011").

The chordal addenda changes relatively quickly over the same range ie (0.2628 - 0.2511 = 0.0117").

Using the same 4DP line I have worked out the chordal thickness for some random tooth numbers - all agree with the table:

http://i200.photobucket.com/albums/aa294/oldtiffie/Sketches/Gearthick1.jpg

I will address back-lash next as it applies equally to small and large hobbers and the 8 DP involute cutter sets.

Later.

TexasTurnado
12-30-2010, 11:16 PM
Ah, I missed that quote John/Wes/Snowman - sorry.

I don't think the pulse division is particularly taxing, for the reasons I described: ~20K fixed-point divisions per second is super simple.

As far as the encoders -- the 1,000 PPR encoder that Brian Thomson used in the MEW article is pretty old. I have some newer Mitsubishi servos/controller that have 100,000 a 1,000,000 pulses per revolution (seriously!) In one of John's posts he mentions a Heidenhain encoder, which would be several thousand pulses per revolution.

But isn't the problem with high count encoders the absolute rate in revolutions per minute that they will reliably put out that many pulses?


But I think the point of Brian's article is that you need enough divided pulses to cut a smooth involute. So the original 1,000 pulses per revolution on the spindle would be divided by 30 for a 30 tooth gear, and the gear blank would only be getting 33 steps per revolution. Not very smooth.

That was the point I was trying to make in my earlier post....

oldtiffie
12-30-2010, 11:30 PM
I'm not so sure about that.

A hobbed gear which turns at one tooth per revolution of the cutter will have as many flats or facets are there are teeth on the cutter per rev.

Unless they are big teeth, 12 or more facets will be so close to a smooth involute curve that it really won't matter for a lot of work otherwise leave a shaping/shaving/grinding allowance.

This is where involute cutters shine as that have smooth (pseudo/corrected) involute curved cutting edges.

OK - involute at the pitch point and a mix of hypo and epicyclic curves otherwise.

TexasTurnado
12-31-2010, 12:53 AM
I have been giving this problem some thought and have come up with a different approach to the electronics:

Instead of a stepper motor, suppose we substitute a DC motor driving a worm gear reducer and put an encoder on the DC motor shaft. Then the pulses from the divider (connected to the encoder on the hob shaft) can be phase compared to the pulses from the encoder on the DC motor shaft and used to tweak the DC voltage to the motor, causing it to rotate at exactly the right speed with constant angular velocity. If necessary to increase the DC motor rotation rate to a more desireable range, the ratio of the gearbox can be increased and a fixed divider (either mechanical or electrical) added to the encoder on the DC motors shaft to offset the ratio change.

In another life, I designed several RF frequency synthesizers, and this method was used there to phase lock a tunable oscillator through a divider to a crystal frequency standard. Motorola even made an IC that would do the phase comparison and drive a "charge pump" that added charge to an integrator as needed to maintain the lock - and it also had circuitry to generate a sweep voltage to find lock. The components were different there, but the principle is the same as I am proposing..... :D

Edit: If lock turns out to be difficult to achieve reliably at the correct rpm, then a lookup table connected to the divider switches can be used to drive a D/A converter to set the DC voltage to approximately the correct voltage initially. This method was used in some synthesizers to avoid 'false' locks.

dp
12-31-2010, 01:06 AM
VCO's will generate jitter around the DAC lsb. The lsb is either 0 or 1. This is measureable by steps of +- 1 / loop gain for reasonable values of loop gain. For unreasonable values the +-1 is seen as a step input and will cause overshoot.

J Tiers
12-31-2010, 01:16 AM
I'm not so sure about that.

A hobbed gear which turns at one tooth per revolution of the cutter will have as many flats or facets are there are teeth on the cutter per rev.

Unless they are big teeth, 12 or more facets will be so close to a smooth involute curve that it really won't matter for a lot of work otherwise leave a shaping/shaving/grinding allowance.



Actually, that isn't all of it, really, since there can be more than just one pitch of the hob cutting on one tooth.... one pitch of the hob cuts on as much of the tooth flank as 'slides past" it.... but it takes at least a couple or three turns for the tooth flank to completely get out of contact.

probably one tooth space is worked on during 5 or 6 turns total, which gives 2 to 3 turns for each flank. maybe more, I have not examined it closely, I just went and looked at a hob vs a gear roughly meshing them.

So you need to look at the number of teeth total that cut on one flank of the tooth, which may not be identical with the number in one revolution. THAT is how many "facets" would be cut, and it could easily be double the number of teeth per rev.

TexasTurnado
12-31-2010, 01:36 AM
VCO's will generate jitter around the DAC lsb. The lsb is either 0 or 1. This is measureable by steps of +- 1 / loop gain for reasonable values of loop gain. For unreasonable values the +-1 is seen as a step input and will cause overshoot.

The DAC is only used to initially set the DC voltage in the ballpark to avoid false lock points - after that the phase lock loop takes over and adjusts the voltage to exactly what is needed for lock at the correct rpm.....

J Tiers
12-31-2010, 01:51 AM
Per the PLL, etc:

Pretty much all feedback systems require an error signal to drive them. Therefore they tend to always lag or lead the correct value.

The problem is to get the loop gain high enough to minimize that without having any overshoot, etc. that can get to be an intractable and problematic feedback function, especially when there are "springy" elements like motors/torque in the system along with masses.

probably a reasonable approximation is possible without getting too far out, but the mechanical system is likely to have a lot more elements than the usual PLL oscillator system. I don't particularly want to analyze it, but with decent modeling one could do the tweaking with a circuit simulator.

djc
12-31-2010, 03:15 AM
Lots of replies rolled up into one post. Shout if the attributions are incorrect.


John has already pointed out that at higher hob RPM's, it is going faster than the electronics are capable of handling

Strictly speaking, it is not the electronics that can't handle the speed, but the mechanics of the stepper motor. The divide-by-n chip (74HC4059) used is good well into the MHz range on its input.


So you only need to do 20,000 fixed-point divides per second

It's not even necessary to tax the processor that much. Do all your calculations in the setup phase, which will give you a 'magic number'. Start counting pulses on your input; when they reach the magic number send a pulse on the output and reset the counter. I have seen an AVR routine written in C (not assembler) that reads quadrature pulses and correctly updates a counter that is 23 machine instructions long.


It is different with servos where a positional back-feed can be compared with the computed position and any difference (error) rectified

A very, very crude closed loop system is simple with a programmable device. Maybe as simple as a single pulse per rev. sensor on the output. You will know in advance how many input pulses should give one index pulse on the output. Every time the index pulse triggers, look at the value of your input counter. If it differs from what it should by zero or a small amount, carry on; any more, take some simple action. To do any more than this gets very complicated...


...with a sufficient PPR encoder I am able to stop and start the spindle in the middle of a thread without loosing position.

I think you need to shout louder about this. Not only what you say, but what you said in your original thread - that you can reverse at will, turn the spindle by hand at varying rates (of change of angular velocity). For threading of any type, you have to do something at the end of the thread. With this electronic version, you just wind back the cross-slide, reverse the motor until you are back to the start, put on your new cut and go again.

Someone else posted the following movie clip and it is very impressive that you can do this with $20 of electronics

http://www.youtube.com/watch?v=pxXGN2SVrhU


The speed of the mcu is only one of the the limitations - apparently the encoders are being run at their upper speed limit also

Could you give some concrete figures from datasheets for this? Just picking one at random,

http://www.quantumdev.com/products/optical_encoders/qd200.html

it seems to be good up to 1/2 MHz. I am guessing here that the figure is per channel, so that's 2MHz if you read all four edges.

macona
12-31-2010, 05:04 AM
This would be a perfect project for a small Mitsubishi servo motor. They have a setting for an encoder input. Then with the electronic gearing parameters you can easily set it to just about any ratio you can think of.

I think I still have a 200w motor and drive for 120v. I know I also have another set from a different company.

Add: Just dug through my servo box. Found a Misubishi MR-J2S-20A1 servo drive with the motor, 200 watt ~1/4HP. It has a 131072 count encoder. The gearing setting allows you to set the ratio as a fraction with the numerator and denominator being any integer between 1 and 65536.

willmac
12-31-2010, 08:50 AM
I think we now have several themes running in this thread.

Responding to OldTiffies posts - I think you may be missing the fact that hobs come in single and multi-threaded forms. For a single thread hob it is true that the hob makes one rev for each tooth of the gear, but this is not so for multi lead hobs. This factor needs to be taken account of in setting up the hob, and is the point that JTiers was making earlier.

I'm not sure nowadays what is the deciding factor for choosing either type, but I think it used to be the case that for highest precision, single lead hobs were preferred, and for higher production efficiency multi thread hobs were better but I may be wrong/out of date onthat.

A factor in determining the degree of faceting is the number of gashes around the hob.

Macona -

I'm not familiar with the type of motor that you describe but if I understand you correctly it would not require any separate electronic dividing to work and would provide smooth operation given the cpr needed from the hob mounted encoder? If so that would be a very interesting way of building a hobber.

John Stevenson
12-31-2010, 08:55 AM
Many commercial hobbing machines are now all electronic and don't rely on gearing.
I suppose it's a very grey area now on what's a hobbing machine and what's a 7 axis CNC ?

MuellerNick
12-31-2010, 09:01 AM
I didn't read the whole thread, just the title.

And now, while browsing through a book, I thought this is a perfect match:
It is "Pfauter Wälzfräsen" Damned Krauts you think? Well, the book says that it has been translated to English too. 606 pages dedicated to hobbing (that is "Wälzfräsen").
I don't think there is a better book. As you might know, Pfauter was and is the German reference for gear cutters.

Pfauter is the author, the book was published by Springer Verlag, year is about 1975. Maybe that helps finding the English version.
It has two volumes, but maybe you can ignore the second volume with the chapters "Accuracy, Economy". Maybe that vol. two didn't even make it to the market.


Nick

John Stevenson
12-31-2010, 09:14 AM
Got one of the Pfauter books, don't know if it's that one, certainly not 600 pages, probably only 120 or so.
good book and covers a lot of the machines so it's not a generic hand book.

Got about 6 foot of shelf space on gearcutting books and associated works as it's something I have been interested in for years,
Not an expert by any means , still learn new things all the while.

One nice book I have is called "On the teeth of wheels" by Cammus, written in 1750 and translated from the French in 1806
This is one of the first texts on gears and is still very valid today.

You can read it on Google.

http://books.google.com/books?id=8StHQTqK64IC&dq=hawkins+teeth+of+wheels&printsec=frontcover&source=web&ots=WIdbAtX57z&sig=Rr_Idd4VOp5Up5FCSiYCqf8bJyA#v=onepage&q&f=false

And modern reprints are available cheaply, mine is an original 1806 copy I got off Ebay for believe it or not £7.00

MuellerNick
12-31-2010, 09:22 AM
Got one of the Pfauter books, don't know if it's that one, certainly not 600 pages, probably only 120 or so.


That '75 book is an extended version of the initial issue of 1933. Don't know how many pages that one had.
Stupid thing is, I ordered that book for a friend and now have to tell him that it was already "sold to someone else". :D


Nick

John Stevenson
12-31-2010, 09:29 AM
That '75 book is an extended version of the initial issue of 1933. Don't know how many pages that one had.
Stupid thing is, I ordered that book for a friend and now have to tell him that it was already "sold to someone else". :D


Nick

Like the Scotch :D

MuellerNick
12-31-2010, 09:46 AM
Like the Scotch


Now obviously the deal that you will get the re-sent bottle won't work anymore.



Nick

lazlo
12-31-2010, 09:57 AM
This would be a perfect project for a small Mitsubishi servo motor. They have a setting for an encoder input. Then with the electronic gearing parameters you can easily set it to just about any ratio you can think of.

From my thread about retrofitting a Bridgeport J-Head with a Mitsubishi servo. November, 2008. I'm behind schedule :D

http://bbs.homeshopmachinist.net/showpost.php?p=394576&postcount=14


"I was cobbling together the various bits I needed for a servo-driven gear hobber, and I was trying to figure a clean way to mount a shaft encoder to the mill head so I could synchronize the hobber spindle. Since a servo motor already has an high-resolution encoder, and the Mitsubishi servo drives are built for Master/Slave (follower) mode, it seemed like a really clean way to add a high performance motor with the big side-benefit that I can literally plug-in an electronic gearing system for gear hobbing, or cutting helix's on the mill."

lazlo
12-31-2010, 10:04 AM
So you only need to do 20,000 fixed-point divides per second

It's not even necessary to tax the processor that much. Do all your calculations in the setup phase, which will give you a 'magic number'. Start counting pulses on your input; when they reach the magic number send a pulse on the output and reset the counter. I have seen an AVR routine written in C (not assembler) that reads quadrature pulses and correctly updates a counter that is 23 machine instructions long.

Clever! You divide the steady-state ratio for the spindle and gear hob RPM once, and you just keeping track of the deviation.

The Mitsubishi and Yaskawa servo controllers with electronic gearing have pretty powerful 32-bit microcontrollers, and the gear ratio is specified as a floating-point value. I have no idea if they have floating-point units, but considering that they're doing the gear divider calculation with 1,000,000 pulses per revolution, it must.

John Stevenson
12-31-2010, 10:41 AM
Just dug the book out, it's called Pfauter Handbook of Hobbing and Hobs, 288 pages

BUT no date, I thought you Germans were organised ?
Looked all thru this, no publication date, no printing date, no idea what edition.

Only clue on the title page, top right it says:-

1276
DK 621,914,1
DK 621,914,2

???????????

MuellerNick
12-31-2010, 10:55 AM
BUT no date, I thought you Germans were organised ?


Naa, that is the English version! They left out the German correctness not to confuse the rest of the world.
I bet, they intentionally included some errors hard to spot. **Harharhar**


... and a happy new year!
Nick

Spin Doctor
12-31-2010, 11:09 AM
Many commercial hobbing machines are now all electronic and don't rely on gearing.
I suppose it's a very grey area now on what's a hobbing machine and what's a 7 axis CNC ?

We had some really ancient Lees-Bradners that where used to cut the drive helical on cam shafts that drove both the distributer and the oil pumps. They got replaced with CNC Phauters. Just like on Cam Grinding machines, they just load a different program in to get a different profile instead of having to tear the work head down to install fresh master cams.

TexasTurnado
12-31-2010, 01:32 PM
From my thread about retrofitting a Bridgeport J-Head with a Mitsubishi servo. November, 2008. I'm behind schedule :D

http://bbs.homeshopmachinist.net/showpost.php?p=394576&postcount=14


"I was cobbling together the various bits I needed for a servo-driven gear hobber, and I was trying to figure a clean way to mount a shaft encoder to the mill head so I could synchronize the hobber spindle. Since a servo motor already has an high-resolution encoder, and the Mitsubishi servo drives are built for Master/Slave (follower) mode, it seemed like a really clean way to add a high performance motor with the big side-benefit that I can literally plug-in an electronic gearing system for gear hobbing, or cutting helix's on the mill."

I didn't realize this project dated back that far. :D I don't feel so bad now to still be working on the restoration of the Colchester and Graziano....;)

lazlo
12-31-2010, 01:44 PM
I didn't realize this project dated back that far. :D I don't feel so bad now to still be working on the restoration of the Colchester and Graziano....;)

Yeah, I suck. :p Between a demanding job and two young children...

You better get a move on John -- I caught up on a bunch of projects over Christmas ;)

LES A W HARRIS
12-31-2010, 02:00 PM
TT.

I have a DivisionMaster which drives a 400 1/2-step stepper drive which is direct-coupled to my 6" Vertex 90:1 rotary table.

360/400 = 0.09 degrees rotation of the stepper and my rotary table worm drive. The 90:1 worm and wheel drive converts that 0.09 degrees to 0.09/90 = 0.001 degrees table movement.

0.001 degrees x 60 = 0.06 minutes x 60 = 3.6 arc seconds the tan of which is 0.0000175 or ~ 1:57,300

I was that pleased with it that I asked the developer if he had a kit for my 8"Vertex" 90:1 rotary table but he said that the rotab was OK but it would not handle the rotab with its 3-jaw chuck on.

If you read the pdf link you will see an example of what Weston Bye is saying.
Hope all had a good Xmas, boy you guys have been busy, trying to catch up.

Old T, numbers clash? 90T X 400Steps = 36,000 steps / 360° = 100 steps per deg of rotation = 0.01° (Y/N) ? = 0.000175 radians times 4" radius of rotab =0.000698" per step at 4" radius. (Y/N) ? Am I having another senior moment?

Cheers, Les.

TexasTurnado
12-31-2010, 02:49 PM
Yeah, I suck. :p Between a demanding job and two young children...

You better get a move on John -- I caught up on a bunch of projects over Christmas ;)

That's great, Robert! Unfortuneately, I have not even been in the shop the past 10 -14 days as I have a cough that I can't get rid of and the doctors have not been especially helpful. I just feel so lousy I don't want to go to the shop, so you know it's serious...:D I just hope this gets sorted out before I come down with pneumonia like I had in '95 - spent a week or so in the hospital and thought I was gonna die....

djc
12-31-2010, 03:26 PM
The Mitsubishi and Yaskawa servo controllers with electronic gearing have pretty powerful 32-bit microcontrollers, and the gear ratio is specified as a floating-point value. I have no idea if they have floating-point units

As Macona says above, the Mitsu. ones at least require the gear ratio to be expressed as a fraction, hence no floating point maths is necessary*. Have a look at p.128 if you're interested in this sort of thing.

http://www.hitex.se/robot/Manuals/MR-J2S-A%20ver.C%20-%20ENG.pdf

*From http://www.guidenet.net/resources/programmers.html "Real Programmers scorn floating point arithmetic. The decimal point was invented for pansy bedwetters who are unable to 'think big'."

macona
12-31-2010, 03:41 PM
Macona -

I'm not familiar with the type of motor that you describe but if I understand you correctly it would not require any separate electronic dividing to work and would provide smooth operation given the cpr needed from the hob mounted encoder? If so that would be a very interesting way of building a hobber.

A servo motor is a closed loop motion actuator. The motor itself can be something as simple as a permanent magnet DC motor, brushless AC, or something more exotic like proportional hydraulics. They can be linear or rotary as well. The key is feedback. As the actuator does its thing there is position data that goes back to a control where the signal is compared against where the control says it should be and the actuator position is changed to meet this set position. So the control always knows where the motor position is unlike a stepper where it's "it should be there".

The motors I am talking about are made by Mitshbishi. The servo drive itself has a microcontroller in it which does the entire feedback loop. The motor is a AC brushless motor with a 131072 line encoder. This means the motor can stop and hold a position at any one of those points. Or at least within a count or two. The A series drives have pulse inputs just like a gecko drive, they also have a quadrature input as well where you can connect them to an encoder or another servo drive.

So you could do a hobber a couple ways. Use a standard motor to run the spindle with an encoder or use a servo motor for the spindle as well. The servo drive outputs a quadrature signal for feedback just like an encoder so you could gear off that as well.

oldtiffie
12-31-2010, 06:15 PM
Originally Posted by oldtiffie
TT.

I have a DivisionMaster which drives a 400 1/2-step stepper drive which is direct-coupled to my 6" Vertex 90:1 rotary table.

360/400 = 0.09 degrees rotation of the stepper and my rotary table worm drive. The 90:1 worm and wheel drive converts that 0.09 degrees to 0.09/90 = 0.001 degrees table movement.

0.001 degrees x 60 = 0.06 minutes x 60 = 3.6 arc seconds the tan of which is 0.0000175 or ~ 1:57,300

I was that pleased with it that I asked the developer if he had a kit for my 8"Vertex" 90:1 rotary table but he said that the rotab was OK but it would not handle the rotab with its 3-jaw chuck on.

If you read the pdf link you will see an example of what Weston Bye is saying.


Hope all had a good Xmas, boy you guys have been busy, trying to catch up.

Old T, numbers clash? 90T X 400Steps = 36,000 steps / 360° = 100 steps per deg of rotation = 0.01° (Y/N) ? = 0.000175 radians times 4" radius of rotab =0.000698" per step at 4" radius. (Y/N) ? Am I having another senior moment?

Cheers, Les.

Thanks for the over-sighting and heads up Les.

You are quite correct.

I was either either tear-ar$ing or careless - or both - and should have been doing neither as I (should?) know better than that. Oh well - burned fingers (and tail-feathers?) - again!!

I will re-do it.

The stepper has 200 steps/revolution = 360/200 = 1.8 degrees/step ie at the worm as the stepper is close-coupled to the worm spindle.

The reduction at the table is 1:90 revs of table : revs of stepper.

The angular rotation of the table per step of the stepper is:
1.8/90 = 0.02 degrees but as the DivisionMaster "half-steps" for postional accuracy an no error (disregarding mechanical error in the rotary table) will be larger that half a step, that accuracy will be 0.02/2 = 0.01 degrees.

(And that is where my error was - out by one decimal place for which I have severely admonished myself for a silly mistake).

Thanks for pointing it out.

0.01 degrees x 60 = 0.60 minutes x 60 = 36 seconds the tan of which is 0.000175 the inverse of which gives the error = 1:5,730 or 0.000175/1,000 = 0.000175 (ie 1.75 "tenths") per inch.

As I presume you've read the pdf file I linked to, you will have seen that that stepper on that rotary table is very accurate and very versatile. No need for dividing plates etc. and the machine just does as its told. As I always rotate the table clock-wise I have no issue with back-lash and the stepper is slowed approaching the required setting and accelerates from the current point to a steady pre-set speed to the next point.

The cutting/milling surface speeds for rotary milling are easily set over a wide range and its relatively quick and easy to set the pulses/second to get the peripheral angular velocity to suit the feed I require in either mm/sec (or minute) - my preferred way - or inches/second (or feet or inches per minute).

It takes no time to set up at all - just bolt it to the mill table, plug it in, set or check settings (it has non-volatile memory) and away I go.

I bought it after a very good report on it by WBleeker (Will) a member here in OZ and I have never regretted it.

It has met all my requirements.

One of the big advantages of it is that I no longer have to work in decimals and get decimal degree answers that I have to convert to degrees:minutes:seconds to use it on a standard rotary table.

But.

I had a job that I didn't want to remove set up on the "electronic" rotab and another job for the rotary table turned up - so I bought another one - and have to use or convert decimal degrees to D:M:S again!!!. But I have to do that with my 8" Vertex rotary table.

oldtiffie
12-31-2010, 07:02 PM
Originally Posted by oldtiffie
I'm not so sure about that.

A hobbed gear which turns at one tooth per revolution of the cutter will have as many flats or facets are there are teeth on the cutter per rev.

Unless they are big teeth, 12 or more facets will be so close to a smooth involute curve that it really won't matter for a lot of work otherwise leave a shaping/shaving/grinding allowance.


Actually, that isn't all of it, really, since there can be more than just one pitch of the hob cutting on one tooth.... one pitch of the hob cuts on as much of the tooth flank as 'slides past" it.... but it takes at least a couple or three turns for the tooth flank to completely get out of contact.

probably one tooth space is worked on during 5 or 6 turns total, which gives 2 to 3 turns for each flank. maybe more, I have not examined it closely, I just went and looked at a hob vs a gear roughly meshing them.

So you need to look at the number of teeth total that cut on one flank of the tooth, which may not be identical with the number in one revolution. THAT is how many "facets" would be cut, and it could easily be double the number of teeth per rev.

That is a very good question JT.

I am still pretty sure I was right - but that was "quite" sure until I read your post and down-graded my confidence level a bit.

You are quite correct that as the cut starts and there will be minimal depth of cuts and only the tips of the hob will be cutting. As the the job is progressively fed into the hob, the arc of the job in which the hob is cutting progressively increases so that more teeth are cutting at any one time.

When full depth of cut is arrived at there will be minimal (1?) teeth cutting (rubbing) as would be the case of a gear tooth in a rack (with neither gear nor rack having other than "basic" teeth) as only one tooth will bear on the the rack teeth at the pitch points. All other teeth of the gear will be clear of the rack.

To take your very well-made point, I concede that there may have been several - but not many - teeth cutting the final stages and as each flank/side of each hobbing cutter is straight that tooth will cut a "flat".

Successive "flat" cutter teeth will cut a succession of flats (facets?).

If the number of flats/facets on the gear teeth is large enough they will form a very close approximation of a smooth involute curve on the gear teeth.

Too many cutter teeth per rev of the cutter may well not have sufficient depth of cut to "bite" and may "rub" and either work-harden the job and/or blunt the cutter (which will "rub" and may work- harden the job and.or blunt the cutter - etc.etc.).

There will always be a compromise between what is "perfect" and what is "adequate" in the circumstances - and will have to be evaluated and assessed on a job-by-job basis.


probably one tooth space is worked on during 5 or 6 turns total, which gives 2 to 3 turns for each flank. maybe more, I have not examined it closely, I just went and looked at a hob vs a gear roughly meshing them.

Good points.

A hob is exactly like a gear at an angle that is normal to the helix angle.

In theory at least a large enough hob will cut a rack with the same DP (rack and gear) as the circular pitch of the hob (gear?) is (must be?) the same as the linear/lineal pitch of the rack (which is really only a very small part of a very large gear).

But back to your post - I need to go back and do a bit more "looking(-up)".

dp
12-31-2010, 11:08 PM
The DAC is only used to initially set the DC voltage in the ballpark to avoid false lock points - after that the phase lock loop takes over and adjusts the voltage to exactly what is needed for lock at the correct rpm.....

That's worth knowing more about. How is the DAC decoupled from the PLL at lock? There have been some exotic dual PLL schemes to do this in the past and I confess to not having done this in at least 20 years. Is there a block diag to peruse?

TexasTurnado
01-01-2011, 02:02 AM
That's worth knowing more about. How is the DAC decoupled from the PLL at lock? There have been some exotic dual PLL schemes to do this in the past and I confess to not having done this in at least 20 years. Is there a block diag to peruse?

I have to confess it has been about that long since I have actually used these. :) Here is what I remember: the part was made by Motorola and was a hybrid: the logic was ECL based to be very fast and the rest of the IC used high speed analog circuitry. The ECL based phase comparator put out a very fast rise time narrow pulse when late compared to the reference pulse, and did nothing it it was early or on time. The narrow pulse injected a few pC of charge into the integration capacitor, thus raising its voltage by a small amount. One pulse would not achieve phase lock, but many pulses of charges in succession would - then it was a matter or adding a pulse of charge as needed to maintain lock.

The integrator was made intentionally lossy by a parallel resistor, to make sure charge would always be periodically required to maintain phase lock. One way of incorporating the DAC output to preposition the integrator was to tie one end of the loss resistor to the DAC output. The lookup table would be selected to place the oscillator always low in frequency compared to where it needed to be for phase lock to occur (for worst case conditions).

However, there was also a digital output that indicated phase lock had occured, which could be used to disconnect the DAC output if desired - in that case two loss resistors were used, one permanently connected. Before lock occurred, there was a ramp generator circuit that could be used to sweep the integrator voltage across the range of voltages where phase lock occurred for the different divider ratio settings.

As for DAC jitter, typically the number of preset voltages needed to avoid false lock (sometimes only two points were used - low end and high end) was much lower than the capacity of even the smallest lookup table host, so the lsb was very insignificant compared to the set voltage. In some cases, it was possible to avoid the DAC altogether, and just tie divider resistors to some of the digital divider input bits. In any event, any DAC jitter was canceled out by the automatic adjustment of the number of phase lock charges.

One potential problem area with using this scheme with a DC motor may be the number of pulses available - the circuits I previously designed typically had millions of pulses available, so the charge added by a single pulse could be very small. OTOH, the bandwidths used in the phase lock loop were much higher there, to achieve the best phase noise sidbands on the output signal. For a gear hobber, I doubt the phase noise sidebands of the blank angular velocity would even be measurable. :D

OTOH, I once had to measure and prove to the customer the noise sidebands on a multigigahertz synthesizer (with cesmium beam reference :) )measured only 1 hz to 150 hz from the carrier were below a specified value, so some seemingly impossible measurements can be made when they have to be......

dp
01-01-2011, 03:15 AM
That sound like an allegory to my solution which was to introduce noise into the DAC/PLL integrator to swamp the long-term lsb jitter. The theory being if you don't like the internal long term signature, put it on a shake table :)

djc
01-01-2011, 04:18 AM
The motor is a AC brushless motor with a 131072 line encoder.

It would be awful embarassing if the little man scribing the graduations lost count and missed one (131071 is prime).

131072 is 2 to the power of 17. Computers like multiplying and dividing by two.

macona
01-01-2011, 04:30 AM
I think the encoder in there is actually a 32768 line encoder which would be 15 bits. With quadrature reading they get the 131072 number. Then its sent serially to the drive. The new MR-J3 series drives have double that amount, 262144 steps per rev. But these are so new that I have only seen maybe two drives show up on ebay.

willmac
01-01-2011, 06:56 AM
OldTiffie-

"When full depth of cut is arrived at there will be minimal (1?) teeth cutting (rubbing) as would be the case of a gear tooth in a rack (with neither gear nor rack having other than "basic" teeth) as only one tooth will bear on the the rack teeth at the pitch points. All other teeth of the gear will be clear of the rack."

I'm not sure what you are saying here, so it may be that I have misundestood. However -- as the hob enters the work, the depth of cut will increase until the centre line of the hob is perpendicular to the end face of the gear. At that point the hob is full depth. It is not true to say that at that depth only one hob tooth is cutting. Generating cutting action continues for that tooth space and is occuring for the teeth on either side. I think that this is the point that J Tiers was making. This is the essence of hobbing.

oldtiffie
01-01-2011, 08:22 AM
Thanks for the extra info Bill.

I've conceded a lot to JT and given that on re-thinking it, I can't but agree with you as well.

To that extent I'm a lot less certain than I was.

As all hob cuts are abutting "flats", the number of those flats (facets?) will determine how close those multi-flats approximates a true involute curve. I guess that the more teeth (per rev) on the hobbing cutter the better the "curve" but there are practical limits on that.

If the curves on a standard involute cutter are true curves (but necessarily true involute curves even though perhaps very close to it) it may well be that the standard involute cutter will be closer to or a better curve than that produced by some hobs.

All of this assumes that all teeth on the hob are cutting equally. This may not be so if the axis of the hob is not that of the milling arbor - ie if the arbor is bent, as is indicated by the dreaded "thrumm - thrumm" etc. as only part of the hob (teeth) is/are cutting, some are rubbing and others are missing altogether. Poor arbor suspender/support brackets and poor journals and bearings don't help either.

But this is not restricted to universal mills either as its common on cantilevered stub arbors held in collets in a vertical mill.

It won't take much of a knock to slightly bend the 8mm arbor and get a similar "thrumm" etc. on the mini-hobber here either. And if that induces rubbing or chatter etc it will quickly ruin the hobbing cutter cutting edges.

Less than rigid set-ups of the work and the hobbing cutter won't help either.

Thios is not a great deal different to an end-milling cutter "running out".

willmac
01-01-2011, 08:43 AM
OldTiffie-

The accuracy of mounting of a hob to the arbor needs to be at least an order of magnitude better than is usually acheived on a horizontal mill arbor. They really are not comparable. A hob needs to be running with an eccentricity of less than a couple of tenths TIR - better if possible. The same tight conditions apply to the gear blank.

That is the reason for one of the criticisms that I had of the original tool being discussed in this thread. Both the hob and work axes are overhung, with very small diameters. I expect the deflections that would occur under normal cutting conditions (excluding plastics) would be more than the TIR above. In that case, the accuracy that is dearly bought in a hobbing cutter would go to waste.

oldtiffie
01-01-2011, 09:04 AM
True Bill - all too true.

When I had to hob on a universal mill it was a solid Cincinnati and we had a special arbor with its own set of spacers and journals which the - again -special support (bronze) bearings were scraped and adjusted to - with at least two arbor supports as close to the hobbing cutter as we could get them without fouling the job.

Hobbing is a big load on a mill as its not a lot different to "ganged" common involute cutters. Its worse than slab milling as the hob has zero top and side rake as do common involute (and most "form") cutters. The slab cutter has positive top rake as well as a pretty good spiral and "shears" the swarf off instead of pushing it off.

Given that involute cutters are as cheap, effective and accurate as they are I can't ever see the need for a hobber in my shop.

That is especially so as regards the mini-hobber which is the subject of this thread.

J Tiers
01-01-2011, 10:44 AM
T

Given that involute cutters are as cheap, effective and accurate as they are I can't ever see the need for a hobber in my shop.

That is especially so as regards the mini-hobber which is the subject of this thread.

The first big advantage is that you need only one hob per pitch , and can then generate ANY gear with it.

if, as someone mentioned, a hob for the little machine can cost only about $50, while the involute cutter set will cost about $200 for the set of 8, it could be a lot cheaper in cutters to do the hob.

if the machine is a lot more expensive, you'd need to be making a lot of gears to make any sense economically.

The other advantage is "generating" the gear tooth shape, as opposed to form-milling, which is , as pointed out repeatedly here, is truly correct (if ever) for only one tooth count. Even if faceted to some degree, and I suspect that is not so bad as it might be, the basic shape is good, and some wear will likely knock down the "odd bits over".

The form cutter route can be either 'store boughten", or "shop made". The "store boughten" type supposedly are better, and maybe so if B&S made them, but maybe they were nade "as good as needed by amateurs", or DP types made "to the nearest metric equivalent". In that case, perhaps they are not so good.

Shop made might be made, as I have done, to fit an existing gear which has at least one toothspace undamaged (single tooth, used on shaper). or it might be made by the "button cutter" method. Both are approximations, one at least can theoretically reproduce the gear, one will "get close".

Any way you slice it, the hob will do at least as well, and probably considerably better, to say nothing of faster.

lazlo
01-01-2011, 11:07 AM
I think the encoder in there is actually a 32768 line encoder which would be 15 bits. With quadrature reading they get the 131072 number. Then its sent serially to the drive. The new MR-J3 series drives have double that amount, 262144 steps per rev. But these are so new that I have only seen maybe two drives show up on ebay.

The optical encoders have 1/4 of the lines (hash patterns, actually) of the rated resolution/pulse rate. Each quadrature channel (A, B) is generating pulses at the actual resolution of the optical encoder. The B channel is the A channel phase shifted by 90°. Most high-end servo controllers, including the Mitsubishi, count the rising and falling edge of each pulse on both channels, generating a 4X total pulse count.

So the optical encoder in the Mitsubishi J2's has 16K "lines" (quadrature hashes), and the J3 has 32K "lines".

lazlo
01-01-2011, 11:19 AM
As Macona says above, the Mitsu. ones at least require the gear ratio to be expressed as a fraction, hence no floating point maths is necessary

The "Fraction" as you say is specified as the RPM of the driver (Master) and the driven (slave) controllers.

It's certainly possible that's done with fixed-point math, especially since you're talking 100's of thousands of pulses per revolution, but it's not clear from the manuals or the sales literature. You need floating-point math to calculate the PID loop, and the Mitsubishi's do real-time resonance/vibration analysis, so it wouldn't surprise me if they're using a 32-bit controller with a floating-point unit.


*From http://www.guidenet.net/resources/programmers.html "Real Programmers scorn floating point arithmetic. The decimal point was invented for pansy bedwetters who are unable to 'think big'."

LOL! That's a hilarious page! Some other quotes:


"Real Programmers don't write specs -- users should consider themselves lucky to get any programs at all, and take what they get.

Real Programmers don't comment their code. If it was hard to write, it should be hard to understand.

Real Programmers disdain structured programming. Structured programming is for compulsive neurotics who were permanently toilet trained. They wear neckties and carefully line up sharp pencils on an otherwise clear desk."



By the way, I'm very familiar with floating-point math. I designed the Tejas New Instructions, which eventually became SSE 4.1 :)

John Stevenson
01-01-2011, 11:26 AM
"Real Programmers don't write specs -- users should consider themselves lucky to get any programs at all, and take what they get.

Real Programmers don't comment their code. If it was hard to write, it should be hard to understand.

Real Programmers disdain structured programming. Structured programming is for compulsive neurotics who were permanently toilet trained. They wear neckties and carefully line up sharp pencils on an otherwise clear desk."



Well that describes EMC and Linux to a TEE, was going to put a smily in but then realsised how true and serious this was - so :mad:

lazlo
01-01-2011, 11:32 AM
Well that describes EMC and Linux to a TEE, was going to put a smily in but then realsised how true and serious this was - so :mad:

John, you Linux kernel hacking now? Isn't there a maximum age for that? :D

How are you liking EMC compared to Mach?

John Stevenson
01-01-2011, 11:43 AM
It's on hold, keep giving up on it then trying it again.
It works but it's not easy for anyone who isn't a techno geek.
I have a fried who runs a bit IT section and he's put it on his lathe and router.

He's had a lot of problems getting it setup, lathe mainly and every question he's asked on the Linux group he's had to re ask it at least 2 or 3 times because the freely given answers are so technical.
The problem is these guys understand everything about EMC and Linux and expect everyone else to do so.

It's written from a programmers point of view, not an engineers.
I think it's only just been put in the latest release, the ability to change a broken tool part way thru a program.
Up until recently you had to bail out and start again with a new tool.
When questioned on this they said Why would you need to change a tool part way thru a program.

When it was explained why there stock answer is the source code is there, change it to suit.

many people have with the result there are many flavours out there.

Les Newall of Sheetcam fame put the toolchange in and passed it up the line over two years ago only for it to suit there because it wasn't on the committee agenda.

TexasTurnado
01-01-2011, 02:11 PM
This would be a perfect project for a small Mitsubishi servo motor. They have a setting for an encoder input. Then with the electronic gearing parameters you can easily set it to just about any ratio you can think of.

I think I still have a 200w motor and drive for 120v. I know I also have another set from a different company.

Add: Just dug through my servo box. Found a Misubishi MR-J2S-20A1 servo drive with the motor, 200 watt ~1/4HP. It has a 131072 count encoder. The gearing setting allows you to set the ratio as a fraction with the numerator and denominator being any integer between 1 and 65536.

I suspect you are quite right, but I have not yet developed a conversation level of understanding with servo design and implementation. :D I looked through the manual posted earlier and was thinking the same thing, but there are so many things in there I don't understand well enough to proceed on my own. Would you like to elaborate on how this might be accomplished and the pitfalls to watch out for?

It would appear that line of servos is capable of running to at least 3000 rpm - is the encoder used capable of outputing 3000 x 131072 =393,216,000 pulses per minute (6,553,600 pulse per second)? That seems very fast to me for a mechanical device, even with 4x electronic multiplication - is there something else going on?

Which brings another question to mind: if the encoders in that line of servos is really that fast, would it be possible to use one servo in a passive mode (motor power off) just for the encoder and connect it to the hob shaft (I'm thinking of the setup Sir John used here) - then use a second servo with its electronic gear function for the motor to drive the blank through the gearbox?

Zahnrad Kopf
01-01-2011, 03:07 PM
Hobbing is a big load on a mill as its not a lot different to "ganged" common involute cutters. Its worse than slab milling as the hob has zero top and side rake as do common involute (and most "form") cutters. The slab cutter has positive top rake as well as a pretty good spiral and "shears" the swarf off instead of pushing it off.

:rolleyes: One cannot help but wonder how one could actually qualify such an assertion. Not to ask what means one would use to do so. Rather, how one might expect it to be accepted.


Given that involute cutters are as cheap, effective and accurate as they are I can't ever see the need for a hobber in my shop.

The contents of that statement would be completely relative on each, at best, given that hobs can be had equally as inexpensively, certainly more "effective", undeniably more accurate (all other concerns being equal), and "need" being a personal choice or concern. In fact, given these hobby level machines' availability and the discussed electronic options like John and others have produced, one could posit to wonder why another would bother with involute cutters at all...

lazlo
01-01-2011, 03:32 PM
It would appear that line of servos is capable of running to at least 3000 rpm - is the encoder used capable of outputing 3000 x 131072 =393,216,000 pulses per minute (6,553,600 pulse per second)? That seems very fast to me for a mechanical device, even with 4x electronic multiplication - is there something else going on?

No, that's right. Far more impressive is that they're doing the PID loop at that frequency!
That's why they have powerful 32-bit microprocessors. They never mention which one, but I still have on in my office that I took apart when the iGBT died.


Which brings another question to mind: if the encoders in that line of servos is really that fast, would it be possible to use one servo in a passive mode (motor power off) just for the encoder and connect it to the hob shaft (I'm thinking of the setup Sir John used here) - then use a second servo with its electronic gear function for the motor to drive the blank through the gearbox?

Sure, you could: I have two Mits servo set-up on my bench in Master/Slave mode: the quadrature output of the "Master" servo/controller is fed as input to the "Slave" servo/controller. With the motors commanded to 0 RPM, I can turn the spindle on the master, and the slave will follow.

This is all turn-key with no external electronics.

But there's no point to using the servo as an encoder. Just buy a Heidenhahn, or Tamagawa optical encoder on Ebay for $100. That's what John shows in one of his pictures on the Victoria. They are the optical encoders used inside the Mits and Yaskawa servos. You can buy Heidenhan's with several hundred thousand pulses per revolution, if that turns you on :)

John Stevenson
01-01-2011, 03:37 PM
I have posted gears cut with involute cutters on here.
I have shown how to make button cutters and then the cutter, stub tooth to boot on here.
I have shown hobbed gears on here
I have shown hobbed helicals and specials on here
I have shown worm and wheels on here.
I have shown spur gear cut with just an end mill on here.
I have shown helicals cut with just an end mill on here.

I am quite prepared to help other who want to help themselves and to have conversations with people who genuinely are interested in gears and gear cutting.

I am NOT interested in having theoretical discussions about hobbing with people who make out they know what they are talking about but in fact can't understand what they are reading.

Zahnrad Kopf
01-01-2011, 03:52 PM
I am quite prepared to help other who want to help themselves and to have conversations with people who genuinely are interested in gears and gear cutting.

That would be awesome... count me in. :) http://bbs.homeshopmachinist.net/images/icons/icon14.gif

How's this - I'm interested in playing around with this electronic hobbing, having all the bits and pieces seemingly needed (Bridgy, horizontal milling attachment, dividing head, plenty of hobs, etc...), save for the "electronics". Has there been produced a simple grocery list of bits and baubles needed for a simple start to make the electronic part and assemble it, anywhere? Originally, I'd mused about doing similar in a mechanical fashion by attaching a gear to the hob's arbor and driving the dividing head (or gear reducer) from it but having a hobber have no real need for it, but do remain interested in the electronic version...

Happy New Year.

snowman
01-01-2011, 05:02 PM
I am quite prepared to help other who want to help themselves and to have conversations with people who genuinely are interested in gears and gear cutting.

Well, if it's any consolation, I consider myself one of your stalkers. In my gear cutting folder I have forum conversations that you have had here, modelgeeks, MEW, etc. I also have conversations from Forest Addy from PM.

I just ordered Ivan Law's book, just downloaded A practical Treatise on the milling machine, and have just ordered my spindle encoder (already have the hobs in the basement)

Weston doesn't know it yet, but I'm about to start stalking him too...and he should be scared, there's only about 60 miles separating us...and I drive 50 of those miles on my way to work!

I can however tell you one poster that hasn't had a single note printed or written down.

So, your intended audience is thankful, and echo's of your sentiments can be heard reverberating throughout the community.

snowman
01-01-2011, 05:05 PM
Has there been produced a simple grocery list of bits and baubles needed for a simple start to make the electronic part and assemble it, anywhere?

There is. Model Engineers Workshop edition 108 & Errata in 110.

If you email me, I can get you the required files, and a domestic supplier (one of the chips is harder to find). jrunyan02 at yahoo . com

lazlo
01-01-2011, 05:29 PM
I am NOT interested in having theoretical discussions about hobbing with people who make out they know what they are talking about but in fact can't understand what they are reading.

Um, are you referring to the sub-thread that wjr, TexasTurnado, Macona and I are having? We're basically talking about how to replicate your electronic divider hobber with off-the-shelf servo components :)

MuellerNick
01-01-2011, 05:51 PM
(6,553,600 pulse per second)? That seems very fast to me for a mechanical device, even with 4x electronic multiplication - is there something else going on?


You can get encoders with something like 18000 lines / revolution. But that is a sine/cosine-signal. It can be scaled up by flash-ADCs. Increase is up to 50 times with a chip from iC-Haus. But then, you are reaching the maximum frequency. That chip can have 200 kHz input with 5* interpolation, or 20 kHz with 50*.
3000 RPM with that resolution? Forget it!
18000 * 50 * 3000 / 60 = 45 MHz! :D

Maybe next week, I'll get a used Heidenhain rotary encoder with that resolution. In return, I'll have to make the electronics to convert from sine 1Vss to digital (A, B & I).

Such an encoder costs well 2000 € new. :eek:



Nick

MuellerNick
01-01-2011, 06:04 PM
Addition:
The A, /A, B, /B, I, /I signal is transmitted over RS422. The fastest chip I found does 30MBits. Standard is up to 10MBit.


Nick

John Stevenson
01-01-2011, 07:04 PM
Um, are you referring to the sub-thread that wjr, TexasTurnado, Macona and I are having? We're basically talking about how to replicate your electronic divider hobber with off-the-shelf servo components :)

Not at all, I'm following that as well, in a private email to wrj tonight to give him the circuit information on the hobber which was in MEW108 his email address gave him away as running a gearcutting company in the US so how can I accuse him of not knowing what he's talking about.

Macona is well know for dealing with servo's

BTW if anyone wants this info from MEW108 written by Brian Thompson I have permission from both David Clark and Brian to circulate this with no copy write issues. It is the full version of just the article from MEW108 with a hand written alteration on the veroboard for an omission in the original. PM me.

TexasTurnado
01-01-2011, 07:37 PM
No, that's right. Far more impressive is that they're doing the PID loop at that frequency!
That's why they have powerful 32-bit microprocessors. They never mention which one, but I still have on in my office that I took apart when the iGBT died.



Sure, you could: I have two Mits servo set-up on my bench in Master/Slave mode: the quadrature output of the "Master" servo/controller is fed as input to the "Slave" servo/controller. With the motors commanded to 0 RPM, I can turn the spindle on the master, and the slave will follow.

This is all turn-key with no external electronics.

But there's no point to using the servo as an encoder. Just buy a Heidenhahn, or Tamagawa optical encoder on Ebay for $100. That's what John shows in one of his pictures on the Victoria. They are the optical encoders used inside the Mits and Yaskawa servos. You can buy Heidenhan's with several hundred thousand pulses per revolution, if that turns you on :)

Cool.:cool: I need a shopping list to start looking for. :D

I was contemplating just using another servo because it already has a robust package with bearings, looks oil-tite, and has the exotic multiplication circuitry built in - thus would be more "turn key" than just getting the encoder. The power rating of this "master" servo would not be an issue, since it would be driven, so any of the models ( as in least costly :) ) could be used.

As for the "slave" servo, what power rating is needed? That would seem to depend on the ratio of the gearbox, and the size of blank/tooth size. I am interested in making gears in the 24 to 10DP range, so I think I probably need a larger size than most HSMs, who appear to be making smaller gears.

If the same 40 to 1 ratio gearbox is used as John did, and the "master" is coupled to the hob with a 1 to 1 ratio, then the smaller toothed gears are going to need the "slave" servo to turn faster than the "master". The posted manual seems to indicate this is possible, as the allowed ratios of the electronic gear are 1/50 to 50 - did I read this correctly?

TexasTurnado
01-01-2011, 07:41 PM
Well, if it's any consolation, I consider myself one of your stalkers.

Count me in that category also! :D I freely confess to being guilty of buying all of the gears I have used to this point....

TexasTurnado
01-01-2011, 07:50 PM
You can get encoders with something like 18000 lines / revolution. But that is a sine/cosine-signal. It can be scaled up by flash-ADCs. Increase is up to 50 times with a chip from iC-Haus. But then, you are reaching the maximum frequency. That chip can have 200 kHz input with 5* interpolation, or 20 kHz with 50*.
3000 RPM with that resolution? Forget it!
18000 * 50 * 3000 / 60 = 45 MHz! :D
Nick

So how do they get these servos to run that fast? Does the resolution automatically decrease as the shaft rpm increases? I did not see any documentation of that in the manual....:confused:

macona
01-01-2011, 07:52 PM
The optical encoders have 1/4 of the lines (hash patterns, actually) of the rated resolution/pulse rate. Each quadrature channel (A, B) is generating pulses at the actual resolution of the optical encoder. The B channel is the A channel phase shifted by 90°. Most high-end servo controllers, including the Mitsubishi, count the rising and falling edge of each pulse on both channels, generating a 4X total pulse count.

So the optical encoder in the Mitsubishi J2's has 16K "lines" (quadrature hashes), and the J3 has 32K "lines".


Yep, 4x32k = 128k -> J2S
4x64k = 256k -> J3

Too much champagne last night??? ;)

-Jerry

lazlo
01-01-2011, 07:57 PM
Too much champagne last night??? ;)

Guilty as charged! http://www.practicalmachinist.com/vb/images/smilies/drink.gif

lazlo
01-01-2011, 08:05 PM
So how do they get these servos to run that fast? Does the resolution automatically decrease as the shaft rpm increases? I did not see any documentation of that in the manual....:confused:

Most of the high-end servos use proprietary differential interfaces, not RS-422.

Servo encoder resolution has become a big marketing bullet point, so they're on the Moore's Law treadmill, and doubling every two years.
The Mits J3 series is getting long in the tooth (I have several from Ebay, so they've been around long enough to be common in the surplus channels), so you can expect the up-coming J4 series to be in the 500K PPR

Standard Fanuc servos have 1 million PPR, and 16 million PPR is an option (never seen one on Ebay though).

I think they're way past the point of diminishing returns, but you have to keep up withe the Jones'...

macona
01-01-2011, 08:27 PM
I suspect you are quite right, but I have not yet developed a conversation level of understanding with servo design and implementation. :D I looked through the manual posted earlier and was thinking the same thing, but there are so many things in there I don't understand well enough to proceed on my own. Would you like to elaborate on how this might be accomplished and the pitfalls to watch out for?

It would appear that line of servos is capable of running to at least 3000 rpm - is the encoder used capable of outputing 3000 x 131072 =393,216,000 pulses per minute (6,553,600 pulse per second)? That seems very fast to me for a mechanical device, even with 4x electronic multiplication - is there something else going on?

Which brings another question to mind: if the encoders in that line of servos is really that fast, would it be possible to use one servo in a passive mode (motor power off) just for the encoder and connect it to the hob shaft (I'm thinking of the setup Sir John used here) - then use a second servo with its electronic gear function for the motor to drive the blank through the gearbox?

The small servos up to 400w will actually run to 4500RPM. Their rated spec is 3000, maximum 4500 RPM. The larger motors can be over driven but not as fast.

Like others have mentioned its no problem for an encoder to run that fast. Its only 6mhz. And encoders are not mechanical, they are electro-optic, well, usually. In the case of the ones in the Mitsubishi servos they have a glass disc with a whole bunch of tiny lines. Light is passed through to a sensor that reads it. I am rather certain that the encoders are actually absolute encoders which know in one turn where it is exactly even after power loss.

The data is transferred back to the drive through a serial connection. Only 6 wires are needed for the H, J2, J2S and J3 series drives for connection. These not only provide the encoder feedback but configure the drive for the particular motor you are attaching.

You could use another servo motor as an encoder but that would be a waste of money. You can set the encoder counts per turn in the parameters up to about 4000 pulses per turn if I recall. I have never needed this so I am not to sure.

Whatever you do I would use a gearbox on the output of the servo motor. This will stiffen the output and give a major torque advantage. And you cant run the hob that fast anyway, its just HSS. I have a little 3:1 Bayside on the small 200w motor I have.

Generic quadrature encoders are pretty cheap. USDigital has a line of encoders and you can shop ebay for them as well. I have gathered piles of encoders now...

I am using a linear encoder similar to the one Nick was talking about in my laser cutter I am building. Heidenhain 478 with a IBV606B interpolator box.

I am also getting ready to re-retrofit my CNC mill. I have a couple of the MR-H series drives with 1kw motors for the X and Y axix and an older MR-J series 1.5kw for the knee. I am also going to replace the spindle motor with a servo motor as well but I am not sure which. I have a MR-H 2kw set and a MR-J2S 3.5Kw set that I had bought to put on the Monarch. But I also have a MR-H 5kw set that I may stick on the monarch and then use the 3.5kw motor on the mill and do a fixed 3:1 belt ratio to the spindle.

The main thing to watch out for is you really want one of the newer drives and motors. They have much better auto-tuning than the older drives. With mitsubishi I would say a J2 or J2S drives. At the end of the manual that was posted earlier there is a compatibility chart of motors and drives. I have found that the motors intended for the J2 series drives do work on the J2S. Motors for the J2 usually have a part number like HC-MF23 where the JS2 version of that motor is HC-MFS-23. The difference is the encoder. The J2 has lower count encoders (I think 16384 per turn) and the autotuning is more advanced on the J2S.

I have not messed with the J3 series. They look really cool though. They have some pretty neat anti-vibration features.

http://www.youtube.com/v/8JkFqzwlyiI

http://www.youtube.com/v/QBtrFDoqWTM

TexasTurnado
01-01-2011, 10:37 PM
The small servos up to 400w will actually run to 4500RPM. Their rated spec is 3000, maximum 4500 RPM. The larger motors can be over driven but not as fast.

Like others have mentioned its no problem for an encoder to run that fast. Its only 6mhz. And encoders are not mechanical, they are electro-optic, well, usually. In the case of the ones in the Mitsubishi servos they have a glass disc with a whole bunch of tiny lines. Light is passed through to a sensor that reads it. I am rather certain that the encoders are actually absolute encoders which know in one turn where it is exactly even after power loss.

The data is transferred back to the drive through a serial connection. Only 6 wires are needed for the H, J2, J2S and J3 series drives for connection. These not only provide the encoder feedback but configure the drive for the particular motor you are attaching.

You could use another servo motor as an encoder but that would be a waste of money. You can set the encoder counts per turn in the parameters up to about 4000 pulses per turn if I recall. I have never needed this so I am not to sure.

Whatever you do I would use a gearbox on the output of the servo motor. This will stiffen the output and give a major torque advantage. And you cant run the hob that fast anyway, its just HSS. I have a little 3:1 Bayside on the small 200w motor I have.

Generic quadrature encoders are pretty cheap. USDigital has a line of encoders and you can shop ebay for them as well. I have gathered piles of encoders now...

I am using a linear encoder similar to the one Nick was talking about in my laser cutter I am building. Heidenhain 478 with a IBV606B interpolator box.

I am also getting ready to re-retrofit my CNC mill. I have a couple of the MR-H series drives with 1kw motors for the X and Y axix and an older MR-J series 1.5kw for the knee. I am also going to replace the spindle motor with a servo motor as well but I am not sure which. I have a MR-H 2kw set and a MR-J2S 3.5Kw set that I had bought to put on the Monarch. But I also have a MR-H 5kw set that I may stick on the monarch and then use the 3.5kw motor on the mill and do a fixed 3:1 belt ratio to the spindle.

The main thing to watch out for is you really want one of the newer drives and motors. They have much better auto-tuning than the older drives. With mitsubishi I would say a J2 or J2S drives. At the end of the manual that was posted earlier there is a compatibility chart of motors and drives. I have found that the motors intended for the J2 series drives do work on the J2S. Motors for the J2 usually have a part number like HC-MF23 where the JS2 version of that motor is HC-MFS-23. The difference is the encoder. The J2 has lower count encoders (I think 16384 per turn) and the autotuning is more advanced on the J2S.

I have not messed with the J3 series. They look really cool though. They have some pretty neat anti-vibration features.

http://www.youtube.com/v/8JkFqzwlyiI

http://www.youtube.com/v/QBtrFDoqWTM

OK, I am impressed! In fact, I am blown away! Those are very impressive videos.... :D

I can also see I have a steep hill ahead of me to climb.... Thanks for all of the info in this post, macona!

I quoted the speed of 3000 rpm as a reference point: as you say, the hob cannot be turned that fast as it is HSS. And the "slave" servo may be tuning faster than the hob, but not the blank, which is after the gearbox.

I agree wholehartedly about using the gearbox - I just need to come to a decision of what ratio is best for this application. I have a 40 to 1 box I bought to rob the gears out of to make an indexing head that I could use, but procuring another with a different ratio is not a big deal. I believe the basic limitation is selecting a ratio that will give whole numbers within range for the values of the numerator and denominator in the electronic gear function - isn't that correct?

I looked through the manual again, and saw where the number of output pulses can be set - but it appears to me the electronic gear function has control over the number of input pulses per rev. If this is the case, then using an encoder of less than 131072/4 pulses will cause the servo to rotate at a slower speed (unless offset by the EG function setting) - is this your understanding?

macona
01-02-2011, 02:24 AM
You are not going to find an encoder that will have that high of count without spending quite a bit. In all reality you will only need an encoder around 1000 lines or so. Then you just figure out the ratio including the gear box and go from there.

lazlo
01-02-2011, 08:55 AM
I am rather certain that the encoders are actually absolute encoders which know in one turn where it is exactly even after power loss.

The J3 servos are all 18-bit absolute encoding. The J2 and earlier were available in absolute or relative encoders.

The battery option is a faux absolute position: the controller zeros-out a particular location, and then keeps track of the relative position from the encoder. That's why you need the battery. Your J2 does this -- see Chapter 5.


it appears to me the electronic gear function has control over the number of input pulses per rev. If this is the case, then using an encoder of less than 131072/4 pulses will cause the servo to rotate at a slower speed (unless offset by the EG function setting) - is this your understanding?

That's right - the electronic gear function is a fraction that you apply to the command pulse frequency coming from the Master (the gear cutter spindle motor). So if the Master is running at 300 RPM, and you set CMX to 20, and CDV to 1, the slave will be synchronized at 1/20th of the Masters' RPM, and you'll hob a 20 tooth gear.

lazlo
01-02-2011, 08:55 AM
You are not going to find an encoder that will have that high of count without spending quite a bit.

Yes, but he has a friend who has a pile of J2 and J3 servos with the matching controllers. That's why he's asking :)

TexasTurnado
01-02-2011, 01:06 PM
You are not going to find an encoder that will have that high of count without spending quite a bit. In all reality you will only need an encoder around 1000 lines or so. Then you just figure out the ratio including the gear box and go from there.

I have no problem using a 1024 line encoder (they seem rather common on ebay - and I saw some that were as high as 4096) if it will get the job done accurately and the ratios work out to be within the limits of the EG function. Perhaps this is a good point to discuss costs - if the encoders can be had for $100 (I saw some at that price, but they were 1024 or fewer ppr), how much do servo motors and their controllers go for?

I saw several encoders where the asking price was close to $2000 and it did not appear the line count was anything like 32k - so is it really cheaper to buy a high count encoder vs using another servo?

And, if using a servo motor for its encoder, does one also need a controller package for that servo? I assume the answer is yes, but I didn't think it would hurt to ask....:)

TexasTurnado
01-02-2011, 01:22 PM
The J3 servos are all 18-bit absolute encoding. The J2 and earlier were available in absolute or relative encoders.

The battery option is a faux absolute position: the controller zeros-out a particular location, and then keeps track of the relative position from the encoder. That's why you need the battery. Your J2 does this -- see Chapter 5.

For the proposed application, is absolute position needed? I would assume (always dangerous :) ) that as long as the power was not turned off to the servos, they would restart still in the same realtive position - so one could stop the hob to measuse the width of the teeth, for example, and then restart the hobbing process.

But if the power was shut off (or lost) overnight could one restart the next day and have the blank still in the correct position relative to the hob?



That's right - the electronic gear function is a fraction that you apply to the command pulse frequency coming from the Master (the gear cutter spindle motor). So if the Master is running at 300 RPM, and you set CMX to 20, and CDV to 1, the slave will be synchronized at 1/20th of the Masters' RPM, and you'll hob a 20 tooth gear.

I had considered the possibility that setting the output pulses to a new value would also cause a corresponding change in the number of pulses expected at the input, but could find no confirmation of that in the manual. Do you know if that is the case?

TexasTurnado
01-02-2011, 01:25 PM
Yes, but he has a friend who has a pile of J2 and J3 servos with the matching controllers. That's why he's asking :)

Wow, I can see 2011 is going to be a great year! How could it be otherwise with such a good friend? :D

Weston Bye
01-02-2011, 01:40 PM
Well, if it's any consolation, I consider myself one of your stalkers. In my gear cutting folder I have forum conversations that you have had here, modelgeeks, MEW, etc. I also have conversations from Forest Addy from PM.

I just ordered Ivan Law's book, just downloaded A practical Treatise on the milling machine, and have just ordered my spindle encoder (already have the hobs in the basement)

Weston doesn't know it yet, but I'm about to start stalking him too...and he should be scared, there's only about 60 miles separating us...and I drive 50 of those miles on my way to work!

I can however tell you one poster that hasn't had a single note printed or written down.

So, your intended audience is thankful, and echo's of your sentiments can be heard reverberating throughout the community.

Wave as you pass, as I drive 25 miles south on I-75 on my way to work near Pine Knob. PM me at your convenience and we'll compare notes.

Someday I would enjoy a get-together with the gang around here, but I'm not an organizer. If nothing else, I will be exhibiting at NAMES and invite you guys to stop by and say hello. Last year I had my Electronic Threading outfit operating. This year I hope to have my CNC efforts on display and running.

TexasTurnado
01-02-2011, 02:34 PM
Having given this some thought, I would like to see if you guys agree:

The smallest gear I think I might like to cut is 10 teeth. So, if I use a gearbox with a 10 to 1 ratio, then the master and slave servos would be turning at the same speed for this 10 tooth gear, and the EG value would be 1.

For a 11 tooth gear, the EG fraction would be 11/10. Thus, it would seem for a gear with n teeth the EG fraction would be n/10.

Now, if an encoder with less than 131072/4 line count was used to drive the slave servo, the numerator of the fraction would need to be multiplied by the ratio of the encoder actual output per rev to 32,768. Thus, for a 1024 ppr encoder, the numerator would need to be multiplied by 32.

Do you guys agree?

sansbury
01-02-2011, 02:50 PM
I am quite prepared to help other who want to help themselves and to have conversations with people who genuinely are interested in gears and gear cutting.

So I have a CNC mini-mill planning to add a 4th axis, plus a CNC mini-lathe. Based on that sort of equipment, what do you consider:

1. the most convenient way to quickly knock out a couple of ordinary spur gears with low to moderate precision; and,

2. The most precise way to make a spur gear without dedicated machinery (dedicated cutters ok) where time is not an issue.

While I cant accommodate any more equipment i wojld not object to spending up to say $200 in order to be able to reliably produce ordinary gears. The programmer in me really likes the gearotic approach but it sounds a bit slow, though that is perhaps made up for by reduced setup time.

macona
01-02-2011, 03:03 PM
I have no problem using a 1024 line encoder (they seem rather common on ebay - and I saw some that were as high as 4096) if it will get the job done accurately and the ratios work out to be within the limits of the EG function. Perhaps this is a good point to discuss costs - if the encoders can be had for $100 (I saw some at that price, but they were 1024 or fewer ppr), how much do servo motors and their controllers go for?

I saw several encoders where the asking price was close to $2000 and it did not appear the line count was anything like 32k - so is it really cheaper to buy a high count encoder vs using another servo?

And, if using a servo motor for its encoder, does one also need a controller package for that servo? I assume the answer is yes, but I didn't think it would hurt to ask....:)


There are various qualities of encoders out there and many, many different styles. Part of the reason in differences in prices. There are highly accurate ones like the Heidenhains and cheep ones that are for panel mount encoder for a shuttle knob. You might look at encoders in the range of 512, 1024, and 2048. These are all evenly dividable to the count of the motors encoder so the fractions will be easier. You dont need a super high count encoder. Thats what the gearing is for.

You will need the motor with the drive to use it as an encoder. Like I mentioned before, they are serially connected to the drive and the drive needs to decode that signal.

The prices of servo motors and drives are all over the place. I have gotten drives for as cheap as $20 and motors for that price as well. But these are outside of the norm. For this application you could probably use a 100w motor since you are going though a gearhead anyway and you really dont need a J2S. The encoder count in the end probably wont matter.

Here is a chart of drives and compatible motors. For J2 series drives just take the S out of the motor and drive models.

http://i65.photobucket.com/albums/h228/macona/mrj2scompatibilitychart.png

macona
01-02-2011, 03:07 PM
So I have a CNC mini-mill planning to add a 4th axis, plus a CNC mini-lathe. Based on that sort of equipment, what do you consider:

1. the most convenient way to quickly knock out a couple of ordinary spur gears with low to moderate precision; and,

2. The most precise way to make a spur gear without dedicated machinery (dedicated cutters ok) where time is not an issue.

While I cant accommodate any more equipment i wojld not object to spending up to say $200 in order to be able to reliably produce ordinary gears. The programmer in me really likes the gearotic approach but it sounds a bit slow, though that is perhaps made up for by reduced setup time.

Quickest way is a rotary table and a hand ground single point cutter. its works fine. Many on this site have done it.

Most precise is probably rigging up something with a servo or stepper.

TexasTurnado
01-02-2011, 04:33 PM
I understand the need to modify the gearbox to reduce backlash by using eccentric bushings - but am curious if there might be more that can be done.

On a former project I was part of, there was a need to wind antenna wire on a drum under quite a bit of tension. The solution was to use two hydraulic motors working against each other under control of a microprocessor to prcisely control the tension. I'm wondering if there might be an electronic analogy to this approach to maintain a load on the gearbox to keep its gears in contact on one side of the teeth to eliminate backlash. Could this be done with a servo motor?

Any comments?

John Stevenson
01-02-2011, 04:43 PM
TT, When I do the 127 tooth wheels on the hobber to reduce any play and prevent chattering I have two pieces of wood with a hole in them on the parting line just under the blank shaft diameter.

These are put either side the shaft and lightly bolted together, one is longer then the other and catches on the bed as the shaft revolves.

This is then adjusted with two high precision hand scraped coach bolts so it acts as a brake.

I pinched the idea off a French piano action machine we had at the last place, don't like cribbing ideas but I'm sure the patent had expired as it was built in 1888.

John Stevenson
01-02-2011, 04:49 PM
So I have a CNC mini-mill planning to add a 4th axis, plus a CNC mini-lathe. Based on that sort of equipment, what do you consider:

1. the most convenient way to quickly knock out a couple of ordinary spur gears with low to moderate precision; and,

2. The most precise way to make a spur gear without dedicated machinery (dedicated cutters ok) where time is not an issue.

While I cant accommodate any more equipment i would not object to spending up to say $200 in order to be able to reliably produce ordinary gears. The programmer in me really likes the gearotic approach but it sounds a bit slow, though that is perhaps made up for by reduced setup time.

With what you have said I'd use the CNC to profile mill a blank in Gauge plate which is then hardened and used as a fly cutter.
Virtually no outlay, $2 for gauge plate

Next step is to use a profile cutter and your 4th axis as a dividing head, again minimum outlay $30 ? for cutter

Next Gearotic and an of the shelf cutter, $75 for gearotic, $5 for a cutter

TexasTurnado
01-02-2011, 05:49 PM
http://i65.photobucket.com/albums/h228/macona/mrj2scompatibilitychart.png

I spent some time rooting around on ebay and see what you mean about prices. :rolleyes: Unfortuneately, most of the listings seem to be from sellers on the other side of the globe....:(

macona
01-02-2011, 05:56 PM
I understand the need to modify the gearbox to reduce backlash by using eccentric bushings - but am curious if there might be more that can be done.

On a former project I was part of, there was a need to wind antenna wire on a drum under quite a bit of tension. The solution was to use two hydraulic motors working against each other under control of a microprocessor to prcisely control the tension. I'm wondering if there might be an electronic analogy to this approach to maintain a load on the gearbox to keep its gears in contact on one side of the teeth to eliminate backlash. Could this be done with a servo motor?

Any comments?


There are gearboxes like those made by bayside that mount right to the end of the motor. The backlash is measured in arc seconds. Now would I mount the gear blank on a arbor hanging off the shaft, I dont know. If I were to do it I would probably make a preloaded spindle that would support the blank just to get some extra stiffness and drive that from the gearbox with a zero backlash coupler like a bellows coupling.

Here is a pic I just took of the 200w motor and drive I have, runs on 120v, good for about 1/4HP continuous and up to three times that for a short interval. It has a Bayside 3:1 gearbox on it rated for 4 arcmin backlash. You want to get these off ebay. These things are stinkin expensive, Bayside sells that model for $785. Paid $16.50 for the drive (Had to repair it) and $36 for the motor/gearbox.

http://farm6.static.flickr.com/5250/5317217895_4679226427_z.jpg (http://www.flickr.com/photos/67292116@N00/5317217895/)
Mitsubishi 200w servo (http://www.flickr.com/photos/67292116@N00/5317217895/) by macona (http://www.flickr.com/people/67292116@N00/), on Flickr

macona
01-02-2011, 05:58 PM
I spent some time rooting around on ebay and see what you mean about prices. :rolleyes: Unfortuneately, most of the listings seem to be from sellers on the other side of the globe....:(

Dont let that stop you. Many of those places offer free shipping. There are a lot of semiconductor fabs over in asia and thats where the used drives and motor come from, decommissioned equipment.

TexasTurnado
01-02-2011, 06:02 PM
TT, When I do the 127 tooth wheels on the hobber to reduce any play and prevent chattering I have two pieces of wood with a hole in them on the parting line just under the blank shaft diameter.

These are put either side the shaft and lightly bolted together, one is longer then the other and catches on the bed as the shaft revolves.

This is then adjusted with two high precision hand scraped coach bolts so it acts as a brake.

I pinched the idea off a French piano action machine we had at the last place, don't like cribbing ideas but I'm sure the patent had expired as it was built in 1888.

Ah, those hand scraped bolts must be your secret edge to success.... :)

I used a similar arangement to provide tension when coiling wire for kiln elements, but the blocks wore so rapidly I had to use the other hand to adjust the tension....:D

Patented in 1888?? That must be when they opened the patent office - I'm sure someone thought of it before then. :rolleyes:

TexasTurnado
01-02-2011, 06:37 PM
There are gearboxes like those made by bayside that mount right to the end of the motor. The backlash is measured in arc seconds. Now would I mount the gear blank on a arbor hanging off the shaft, I dont know. If I were to do it I would probably make a preloaded spindle that would support the blank just to get some extra stiffness and drive that from the gearbox with a zero backlash coupler like a bellows coupling.

Here is a pic I just took of the 200w motor and drive I have, runs on 120v, good for about 1/4HP continuous and up to three times that for a short interval. It has a Bayside 3:1 gearbox on it rated for 4 arcmin backlash. You want to get these off ebay. These things are stinkin expensive, Bayside sells that model for $785. Paid $16.50 for the drive (Had to repair it) and $36 for the motor/gearbox.

http://farm6.static.flickr.com/5250/5317217895_4679226427_z.jpg (http://www.flickr.com/photos/67292116@N00/5317217895/)
Mitsubishi 200w servo (http://www.flickr.com/photos/67292116@N00/5317217895/) by macona (http://www.flickr.com/people/67292116@N00/), on Flickr

Don't happen to know where I can find one of those in 10 to 1 for about the same price (or even 5x that price), do you? :)

I haven't given the blank mounting much thought yet, but for the hob I am thinking of making a large cast iron block to clamp to the exposed dovetail above the knee on my Lagun FTV-1S. The front surface of this block would have T-slots (or similar) to rigidly clamp a yoke with the hob between the ends of the yoke, mounted in precision bearings. By tilting the head of the mill to the helix angle of the hob, and sliding the ram in or out to align the yoke to the block on the dovetail, I would have a very rigid hob mounting.

It appears I can extend the quill of the machine about 3/4 inch to expose a surface to mount the servo/encoder and couple it to the spindle with a notched belt. I haven't decided yet whether the spindle can be direct coupled to the shaft through the yoke, or if I should use something like large bellows for a backlash free coupling that puts no extra load on the hob bearings.

At this point, I only know the blank will be mounted on the table....:D

dm1try
01-02-2011, 08:38 PM
i've been playing with mitsubishi J2S servos lately. if someone's wondering what do they look like inside - here's a couple of pictures. encoders are absolute, they need a battery in the amplifier and a supercapacitor in the motor to keep track of full turns counts, imho
http://i1143.photobucket.com/albums/n629/Dmitry_Yurtaev/hc-kfs73-xplode.jpg
http://i1143.photobucket.com/albums/n629/Dmitry_Yurtaev/kfs-enc.jpg

TexasTurnado
01-02-2011, 09:13 PM
You want to get these off ebay. These things are stinkin expensive, Bayside sells that model for $785. Paid $16.50 for the drive (Had to repair it) and $36 for the motor/gearbox.

[

I looked there and at the mfgs website and those gearboxes look very well made and robust. The ratios go as high as 100 to one, and 10 to 1 seems to be readily available for around $200. Of course, there is no way to know how worn they are and thus what the real backlash is unless you get a new one and then the price goes up in the $500 range.... :(

Peter.
01-02-2011, 09:24 PM
i've been playing with mitsubishi J2S servos lately. if someone's wondering what do they look like inside - here's a couple of pictures. encoders are absolute, they need a battery in the amplifier and a supercapacitor in the motor to keep track of full turns counts, imho


Your image links are bringing up a certificate error on i.e.