PDA

View Full Version : New Electronic Leadscrew Project



Paul Alciatore
04-21-2019, 03:28 AM
This looks promising. I like his thinking.

https://www.youtube.com/watch?v=FTs9GygRQ-U

Sparky_NY
04-21-2019, 06:42 AM
This looks promising. I like his thinking.

https://www.youtube.com/watch?v=FTs9GygRQ-U

Electronic leadscrews are very old news for lathes. There have been countless build projects and forum threads on the subject over the years. The concept is sound. This guy is not developing any thing new, far from it.

This guy sounds like he has plans for a commercial product in the future, not just a one-off for his lathe. A commercial product is a different matter, there are so many hardware configurations for the different lathes out there that every install will be unique (hardware wise) When developed, the cost of a commercial product will likely be beyond what most will be wiling to pay.

As I said, electronic leadscrews are old news YET no commercial products have seem to appeared over the years. There is probably good reasoning for that.

Doozer
04-21-2019, 08:18 AM
I am also interested in buying (or making) an ELS for my lathe.
For me, I am pretty good with electrical and electronics, but I
just don't have the "hobby" time to make my own boards and
solder in 100+ components. So I would like to buy one.
One product that I have bought for a CNC lathe that I made
was the servo motor and drive package from AutomationDirect.
If you look on their site, you can get them in different Watt ratings,
which is horsepower. You get to buy the motor and matching cable
that plugs into their drive. You also buy an encoder (for attaching
to the lathe spindle) and it also had a cable that plugs into their drive.
The drive connects to a laptop and they give you software to setup
the drive parameters. There are 2 parameters, which are a fraction
ratio that you can input. This ratio is the setup between the input from
the encoder, and the output to the motor. (The motor you would connect
to the leadscrew). So 1 to 1, the leadscrew will spin the same speed
as the spindle. You need to figure your math to cut whatever threads per
inch or millimeter pitch that you want, so there is some manual math
needed to find what ratio you need to input into the drive. (Example,
1:1 ratio, the lathe will cut threads that match the pitch of the leadscrew,
and from there adjust your math accordingly.) Pretty easy to calculate,
no sexy front end screen with this drive, just manually input 2 numbers
for the lead-follower fraction. But I thought it was a highly usable option
for a purchased ELS system for the home shop. I do not know much
about the Raspberry Pie, but I think it is a small computer. Thinking it
might be able to run the servo drive software in a neat little package,
without the need to have a laptop always sitting near your lathe.
What I do know, is that AutomationDirect does sell these servo drive
and motor packages for around $600 to $800 and they work very well.
At least the lathe I built with them, it was a good fit for the application.
Maybe for you cheap home shop bastards, the price might seem a
little bit high, but having used one, I can assure you that the Automation
Direct motors and drives work well and are not difficult to program.
I know it seems like a lot of money to outlay on a possible unknown
where you have to rely on yourself to sort out the details, but the
CNC conversion guys do this every day. I am just here to testify that
their servo drive works and it is nice to have a motor and an encoder
that will plug directly into their drive and it works. No figuring out pins
and wire connections. That is worth something to me. Actually I called
AutomationDirect once, an asked if I could program the ratio in their
servo drive software using one of their touch screens, without connecting
the touch screen to a PLC first. They told me yes, that they offer a touch
screen that would plug the numeric values into the drive software, and no
need to go through a PLC, but you could not do much math with the touch
screen alone, just enter the variables. With a PLC you could do the math
like enter the TPI and it puts in the ratio for you, but it would be another
piece of hardware to buy. The Babin ELS on the Hardinge lathes uses
a touch screen with a PLC built in to do what they do. The Babin touch
screen is not from AutomationDirect. It is another brand. I looked into
it at one time, and I figured out what Babin uses for a screen, but I forgot
what exactly which one it was.
When I get some time, I am going to buy another AutomationDirect servo
drive and motor and encoder, and convert one of my toolroom lathes
for ELS. If any of you guys have some ideas, I will be watching this thread.

--Doozer

CCWKen
04-21-2019, 09:27 AM
I'd buy one although I don't think I'd pay $600 for it. That would buy a lot of the taps and dies I use now. And keeping stock around for off-the-shelf parts like special boring head mounts seeps wasteful. My change gears stay on my favorite feed. I would like to change that on the fly. :cool:

BCRider
04-21-2019, 10:49 AM
After watching about half that video I feel a serious urge to go out to the shop and give the full featured thread or feed QCGB on my lathe a hug....

lakeside53
04-21-2019, 10:58 AM
I followed the original ELS project (was a yahoo group) for years, and came to the conclusion that although was interesting at the time, today you may as well just CNC the entire lathe by adding in the second axis. Use a simple conversational interface if the "programming" seems "daunting".

nickel-city-fab
04-21-2019, 01:26 PM
Was gonna say, isn't this basically what the Hardinge HLVH does with analog electrics? I imagine designing the analog way would be much easier/fewer components, for me anyway. At the end of the day, its all math.

skunkworks
04-21-2019, 02:25 PM
I agree with lakeside... Linuxcnc has had encoder threading for 15 years or more. I just works. Lead-in and lead out plus tons of other functionality...

Reminds me of this
https://www.youtube.com/watch?v=D_aGsBCorww

'I picked a cnc controller that doesn't thread - so I added threading with an arduino...'

sam

elf
04-21-2019, 04:11 PM
I don't know what the big deal about an ELS is. All you need is synchronized steppers (or servos). My rose engine can do threads with any pitch, multi-start threads, left hand, right hand, or just plain helixes. It's driven with a Teensy microprocessor (https://www.pjrc.com/store/teensy35.html) (Arduino compatible). A library, TeensyStep (https://github.com/luni64/TeensyStep), does the synchronization. My code just needs user input for number of revolutions and length to thread. I also designed the driver board (https://github.com/elfren/RoseEngine_SpindleAndAxis/tree/master/PCB/Four_Axes) that holds the Teensy and four stepper drivers.

RB211
04-21-2019, 04:25 PM
Yep, might as well go full CNC.

Baz
04-21-2019, 05:45 PM
Yep, might as well go full CNC.
Anyone here own a bicycle? Why? You might as well get a 6x6 100ton dump truck for moving the groceries or an F1 racer if in a hurry.
An ELS can potentially be a substitute for a pile of gears (cost and hassle) or a complex QCGB (eg Harrison M300 one seen on ebay for £750) but stepper and psu and processor costs do mount up so full CNC does get expensive.
I got a little Arduino last year with the intention of doing an ELS and schemed the software but then got waylaid. Two aspects that I planned were
a) no screen and fiddly menus - use two rotary 16way BCD switches so I can go from thread to feed in one second flat. (trouble is they are expensive)
b) don't try to do any maths in the mini processor. Work it out beforehand and just have lookup tables. No danger of lost cycles.

RB211
04-21-2019, 09:20 PM
Anyone here own a bicycle? Why? You might as well get a 6x6 100ton dump truck for moving the groceries or an F1 racer if in a hurry.
An ELS can potentially be a substitute for a pile of gears (cost and hassle) or a complex QCGB (eg Harrison M300 one seen on ebay for £750) but stepper and psu and processor costs do mount up so full CNC does get expensive.
I got a little Arduino last year with the intention of doing an ELS and schemed the software but then got waylaid. Two aspects that I planned were
a) no screen and fiddly menus - use two rotary 16way BCD switches so I can go from thread to feed in one second flat. (trouble is they are expensive)
b) don't try to do any maths in the mini processor. Work it out beforehand and just have lookup tables. No danger of lost cycles.

You are only 1 ballscrew and stepper away from full cnc. Might as well go all the way.

DFMiller
04-22-2019, 03:17 AM
I can see this is a useful project.
For instance if your trusty lathe develops a problem like a busted fibre gear all of a sudden and you canít just run out and get a new gear.
In fact my nice older Taiwanese Myford clone did exactly that yesterday. If the gear turns out to be unobtainium I will be looking at this or something similar.

Dave

RB211
04-22-2019, 06:54 AM
I can see this is a useful project.
For instance if your trusty lathe develops a problem like a busted fibre gear all of a sudden and you can’t just run out and get a new gear.
In fact my nice older Taiwanese Myford clone did exactly that yesterday. If the gear turns out to be unobtainium I will be looking at this or something similar.

Dave

3D print a gear or make one, this is a home shop machinist forum, right?

DFMiller
04-22-2019, 09:33 AM
Or make and ELS.
All are options. I suspect ELS is more versatile and much simpler and reliable.

J Tiers
04-22-2019, 09:33 AM
You are only 1 ballscrew and stepper away from full cnc. Might as well go all the way.

Agree.

The costs seem to be close enough that it makes more sense to do CNC as opposed to spending a considerable amount on a more seldom used function in order to obtain the capability to obtain an even less often used functionality.

At least CNC would be usable for nearly every task you would use the lathe to do, instead of for just one rarely used function.

Dan Dubeau
04-22-2019, 11:06 AM
Most lathe stuff (that I do anyway) is quicker to just crank handles, than to write a program and run it. An ELS would be a very nice compliment to my lathe, and I honestly wouldn't want a full blown CNC conversion. Different people, different needs. The guy seems very knowledgeable in ways that I'm not. I've enjoy following this video series.

jackary
04-22-2019, 12:46 PM
I did this a while ago and it still works ok
https://www.youtube.com/watch?v=ZH0ZjWauMZ0&t=3s

Alan

skunkworks
04-22-2019, 01:06 PM
Wow - tubocnc - I got my feet wet with that software!

I moved on to linuxcnc because turbocnc doesn't have any look ahead - constant velocity tool paths.


I did this a while ago and it still works ok
https://www.youtube.com/watch?v=ZH0ZjWauMZ0&t=3s

Alan

Baz
04-22-2019, 04:13 PM
Agree.
a more seldom used function in order to obtain the capability to obtain an even less often used functionality.
.
A big attraction for the UK user is that we are caught in the middle of metric and imperial threads. Nothing smaller than an M300 provides both without fiddling with changewheels, even if you have a full Norton box to provide all the imperials. The ELS offers both with feeds thrown in. Even if you only do one thread per day the chances are it will be the opposite (imp or metric) to what you last did.

Lew Hartswick
04-22-2019, 11:34 PM
Most lathe stuff (that I do anyway) is quicker to just crank handles, than to write a program and run it. An ELS would be a very nice compliment to my lathe, and I honestly wouldn't want a full blown CNC conversion. Different people, different needs. The guy seems very knowledgeable in ways that I'm not. I've enjoy following this video series.
I agree 100 pct. It would take ten times the time to write the silly CNC instructions than to just turn the cranks.knobs for nearly all one-off things I do on a lathe.
...lew...

J Tiers
04-23-2019, 12:01 AM
A big attraction for the UK user is that we are caught in the middle of metric and imperial threads. Nothing smaller than an M300 provides both without fiddling with changewheels, even if you have a full Norton box to provide all the imperials. The ELS offers both with feeds thrown in. Even if you only do one thread per day the chances are it will be the opposite (imp or metric) to what you last did.

I do about one a week, and even at one a day, I can fiddle with the change gears while spending my time and money on something that pays back faster, like a CNC mill.

For most regular threads, it is easier to use a die or tap.

3 Phase Lightbulb
04-23-2019, 09:42 AM
I really like the idea of having a simple ELS system for feeding and threading, and it's a great captivating project to work on.

JCByrd24
04-23-2019, 10:21 AM
I really like the idea of having a simple ELS system for feeding and threading, and it's a great captivating project to work on.

I'd agree. My lathe like many others of our class has a somewhat non-lubricated, open drivetrain for the quick change gearbox. This thing makes a racket at anything over a couple hundred RPM and I truly think was not designed for much use above 1000 rpm. I added a simple variable speed motor drive to my power feed rod and now only use the lathes feed for threading via the lead screw. In this process I lost the fixed ratio between the spindle and feed rod, and thus knowing what my feed rate is. A system such as this could add it back without the racket and power loss of the open gear train.

Lew Hartswick
04-23-2019, 10:52 AM
I really like the idea of having a simple ELS system for feeding and threading, and it's a great captivating project to work on.

:-) From another perspective; I was in the "electronics" field my whole working life (1950 to 1997) and the failure rate for such devices has always been such that I'll take the "gear driven" approach every time. :-)
...lew...

mattthemuppet
04-23-2019, 11:08 AM
even a simple DC motor driven feedscrew would be nice, just to have variable feeds. I've seen a bunch of those.

3 Phase Lightbulb
04-23-2019, 11:25 AM
:-) From another perspective; I was in the "electronics" field my whole working life (1950 to 1997) and the failure rate for such devices has always been such that I'll take the "gear driven" approach every time. :-)
...lew...

Is that what you really wanted to say? :)

J Tiers
04-23-2019, 12:28 PM
There are indeed many failure mechanisms for electromechanical devices. And in general, most folks cannot repair them, and are therefore dead in the water if they fail. They tend to be costly.

Reliability is generally good, however, at least for the first 15 years or so. After that it becomes more and more "iffy", and parts become unavailable, as well as "un-makeable".

Gears do have the advantage that they work reliably, have known failure mechanisms, and are not hard to repair for anyone who has a shop and some know-how. Motors are less makeable, but are plentiful and not as likely to be obsoleted.

It is a trade-off. Capability vs the problems of repair and potential failures, and of cost. It would appear that the cost is close enough to CNC to make the move to a true CNC reasonable, given the step-up in capability.

Baz
04-23-2019, 01:04 PM
Post apocalypse the geared drive on your treadle lathe enables you to make the vital bolt to repair the wind generator.....,.

3 Phase Lightbulb
04-23-2019, 01:39 PM
Post apocalypse the geared drive on your treadle lathe enables you to make the vital bolt to repair the wind generator.....,.

Or maybe a more positive outlook? Maybe the ELS conversion that you successfully built yourself pays you back handsomely with new lucrative abilities and you easily amass a fortune and get to focus on what you like to do, vs what others want you to do. :)

Dan Dubeau
04-23-2019, 03:10 PM
Add a micrometer adjustable depth stop rod to trigger a limit switch on the carriage, coupled with the knob adjustable feedrates and an ELS can turn any lathe into the holy grail of single point threading lathes....the HLVH. Ya, the iron and spindle wont be as good, but user friendliness is right there with the Hardinge. Add a 3phase motor with vfd and youd be even closer.

PStechPaul
04-23-2019, 03:27 PM
I made the following comment on the video:

I like this idea. I am comfortable with SMT devices, at least 0805 and larger, and 1.27 mm SOIC, and such components are often cheaper and more available than thru-hole. It would also be nice to be able to drive the cross-slide as well, and add some CNC capability. But then it might be necessary to use ball-screws to eliminate backlash. Perhaps it could use DROs to obtain absolute position data for the X and Z axes of the lathe. I might consider making a much simpler version only for power feed (which does not require synchronization), and use the existing gears for threading. Thanks.

I might pursue the idea of a simple electronic motorized leadscrew. The biggest hassle is changing gears between power feed and threading, but it should be possible to just disconnect the gears from the spindle and drive the leadscrew with a stepper motor at various RPM values. And it would be easy to make an electronic limit switch to stop the power feed instantly.

jhovel
04-24-2019, 12:19 PM
Why are people reinventing wheels again and again?
13 years ago: https://www.youtube.com/watch?v=RRaVIBhLgF0 (ELS group development project, lead by John Dammeyer)
9 years ago: https://www.youtube.com/watch?v=-QyjAuhhGL8 (Warco Lathe ELS)
4 years ago: https://www.youtube.com/watch?v=RTPjz6j-QsE (ELS II Rocketronics)

All do every know standard thread by selecting on the disply, all known standard tapers by selecting on the display, power facing, power chamfering, power turning - all by driving the tool to the start point, pressing a button to identify it, diving the tool to the end point, pressing another button. Then select how deep each cut should be, what feed rate, how many spring passes and then drive the tool clear of the work. You then press start and off she goes - look mum, no hands. You can select the best speed for your tool, taking your concentration and reflexes out of the equation. Want a non standard thread? No problem type in the pitch in TPI or mm, fit the tool with the right geometry. Go. Non-standard taper? Type in the angle, diametrical difference or cosine. Go.

CNC programming? Conversational programming? Arduino? Huh? Where?

I built mine about 10 years ago from the kit that John Dammeyer still supplies. I've worn out and replaced a couple of touch pad switches, John just sent me new ones. No bill....

Oh, the ELSII can do radius and ball turning. The processor in the ELS does not have the capacity to add that. But John is working on a piggy-back board to fit a newer more capable processor. The ELS community is still alive and well on Yahoo and several people have developed new applications for the ELS, e.g. driving rotary tables, partial milling machine control etc.

3 Phase Lightbulb
04-24-2019, 12:39 PM
Why are people reinventing wheels again and again?

Might be that some folks really don't understand the benefits of stone wheels....

http://www.bbssystem.com/pictures/2019_zr1.jpg

skunkworks
04-24-2019, 12:45 PM
Still seems limited... Why not go with a cheap computer and linuxcnc? The examples above don't seem to allow for synced lead-in/lead-out for the threading operations?


https://www.youtube.com/watch?v=_7skJhKzU7Y


Why are people reinventing wheels again and again?
13 years ago: https://www.youtube.com/watch?v=RRaVIBhLgF0 (ELS group development project, lead by John Dammeyer)
9 years ago: https://www.youtube.com/watch?v=-QyjAuhhGL8 (Warco Lathe ELS)
4 years ago: https://www.youtube.com/watch?v=RTPjz6j-QsE (ELS II Rocketronics)

All do every know standard thread by selecting on the disply, all known standard tapers by selecting on the display, power facing, power chamfering, power turning - all by driving the tool to the start point, pressing a button to identify it, diving the tool to the end point, pressing another button. Then select how deep each cut should be, what feed rate, how many spring passes and then drive the tool clear of the work. You then press start and off she goes - look mum, no hands. You can select the best speed for your tool, taking your concentration and reflexes out of the equation. Want a non standard thread? No problem type in the pitch in TPI or mm, fit the tool with the right geometry. Go. Non-standard taper? Type in the angle, diametrical difference or cosine. Go.

CNC programming? Conversational programming? Arduino? Huh? Where?

I built mine about 10 years ago from the kit that John Dammeyer still supplies. I've worn out and replaced a couple of touch pad switches, John just sent me new ones. No bill....

Oh, the ELSII can do radius and ball turning. The processor in the ELS does not have the capacity to add that. But John is working on a piggy-back board to fit a newer more capable processor. The ELS community is still alive and well on Yahoo and several people have developed new applications for the ELS, e.g. driving rotary tables, partial milling machine control etc.

Noitoen
04-24-2019, 04:08 PM
Some time ago I saw a guy online with a very simple pipe threading contraption. It used an x/y system driven by 2 steppers and a rotary cutting tool mounted on a small angle grinder. The pipe held on an hand cranked chuck with an encoder and as the pipe slowly rotated the high speed cutter would approach the pipe in sync, cutting a neat thread on different size of pipes or bolts.
I tried to find this video again but since I didn't bookmark it, I couldn't find it.

jhovel
04-25-2019, 06:58 AM
All three examples shown above can move both x and z axes and do synced movements for threading Lead ins, cutter retractions, automatic depth of each cut and spring passes.... as I described.
But if you really want to do CNC programming to cut a few threads, these are not for you. If you want to do small or large production runs, then this whole thread is definitely not for you.

Mcgyver
04-25-2019, 08:38 AM
A lot of the responses here seem along the lines of "ELS has been done before". So? I haven't studied this subject, but lets say the above link is a really slick solution, will cut every thread perfectly at high spindle speeds, needs no computer and is cheap and robust. Why does it matter that there other ELS out there? Are you saying this one is inferior to these other ELS's? If so, please give details, why, etc.

skunkworks
04-25-2019, 11:15 AM
Can you give some examples? I cannot find any thing that show a synced lead-in or out.. The manual I found doesn't show it. (unless I am missing something..)

https://www.rocketronics.de/download/zyklo/Manual-ELS-II-Rev-5-English.pdf


All three examples shown above can move both x and z axes and do synced movements for threading Lead ins, cutter retractions, automatic depth of each cut and spring passes.... as I described.
But if you really want to do CNC programming to cut a few threads, these are not for you. If you want to do small or large production runs, then this whole thread is definitely not for you.

lakeside53
04-25-2019, 11:25 AM
I made the following comment on the video:


I might pursue the idea of a simple electronic motorized leadscrew. The biggest hassle is changing gears between power feed and threading, but it should be possible to just disconnect the gears from the spindle and drive the leadscrew with a stepper motor at various RPM values. And it would be easy to make an electronic limit switch to stop the power feed instantly.

That's the original ELS from a decade back.. don't forget you need to read spindle speed and adjust feed to match, or vice versa.

For all that are reinventing the wheel, go back and read the original ELS project forum design notes and interactions. There is s lot that can be learned.

jhovel
04-27-2019, 10:16 AM
In that case I don't know what you mean by "lead in and lead out".... I understand the terms to mean set the tool to a cutting position in X axis to start cutting on the fly and remove it at the end of the cut in exactly the same position for each cut - synchronised with the spindle. Thats what the ELSs do, as described and shown.
You obviously mean something different.

@Mcgyver: no reason to invent another one at all! You might even hit on a solution that nobody else has thought of.
It just seems a lot of energy to spend when you can just go ahead a buy the (cheap) kit that allows you do make swarf straight away, instead of designing the gear to make swarf one day in the future....
Now if you got a kit and came up with a simple solution to upload more complex profiles straight from a CAD package as a "user-defined profile" and just run it to - say - turn ball handles, balanced machine handles, shoulder bolts or other multi-step shapes, then you would have my full attention and support!

Doozer
04-27-2019, 10:07 PM
Stop stop stop with the suggestion to just CNC it.
Not everyone wants CNC. Me included.
A good ELS system would complement a good manual lathe.
To convert the lathe to CNC is a lot more complexity and a
larger learning curve than many people care to take on.
Again, me included. With CNC, if you have anything, a decimal
point or whatever, out of place, the whole thing can come
crashing down. That is a reason why CNC is BS in a home shop
for many of us. Just don't need the extra headache and stress
of CNC. Yes stress of crashing the damn thing all the time with
the one off things HSMers do. So ELS would be great. Whole
lathe CNC conversion, no no no. Just stop it. There is another
whole forum on this board for CNC. There is a reason they are
not all lumped together like marshmallows in a bag. Keep it.

-Doozer

RB211
04-27-2019, 10:25 PM
Stop stop stop with the suggestion to just CNC it.
Not everyone wants CNC. Me included.
A good ELS system would complement a good manual lathe.
To convert the lathe to CNC is a lot more complexity and a
larger learning curve than many people care to take on.
Again, me included. With CNC, if you have anything, a decimal
point or whatever, out of place, the whole thing can come
crashing down. That is a reason why CNC is BS in a home shop
for many of us. Just don't need the extra headache and stress
of CNC. Yes stress of crashing the damn thing all the time with
the one off things HSMers do. So ELS would be great. Whole
lathe CNC conversion, no no no. Just stop it. There is another
whole forum on this board for CNC. There is a reason they are
not all lumped together like marshmallows in a bag. Keep it.

-Doozer

A decimal place out and it comes crashing down? Oh I see, you want nothing to do with CAD/CAM, just a little magic box that allows you to select a feed rate or TPI. Hardware wise, you are still just one axis short of full CNC capability, and why you don't see many ELS threads. Majority of people just do CNC.
And your ELS will still crash without added safeguards like limit switches. It's like buying a Tesla without the computer and only half the motors

Paul Alciatore
04-28-2019, 12:37 AM
I started this thread because I have thought about the math needed to do this right. By that I mean without any slippage due to approximations. For instance, division by three in a binary system is not a trivial matter. 1/3 = 0.333, doesn't it? Well, no. 0.3333 is closer but still not the actual value. In fact, using base ten numbers you can never actually state a single number that is equal to 1/3. You must use the fraction (two numbers) if you want it to be exact. The same thing happens with binary numbers that all computing devices use. In binary 0.1 is one half and that is too big. So you start with 0.01 which is 1/4. That leaves 1/12 remaining. So you have to add 1/16 or 0.0001 and you have 0.0101. But you are still short and additional binary places will have to be added: 0.010101, etc. But, as in base 10 numbers, you never get to the actual 1/3 value. No matter how many places you carry it to, it will always be a bit short. And this leads to cumulative errors in the binary math that computing devices use.

The thinking that I saw in this YouTube video seems to show that he is going to tackle this problem. There are ways to do it, but you have to go beyond just using floating point math.

Another thing that I liked in his approach is the fact that he is using a 1024 division encoder. I have seen ELS systems based on a ONE division encoder. They had to rely on the spindle speed being constant to get a decent thread. This can be done, but it is very prone to errors if the metal being cut has areas that are harder and softer than the average speed. I don't like that.

He also talks about a TI processor that is a lot more powerful than the average PIC style device.

As I said, I like the way he is thinking. I intend to follow his progress just to see where he goes with it.

RB211
04-28-2019, 01:16 AM
I started this thread because I have thought about the math needed to do this right. By that I mean without any slippage due to approximations. For instance, division by three in a binary system is not a trivial matter. 1/3 = 0.333, doesn't it? Well, no. 0.3333 is closer but still not the actual value. In fact, using base ten numbers you can never actually state a single number that is equal to 1/3. You must use the fraction (two numbers) if you want it to be exact. The same thing happens with binary numbers that all computing devices use. In binary 0.1 is one half and that is too big. So you start with 0.01 which is 1/4. That leaves 1/12 remaining. So you have to add 1/16 or 0.0001 and you have 0.0101. But you are still short and additional binary places will have to be added: 0.010101, etc. But, as in base 10 numbers, you never get to the actual 1/3 value. No matter how many places you carry it to, it will always be a bit short. And this leads to cumulative errors in the binary math that computing devices use.

The thinking that I saw in this YouTube video seems to show that he is going to tackle this problem. There are ways to do it, but you have to go beyond just using floating point math.

Another thing that I liked in his approach is the fact that he is using a 1024 division encoder. I have seen ELS systems based on a ONE division encoder. They had to rely on the spindle speed being constant to get a decent thread. This can be done, but it is very prone to errors if the metal being cut has areas that are harder and softer than the average speed. I don't like that.

He also talks about a TI processor that is a lot more powerful than the average PIC style device.

As I said, I like the way he is thinking. I intend to follow his progress just to see where he goes with it.

Most of these 8 bit risc processors do not like floating point math, better at integer math, but who really cares? Any rounding errors with an integer sized variable is still going to be insignificant compared to any machine error.
As for the processor you use, for me at least, it is based on what IDE I like better. I am partial to Atmel. Even the dumbed down Arduino ecosphere is plenty good for this application. Perhaps the TI launchpad has floating point co-processor, but would be overkill still.
For reading the quadrature output of a spindle encoder, you are most definetley going to want to use at least two hardware interrupts to pick up the signal accurately. So go with a mcu that has more than a couple. You'll want another or two for reading a user interface rotary encoder, and the stepper driver might want one as well.

PStechPaul
04-28-2019, 02:23 AM
Unless you have accurate DROs on the carriage and the cross-slide, OR zero backlash ball screws, all the computing power of a Cray will not provide any better results than a properly chosen and programmed PIC or Atmel or other similar microcontroller. I think it would be valuable to add power feed for turning, facing, and parting, none of which require precise synchronization to spindle speed. The existing change gears and QCGB are adequate for occasional threading on a hobby grade lathe.

nickel-city-fab
04-28-2019, 08:02 AM
You know, there are analog computers that can divide to any arbitrary number you want. You get a voltage out. There is no LSB error because there are no bits to drop. Take that for your input to encode it for your steppers, for example with a voltage-to-frequency converter and some wave shaping to get your square wave. Even if the V-F conversion drops some, the error will disappear inside the overall tolerances of the entire machine. FWIW, modern DVD players and sound cards are all using 24-bit resolution and you can get the chips for pennies.

vpt
04-28-2019, 09:18 AM
While none of this is for me, I like my mechanical lathe, I still enjoy seeing any sort of projects people are working on.

I agree with the home shop one offs being much more work and possible problems with cnc on the lathe. Even threading relying on ELS to be right all the time is to risky imo.

CNC on a mill is a bit more feasible but still lots of piddling around with programs and still a decent risk of crashing the tool and part. Again one off home shop stuff I don't see any benefit to a cnc mill.

Now a cnc cut table (plasma in my case) I feel is great! 2D parts so almost never a crash, 2D takes very little time to draw up what part a guy needs. One offs are done in the same amount if not less time than manually cutting a part and the part comes out great with rounded corners and whatnot if you choose, these features take quite a while to get manually cutting and grinding. A great asset to a home shop if within the budget.

Sorry for going off on cnc stuff. I have adhd and stuff.

Anyway, looking forward to some build pics and videos!

Glug
04-28-2019, 09:43 AM
Most of these 8 bit risc processors do not like floating point math, better at integer math, but who really cares?

The latest $35 Raspberry Pi is 64 bit, with 4 cores and of course runs Linux. Hugely powerful and versatile.

I want to do an ELS on my lathes at some point. For metric threading, faster apron return speed, on the fly feed adjust, etc. A challenge is making it not look like a goofy grafted on hack. In the case of the 10ee, preserving the integrity of the original machine is critical.

wierdscience
04-28-2019, 09:51 AM
A decimal place out and it comes crashing down? Oh I see, you want nothing to do with CAD/CAM, just a little magic box that allows you to select a feed rate or TPI. Hardware wise, you are still just one axis short of full CNC capability, and why you don't see many ELS threads. Majority of people just do CNC.
And your ELS will still crash without added safeguards like limit switches. It's like buying a Tesla without the computer and only half the motors

Problem is the majority that do "just CNC it" never get to the point of threading,or if they do,they find out their hardware is adequate for everything,except CNC threading.
The latter happened to a friend of mine about three years ago.He converted a Grizzly 13x40 to CNC,then found out the 3hp motor and VFD he bought were not up to the task,so upped the anty to 5hp and a vector drive($850)then found out that closed loop stepper or full on servo motors it must be(+$1800) That and a full year of his spare time and about a month of my spare time to get all the hardware sorted out.

It works fine now,but only after a lot of childbirth.He was partially fooled into thinking it would be easy,by having converted a mill previously(cake walk by comparison) and by all the youtube videos of "CNC threading linux..blah,blah,blah" where the video shows a piece of Aluminum,plastic,or shudder even wax being threaded,just long enough for a nut to thread on.

wierdscience
04-28-2019, 10:06 AM
even a simple DC motor driven feedscrew would be nice, just to have variable feeds. I've seen a bunch of those.

That is exactly what I am going to do on my 9x20.I recently bought one of the cheap ($120) blidgeport table feeds off fleabay.I'm going to extend the rigthand end of the lead screw and mount the table feed there.So then I will have fwd/rev and VS feeds and be able to keep the thread change gears in the most used range.

How I wish the Chinese had copied some lathe other than an Emco.

Mcgyver
04-28-2019, 10:08 AM
In the case of the 10ee, preserving the integrity of the original machine is critical.

Slick would be having it buried in the gearbox, using all the original controls. Of course there'd be a pendant hidden close by to over ride the front controls and dial metric or non standard threads.

The guys who'd get the greatest benefits are ones without a quick change box. I think a powerful low cost solution not requiring a computer would be a welcome addition

RB211
04-28-2019, 10:58 AM
I don't understand the fear of CNC. If computers intimidated me, I would be motivated to overcome that to enjoy CNC.

wierdscience
04-28-2019, 11:08 AM
It's not the fear of CNC,it's the pain of interfacing it and making it work right that's the issue.In the case of threading it's not exactly cut and dried.

Glug
04-28-2019, 11:18 AM
Slick would be having it buried in the gearbox, using all the original controls. Of course there'd be a pendant hidden close by to over ride the front controls and dial metric or non standard threads.

Yes. The Navy had my 10ee rebuilt in 1993. The M-G was removed and replaced with a DC drive, in an oversize enclosure bolted to the back. There is a control box on a post above the headstock. I'm not sure how you can make that look better. Most aspects could be integrated with traditional controls.

At least it was very professionally and neatly done. All the wiring is numbered, etc. I recently had access to schematics and was able to take photos. It'd be nice to better integrate it, without losing to good aspects. The Fincor DC drive needs to be updated to something better, I'm thinking a Parker Eurodrive.


The guys who'd get the greatest benefits are ones without a quick change box. I think a powerful low cost solution not requiring a computer would be a welcome addition

The other huge benefit is silent feeds. That transforms a lathe.

Removing the terrible clatter and noise from end gears, especially on machines where the end gears are only partially covered. Many of us use feeds nearly constantly. And let's face it, the typical "quick" change gear selection is beyond cumbersome, especially when you want to make quick changes to feeds.

Ignoring threading for a moment, how awesome would it be to have a 20 position detent pot with different feed/rev settings? Plus a second pot that is a multiplier to increase the feed rate when returning the carriage to the right.

nickel-city-fab
04-28-2019, 11:36 AM
If I wanted a CNC lathe I would just buy a Haas TL-1 and be done with it. Base model costs about the same as a new pickup with no tooling. Problem is, I don't have the space, the electricity, or a big enough compressor to feed it.

lakeside53
04-28-2019, 11:53 AM
How I wish the Chinese had copied some lathe other than an Emco.

It was "which" Emco they copied. It would have been a different world with they had copied the V10.


For those that don't want to deal with the "cnc" aspect, what we need a small continuously variable transmission (couple of of cones?) instead of a quick change gearbox driving the lead screw. They you could cut anything within a range at the turn of a knob

Mcgyver
04-28-2019, 01:27 PM
I don't understand the fear of CNC. If computers intimidated me, I would be motivated to overcome that to enjoy CNC.



sounds more like "I don't understand the other guys position or its different than my own so he mustn't have a good position, its some fault of his"

its zero to do with fear, its that one size doesn't fit all situations.



For those that don't want to deal with the "cnc" aspect, what we need a small continuously variable transmission (couple of of cones?) instead of a quick change gearbox driving the lead screw. They you could cut anything within a range at the turn of a knob

No exact ratio so not good for threading and more mechanical complexity.

J Tiers
04-28-2019, 04:33 PM
......

No exact ratio so not good for threading and more mechanical complexity.

Might be fine for threading. Short threads tolerate small mismatches, and the metric thread approximations are not exact, but work rather well. Analog fire control computers worked well despite using mechanical computers in many cases, as with the "TDC" in submarines.

The problem would be wear and calibration, as well as the fact that there would still probably be ranges, one set of cones would not cover the full gamut of threads in general use. So there would be enough added external to the cone setup that you might as well use the Norton box.

I agree it might not be the best for general use, but it has possibilities. Mechanical complexity is likely a wash as compared to other sorts of complexity with which folks may not be experienced or confident in dealing with.

Baz
04-28-2019, 06:07 PM
The Harrison M300 QCGB uses 4 knobs to get most imperial and close approximations to metric using just 30 gears which are relatively low count, 20 to 50 teeth. In mass production how much does it cost to make a low precision low strength gear? I think Chinese manufacturer's could churn out such a box for less than the cost of these aftermarket ELS designs.

Paul Alciatore
04-28-2019, 07:09 PM
Back in the 1960s I studied some control systems that used analog computers. Precision pots that were a foot in diameter, 1% and better components and they still had to be tweaked to get three place accuracy, etc.

There is a reason why digital computing has taken over in almost all applications. Analog computers do work, but it is hard and expensive to get the accuracy that a digital machine will come by naturally.

Today a $25 digital device can compute with more accuracy than a $100,000 device of that era. Or of the present era, for that matter.




It was "which" Emco they copied. It would have been a different world with they had copied the V10.


For those that don't want to deal with the "cnc" aspect, what we need a small continuously variable transmission (couple of of cones?) instead of a quick change gearbox driving the lead screw. They you could cut anything within a range at the turn of a knob

ulav8r
04-28-2019, 10:04 PM
A simple ELS would be great for my approximately 100 year old pattern makers lathe that has a power feed but does not do threading. It is slightly heavier than an Atlas 6x18 so would be fine for projects such as making 5/8 x 30 threaded dies for Lyman 310 tools.

skunkworks
04-28-2019, 10:19 PM
I don't understand... 3hp wasn't enough to thread? You can make a potato thread.. here is a 500w emco with it's crap steppers threading aluminum....


https://www.youtube.com/watch?v=o2eqf9CGnUY&t







Problem is the majority that do "just CNC it" never get to the point of threading,or if they do,they find out their hardware is adequate for everything,except CNC threading.
The latter happened to a friend of mine about three years ago.He converted a Grizzly 13x40 to CNC,then found out the 3hp motor and VFD he bought were not up to the task,so upped the anty to 5hp and a vector drive($850)then found out that closed loop stepper or full on servo motors it must be(+$1800) That and a full year of his spare time and about a month of my spare time to get all the hardware sorted out.

It works fine now,but only after a lot of childbirth.He was partially fooled into thinking it would be easy,by having converted a mill previously(cake walk by comparison) and by all the youtube videos of "CNC threading linux..blah,blah,blah" where the video shows a piece of Aluminum,plastic,or shudder even wax being threaded,just long enough for a nut to thread on.

J Tiers
04-28-2019, 10:34 PM
Problem is the majority that do "just CNC it" never get to the point of threading,or if they do,they find out their hardware is adequate for everything,except CNC threading.
The latter happened to a friend of mine about three years ago.He converted a Grizzly 13x40 to CNC,then found out the 3hp motor and VFD he bought were not up to the task,so upped the anty to 5hp and a vector drive($850)then found out that closed loop stepper or full on servo motors it must be(+$1800) That and a full year of his spare time and about a month of my spare time to get all the hardware sorted out.

.......

Makes no sense..... I can thread just fine with a motor 10x smaller than that. The problem may be elsewhere.... having 5 HP is not a basic requirement for threading. It sounds more like the CAM program was at fault, as if it was trying to do full depth one pass threading.

There are dozens of videos of CNC conversions threading, NOT by any means threading wax.... Most that I have seen are taking a number of passes, not any of which would take 5 HP.

That whole concept seems just way off the wall to me.

If there was a complaint about the feed motors, I might be less inclined to question the matter, since in most they are required to hold the crossfeed and drive the threading feed. But even there, the cut is adjusted to bring it within the capability of the machine.

There just has to be something wrong with the CAM program, or inputs to it. If I can thread by hand on a 1/3 HP 10" machine, there is no possible way that it needs 5 HP to do it on a machine only slightly larger.

RB211
04-29-2019, 12:05 AM
Why is it the naysayers to home shop CNC most likely are using professional grade industrial machines at work?

skunkworks
04-29-2019, 10:02 AM
BTW - that video I posted was using the printer port.. (with a 100 line encoder + index)

sam

Dan Dubeau
04-29-2019, 10:23 AM
Why is it the naysayers to home shop CNC most likely are using professional grade industrial machines at work?

Would you want to fly airliners for a hobby? Or would you or most professional pilots gravitate more towards an ultralight or home built where everything is pilot controlled?

As someone who fits that description I think it just comes down to personal preference and they type of work you do. The things I make at home are different than at work. I don't have a need for a CNC lathe, nor would I want one. For a lot of stuff, programming a part just isn't faster than cranking the handles. Unless there is a lot of material to come off, or i'm making multiples. I just simply prefer to turn manually. IMPO it's just faster. Even at work there are lots of times I'll just run a few parts off on the hardinge, instead of firing up the Nakamura. If the QTY is under 5, and it doesn't involved threading, tapers, or curves, I'm usually faster with a manual lathe. I've been bit a couple times though when I've made 2 or 3 of something, and the customer has come back and ordered 100. That really sucks.

On the other hand give me a CNC for millwork over a manual all the time. Unless I'm drilling holes, or doing very simple xy stuff I have no interest in standing at a mill cranking handles. Manual millwork brings me no joy :). MOst of what I do with a mill involves curves and surfacing so that influence my opinion. If I never have to setup another rotary table again I'll be a happy man.

RB211
04-29-2019, 10:29 AM
Would you want to fly airliners for a hobby? Or would you or most professional pilots gravitate more towards an ultralight or home built where everything is pilot controlled?

As someone who fits that description I think it just comes down to personal preference and they type of work you do. The things I make at home are different than at work. I don't have a need for a CNC lathe, nor would I want one. For a lot of stuff, programming a part just isn't faster than cranking the handles. Unless there is a lot of material to come off, or i'm making multiples. I just simply prefer to turn manually. IMPO it's just faster. Even at work there are lots of times I'll just run a few parts off on the hardinge, instead of firing up the Nakamura. If the QTY is under 5, and it doesn't involved threading, tapers, or curves, I'm usually faster with a manual lathe. I've been bit a couple times though when I've made 2 or 3 of something, and the customer has come back and ordered 100. That really sucks.

On the other hand give me a CNC for millwork over a manual all the time. Unless I'm drilling holes, or doing very simple xy stuff I have no interest in standing at a mill cranking handles. Manual millwork brings me no joy :). MOst of what I do with a mill involves curves and surfacing so that influence my opinion. If I never have to setup another rotary table again I'll be a happy man.
I do find lathe work much more satisfying than manual milling, but CNC milling takes the cake. I like the CAD/CAM, and fine tuning it to get a nice finished part.
I can see where a hobby lathe converted to CNC could be a recipe for going bald, chasing down why it won't turn to 5 tenths accuracy.

As for flying an airliner as a hobby, I don't even want to think about them when I come home unless it is R/C or a combat flight sim. The ONLY small single engine piston airplane I have any inclination to fly anymore would be a WW2 warbird.

mattthemuppet
04-29-2019, 03:40 PM
That is exactly what I am going to do on my 9x20.I recently bought one of the cheap ($120) blidgeport table feeds off fleabay.I'm going to extend the rigthand end of the lead screw and mount the table feed there.So then I will have fwd/rev and VS feeds and be able to keep the thread change gears in the most used range.

How I wish the Chinese had copied some lathe other than an Emco.

neat, I look forward to seeing the pics. If (and that's a big if) I ever get round to it, something like a 23VDC motor out of an electric scooter and a small DC motor controller would work fine. Direction is a simple DPDT switch deal and full speed (say for reversing the carriage) can be done with a momentary switch joining the 2nd and wiper wire of the variable resistor used to set speed.

To be honest though, there are so many more pressing projects in between now and this one that I'm more likely to buy a new lathe before I get round to it :)

Dan Dubeau
04-29-2019, 03:52 PM
As for flying an airliner as a hobby, I don't even want to think about them when I come home

That's the way I feel about CAD/CAM, and CNC when I go home...... Also the reason I tend to gravitate more towards fabrication projects, and building stuff from my available "junk" pile at home. I tend to knock out my machining projects at work on breaks or stay late, and prefer to hide behind a welding hood when I'm at home.

JCByrd24
04-29-2019, 04:14 PM
neat, I look forward to seeing the pics.

Here's mine. Very simple and about $50 in parts. The speed controller I got has a voltage output but it doesn't read well at low speeds, so the next step is to just add another $20 amazon tachometer to the feed rod like I did on the spindle. I made a chart in excel already that tells me what feed rod RPM I need for a given spindle speed and feed rate (in/rev) so I'm no longer guessing.

http://bbs.homeshopmachinist.net/threads/76196-New-powerfeed-for-lathe-DC-variable-speed

Baz
04-29-2019, 04:35 PM
. Precision pots that were a foot in diameter, 1% and better components and they still had to be tweaked to get three place accuracy, etc.
.
Nowadays one would use a cheap pot and digitally calibrate it to 0.5%. Thanks for the idea. If I get back to my project instead of the expensive BCD encoded switch I will make a clickstop for a pot and calibrate it to act like a switch.

skunkworks
04-30-2019, 08:41 PM
I think the main issue is people pick cnc software that doesn't work very well for threading... Then from that point on they think that pc based cnc suck. (they don't)

sam

current systems that seem to work with encoder threading are

Linuxcnc
acorn
Mach4 with a couple of motion controllers (pokeys and hicon?)...

mattthemuppet
05-01-2019, 10:56 AM
Here's mine. Very simple and about $50 in parts. The speed controller I got has a voltage output but it doesn't read well at low speeds, so the next step is to just add another $20 amazon tachometer to the feed rod like I did on the spindle. I made a chart in excel already that tells me what feed rod RPM I need for a given spindle speed and feed rate (in/rev) so I'm no longer guessing.

http://bbs.homeshopmachinist.net/threads/76196-New-powerfeed-for-lathe-DC-variable-speed

thanks! I even remember reading that, although I had forgotten that I remembered :) Good to hear that those cheap all in one DC speed controls from Amazon do the job with lower power DC motors.