PDA

View Full Version : Scraping Topics



TexasTurnado
11-30-2010, 12:40 PM
I have a question for the group: does anyone have info on how much a typical camelback st edge flexes? Or, to approach it from the opposite direction, how rigid is a 6 ft cameback?

If it is scraped to virtual master plate that is perfect, and then supported at either end, how much will it droop in the middle? I can supply some dimensions for my B&S 72 inch camelback if someone will volunteer put the numbers in an appropriate sw package to come up with at least an approximation...:D

The problem is not too much unlike a suspension bridge: the roadway is the scraped bottom surface, and above that is an arch in compression with an almost solid sheet connecting the arch to the scraped surface instead of suspension cables. The material is cast iron, so the parameters for gray ci should be a good approximation..... I would expect the sag would be less than what would result from a solid rectangular sheet of the same height (approx .5 in thick) on edge, due to the relatively heavy arch in compression.

dp
11-30-2010, 02:21 PM
Body heat on the arch section will expand that portion, too, no? Mine is just 18" and would be more sensitive to this. I doubt I could lift a 72" straight edge.

I have a rough casting I need to get going on but I'm going to cheat and have the faces ground first.

TexasTurnado
11-30-2010, 02:35 PM
Body heat on the arch section will expand that portion, too, no? Mine is just 18" and would be more sensitive to this. I doubt I could lift a 72" straight edge.


I put my arms thru two of the holes to the elbow, and move it around that way....:D

TexasTurnado
11-30-2010, 02:47 PM
Check out this 12 foot model for sale on ebay, weighs in at 350 lbs!

http://i288.photobucket.com/albums/ll168/TexasTurnado/12footSE.jpg

http://cgi.ebay.com/Camel-Back-Straight-Edge-12-Cast-Iron-Camelback-level-/350418368014?pt=LH_DefaultDomain_0&hash=item51968ff60e

whitis
11-30-2010, 03:41 PM
If it is scraped to virtual master plate that is perfect, and then supported at either end, how much will it droop in the middle?
Don't do that. Support it at the Airy Points.
http://en.wikipedia.org/wiki/Airy_points

The Airy points balance so that there is equal weight on each side of the supports (adjusting for the difference between cantilevered and between supports), minimizing deflection.

Note that for a camelback, as opposed to a straight beam shape, the airy points may be differently located (than 5/9ths) since you have removed some weight from the ends.

Look at the pic texasturnado posted. See those pads on the back? Those are the points for supporting the beam when upside down or for lifting. A smaller unit should have thermally insulating handles there.

form_change
11-30-2010, 04:01 PM
It's not an easy one to answer without full information as the answer is very dependent on the exact geometry of the CB and it's mass.
However, I did a calculation for a 6ft truss made of 1" sqr tubing braced with 3/4 x 1/8" strip, supported 22% from each end and got an answer of 3/100 th of a thou for deflection under it's own weight. That formula came from an unconfirmed source and I've yet to check the sums, but seems reasonable.
I think the rough answer for you is 'not much at all'. For home use it can probably be assumed to be neglegible, as thermal effects will start to influence results at that level.
I once came across a story about making extremely high precision surface tables at Mitutoyo. Apparently the plates would be allowed to rest between scrapings because rubbing the high spots would heat them up and make the surface (microscopically) bulge. That was in an underground thermally controlled room too

Michael

lazlo
11-30-2010, 04:20 PM
However, I did a calculation for a 6ft truss made of 1" sqr tubing braced with 3/4 x 1/8" strip, supported 22% from each end and got an answer of 3/100 th of a thou for deflection under it's own weight.

I think the rough answer for you is 'not much at all'. For home use it can probably be assumed to be neglegible, as thermal effects will start to influence results at that level.

That would be my guess as well -- that the amount of sag would be negligible. A camelback truss is specifically designed to minimize sag.

There's a section in Moore's Foundations talking about the geometric limitations of the traditional scraping master shapes (and why they designed the torsion-box reference), but they were scraping to microinch standards.

oldtiffie
11-30-2010, 04:39 PM
I'd have thought that this is a classic use for a good finite element analysis (FEA) software package - often found in high-end CAD or modeling packages.

There are several here who have and presumably use those applications.

I don't have an FEA application/package.

In the practical world (ie in shops and outside Universities and specialised companies and controlled environments) there is no perfect surface plate or reference and even a Class AA plate has its "ups and downs" which needs to have been checked and "mapped".

The map will give the optimum "flat" line on that surface plate.

The surface plate will have an optimum support and maximum load specification.

The camel-back level can only be referred to the plate with the level in full contact (ie fully supported) with the plate.

There will be some "sag" (deflection) if the level is supported at say a single point at mid-span and at the ends (2 points). There are any amount of possible combinations.

The camel-back while very stiff as regards deflection is elastic such that deflection while minimal will not be zero.

Temperature changes will need to be considered as if the solution is to be practical it must be applied over a representative range of temperatures as might reasonably apply in a "shop" environment.

The FEA application should be the quickest and most theoretically accurate solution.

But I guess the real answer is to ask the manufacturer/s of such camel-back levels as they would have classified it as being of a certain grade that had specified limits.

That should not be unreasonable as the camel-back level, if the application requires it, should be checked and re-calibrated and if necessary re-conditioned periodically else it will be "out of test".

Forrest Addy
11-30-2010, 09:18 PM
Let's look at the problem from a couple directions.

The question can be phrased: "how much does a 6 ft camelback straight edge deflect from its own weight?"

There's a couple of basic designs and several maker having the same basic design but of slightly different proporions. These make for small differences in the actual deflection figures. If you have certain knowledge of the deflection characteristics you can cook up an error map of the straight edge's behavior in different attitudes.

You can calculate until the cows come home but in the end you will get a theoretical number that may or may not prove out under actual load tests. The only thing that counts in demensional inspection and load tests is actual numbers you can take to the bank; that is, numbers that are second party confirmable. Anything else is intellectual exercise.

That is, if all these high falutin considerations are important to you.

I propose this: Run a load test as through the straight edge was a bridge or a crane or something.

The home center sells gravel in handy bags all weighed out. Buy one for a trial weight. Retain the receipt if you want to return it. Set the straight edge on solid support on the ends (to start). The solid support should trully solid. I suggest the thick concrete top of a retaining wall. Bed the support (a pair of 1-2-3 blocks) to support with maybe 3" of overhang. I was thnking of a couple blobs of sealing wax on saran wrap. The saran will comply with concrete and the wax (ealing wax is rigid when set) will solidify to provide full support to the blocks. The support will be good only for that partcular exact place on the concrete. Then arrange a 0.0001" reading indicator in the center to register deflection of the straight edge.

Set up the straight edge. Set a zero. Load the straight edge with the bagged gravel. Note the reading. Remove the test load. Observe if the indicator returned to zero. If not tinker with set-up and other variables until you get repeat zero. Note: the heat of your hands and body will will locally expand the straight edge and thus have significant effect on the indicator readings. You do need a concrete wall or a very stiff machine tool as support for this test. A flat floor or a light duty lathe bed will not be adequate.

Weigh the straight edge. Taking the weight of the test load and the deflection readings, do the ratio and proportion math to adjust the deflection to suit the actual weight.

This test returns a value for end support and a concentrated load. You can subdivide the load into smaller bags and distribute it. Move the supports to under the insulator features, etc. Have fun but take notes.

Camelback straight edges are pretty stiff regardless of design refinements. I doubt if you'll get load deflection readings over a couple of tenths (0.0002"). A straight edge can be considered a spring. I the 6 ft unit under test weighed 150 lb and deflected 0.0002" then the rate would be 750,000 lb/in: pretty stiff.

NzOldun
11-30-2010, 11:14 PM
The design of a camel back straight edge is so that, in theory, the amount of deflection at the two extreme ends, is exactly the same as in the middle of the span between the feet (support points) - C.F. Machinery's Handbook, 16th Ed pp388, for the simple, uniform cross section beam case.

For a 3-5 ft straight edge which would weigh in the range of 50 - 100 lbs, the likely deflection under it's own weight would be in the range of 0.00001" (i.e. in the hundred thousandths of an inch). At this level of accuracy, you need a temperature controlled environment and hi-spec laboratory instruments to take the measurements. Rather academic for the HSM, IMHO.

J Tiers
11-30-2010, 11:59 PM
For a 3-5 ft straight edge which would weigh in the range of 50 - 100 lbs, the likely deflection under it's own weight would be in the range of 0.00001" (i.e. in the hundred thousandths of an inch). At this level of accuracy, you need a temperature controlled environment and hi-spec laboratory instruments to take the measurements. Rather academic for the HSM, IMHO.

Which accounts for my comments that any sensible and reasonably good CB straightedge would act as quite a good enough reference to deal with flattening the problems the lathe in the original thread is alleged to have to a point that the OP there would find quite nice.

No percentage in discussing fractional tenths over a boutique beer..... at some point you need to start spreading blue, marking the work, and making CI dust fly.

oldtiffie
12-01-2010, 12:54 AM
I have no issue in general with what Forrest says as regards the direct measurement of deflection related to a specific camel-back level except to say that as the loading on and of the CB bottom chord is more a uniformily distributed load than it is a point or concentrated load that the load should be re-distributed. Further, the loads on the bottom chord should be applied at the beam bottom chord node points (where the struts and the bottom chord centres intersect). If not applied a the node points, but between them, the bottom chord elements between the nodes the beam segment which would be notionally in tension would have a concentrated bending load applied to it and would deflect in addition to that which occurs for the entire truss (ie camel-back level).

But I can't see the flatness of the level being any better than the best surface plate that will accommodate the level in places other than just the diagonals of the plate. I cannot understand the undue emphasis on putting the CB level on and only on the diagonals of the surface plate as that diagonal should be no flatter than the rest or any part or parts of the plate.

Using Starret's catalogue the best - for a 6 foot level - would seem to be a 6 foot (72 inch) x 4 foot (48 inch) surface plate which has a diagonal of 7.2 feet (~ 7' - 2.5") which gives a large area to deal with the CB level as well as improving the options of which points on the plate (from its "map") to use to optimise flatness to scrape the CB level to.

The "Starrett" allowable errors in 6' x 4' surface plates of various grades measured along the diagonals are:

Grade AA (Laboratory): 0.000350" (0.0085mm)

Grade A (Inspection): 0.00070" (0.0170mm)

Grade B: (Tool Room) 0.0014" (0.035mm)

It is unlikely that the CB level will be more accurate than the plate it is scraped to. It is more likely that it will be one grade less.

I can't find any references to the standards for CB levels but my guess is that it would be similar to surface plates and perhaps with scraped-in positive camber so that under load it does not bear unduly on the centre. This is similar to what Starrett does with its precision levels as was discussed at some length not so long ago.

The Starrett standards/limits for surface plates are at page 396 of:
http://ecatalog.starrett.com/Default.aspx#401

oldtiffie
12-01-2010, 01:47 AM
I'd be pretty sure that the CB level sole-plate will only be level and co-planar (ie no "twist(ing))" if used in a vertical position.

I could see that it will work very well on the flat parts of a machine bed, but I'd be a lot more than hesitant to use it it on the prismatic sections - ie the "vee-ways" which not only have to maintain a constant cross-sectional shape but have its axis not only straight but parallel to the machine other vee-way/s but parallel in both the vertical and horizontal to the machine/lathe head-stock spindle axis.

The maintenance of prismatic ways and sections is several orders of difficulty and skill more than for flat surfaces.

Most straight edges are of relatively thin vertical ("x" - "x") section and have a very good resistance to bending in the vertical plane and work very well. Being relatively thin in the "Y"-"Y" horizontal plane they bend very easily under their own weight unless supported externally in-plane.

Most can relate to this happening with a 12" rule or a 3 foot straight edge.

A CB level will increasingly develop a cross-axis "bow" as it moves further from vertical.

I'd be very reluctant to use it on flat faces on a prismatic vee-way - the more so and many prismatic faces are 45 degrees from the horizontal.

Forrest Addy
12-01-2010, 03:13 AM
Inclining amawkwardly handled straight edge poses the risk of twist and sinuosity in the reference edge but if the straight edge is supended by the neutral axis the error can be held to a fracton of that possible.

The error itself is difficult to quantify but I can tell you from personal experience that succeeding prints taknen along an inclined way surface may be surprizingly inconstant - enough to baffle you until you work out the sources of the error and how to control them.

Back in the day when I worked with long heavy straight edges, I used an equalizer rope supension attached to 4 C-clamps secured to the web along what I took to be the neutral axis (a mid-line arc between the face and the back.) With the straight edge so supported there is little twisting moment appled to the straight edge casting and the tool can be inclined to any degree needed with little loss of reference face accuracy (there's always some error, the supension isn't perfect.)

Three short pieces of rope knotted to form a double W, four C-clamps, and a crane hook is all that's required.

oldtiffie
12-01-2010, 04:51 AM
Originally Posted by NzOldun
For a 3-5 ft straight edge which would weigh in the range of 50 - 100 lbs, the likely deflection under it's own weight would be in the range of 0.00001" (i.e. in the hundred thousandths of an inch). At this level of accuracy, you need a temperature controlled environment and hi-spec laboratory instruments to take the measurements. Rather academic for the HSM, IMHO.
Which accounts for my comments that any sensible and reasonably good CB straightedge would act as quite a good enough reference to deal with flattening the problems the lathe in the original thread is alleged to have to a point that the OP there would find quite nice.

No percentage in discussing fractional tenths over a boutique beer..... at some point you need to start spreading blue, marking the work, and making CI dust fly.

Well JT,

I guessed you missed the bit about the lathe-bed and its ways being hardened and with at least (so far as is known from the OP), 0.008" to be removed from the most of the ways. I can't see too much of that being scraped off manually by hand - or with a powered "Bi-ax" scraper either.

A "straight edge" may (or may not) be OK for the two "flats" (one each for the head-stock, saddle and tail-stock) and I'm pretty certain that a straight edge won't be a lot of help on the two prismatic ("vee") ways (again, for the head-stock, saddle and tail-stock).

The OP has pretty well settled on having it ground - if and when anything is done at all.

If the bed is precision ground, I can't see any real need for scraping it other than mottling it for lubrication and "looks".

At the moment it seems that even with the saddle almost or actually touching ("fouling") the bed, the lathe does well enough as it is and will do until something is done.

Perhaps the OP will just grind off part of the underside of the carriage to relieve the fouling when he has the saddle off as an interim measure.

Its all up to the OP.

J Tiers
12-01-2010, 09:45 AM
Well JT,

I guessed you missed the bit about the lathe-bed and its ways being hardened and with at least (so far as is known from the OP), 0.008" to be removed from the most of the ways. I can't see too much of that being scraped off manually by hand - or with a powered "Bi-ax" scraper either.

A "straight edge" may (or may not) be OK for the two "flats" (one each for the head-stock, saddle and tail-stock) and I'm pretty certain that a straight edge won't be a lot of help on the two prismatic ("vee") ways (again, for the head-stock, saddle and tail-stock).



There are many "ways" of making CI fly....

You have to get off your butt and DO (or contract for) one of them...... otherwise nothing happens. perfection is wonderful, but most of the time, "good enough" is just that.

Better to "be a Russian", and get 80% of the way to perfection now, rather than 98% of the way in the sweet by and by. Perfection is for the Czar, the reality of pretty good is for the people now.

Discussions of thous, degrading into tenths, and then onward to microns are just mental exercises and "flatten no dips", although they DO seem to make the "dips" come out of the woodwork.

Mcgyver
12-01-2010, 10:06 AM
I'm pretty certain that a straight edge won't be a lot of help on the two prismatic ("vee") ways (again, for the head-stock, saddle and tail-stock).

.

OT, you're making me mental. Calm down, sit down and listen/read. These 10 million posts and links and quotes on subject that you don't know about make the thread like walking through a swamp. Like the page long set of quote posts at the end of the last scraping thread quoting the entire thread, wtf? who's going to read all that? Its annoying. When you know a subject have it it, when you don't listen/ask questions.

A straight edge is what you scrape a prism way with. A prism way is just two flats at a set angle. You make a pair of perfectly matched scraped blocks to keep the angle and then use said pair of matched blocks to compare the vector of the prism way to the flat way and other prism way using a level and indicator - that gets the prism perfect and parallel to the other ways.

http://i785.photobucket.com/albums/yy132/michael01000/scraping/th_DSC_1650Large.jpg (http://s785.photobucket.com/albums/yy132/michael01000/scraping/?action=view&current=DSC_1650Large.jpg)

http://i785.photobucket.com/albums/yy132/michael01000/scraping/th_DSC_1641Large.jpg (http://s785.photobucket.com/albums/yy132/michael01000/scraping/?action=view&current=DSC_1641Large.jpg)

TexasTurnado
12-01-2010, 10:24 AM
Don't do that. Support it at the Airy Points.
http://en.wikipedia.org/wiki/Airy_points

The Airy points balance so that there is equal weight on each side of the supports (adjusting for the difference between cantilevered and between supports), minimizing deflection.

What you say is true, but I want to know the stiffness from the ends in this case: When oldtiffie posted the map of his surface plate (in the other "cunning plan" thread) with the convex shape on one diagonal and a concave shape on the other, I would like to know if the camelback SE would "sag" into the concave diagonal when spotting (if scraped to the convex and more accurate surface).

His plate was not long enough as made, but imagine it 6 ft on the diagonal and the tolerances multiplied by the same amount - my question was constructed to try to answer whether or not it would conform to the new shape or give different markings.

I realize this is an oversimplification of the problem (there is no data about the surface other than the end points and middle), but it is a place to start to give some idea of how much variation can exist due to the camelback SE.

TexasTurnado
12-01-2010, 10:53 AM
The design of a camel back straight edge is so that, in theory, the amount of deflection at the two extreme ends, is exactly the same as in the middle of the span between the feet (support points) - C.F. Machinery's Handbook, 16th Ed pp388, for the simple, uniform cross section beam case.

For a 3-5 ft straight edge which would weigh in the range of 50 - 100 lbs, the likely deflection under it's own weight would be in the range of 0.00001" (i.e. in the hundred thousandths of an inch). At this level of accuracy, you need a temperature controlled environment and hi-spec laboratory instruments to take the measurements. Rather academic for the HSM, IMHO.

This is just the kind of data I was looking for - assuming it is applicable to my SE. Even if it is off by a factor of two or three, it still indicates the SE should give different readings on the imaginary master plate I proposed earlier - thanks for posting.

It also means one must be careful in selection of the area of the master plate to use when spotting the SE, and why some of the apparent inconsistencies can occur... :)

TexasTurnado
12-01-2010, 11:16 AM
Back in the day when I worked with long heavy straight edges, I used an equalizer rope supension attached to 4 C-clamps secured to the web along what I took to be the neutral axis (a mid-line arc between the face and the back.) With the straight edge so supported there is little twisting moment appled to the straight edge casting and the tool can be inclined to any degree needed with little loss of reference face accuracy (there's always some error, the supension isn't perfect.)

Three short pieces of rope knotted to form a double W, four C-clamps, and a crane hook is all that's required.

This sounds like a great way to support the SE for spotting!

But I must confess, it also makes me even happier I had my beds ground - hardened or not! :D I realize even more now how lucky I am to have a great grinding shop with reasonable prices less than 50 miles away.

MuellerNick
12-01-2010, 11:45 AM
I would like to know if the camelback SE would "sag" into the concave diagonal when spotting (if scraped to the convex and more accurate surface).

Yes, it will. And you will still be able to detect it.
Imagine a bar out of rubber that you lay onto a bent surface. Even if that rubber bar is in full contact with that surface, the pressure in each square mm will not be equal. And that different pressure will give you different reading of the blue.

Nick

oldtiffie
12-01-2010, 07:15 PM
Originally Posted by oldtiffie
I'm pretty certain that a straight edge won't be a lot of help on the two prismatic ("vee") ways (again, for the head-stock, saddle and tail-stock).

OT, you're making me mental. Calm down, sit down and listen/read. These 10 million posts and links and quotes on subject that you don't know about make the thread like walking through a swamp. Like the page long set of quote posts at the end of the last scraping thread quoting the entire thread, wtf? who's going to read all that? Its annoying. When you know a subject have it it, when you don't listen/ask questions.

A straight edge is what you scrape a prism way with. A prism way is just two flats at a set angle. You make a pair of perfectly matched scraped blocks to keep the angle and then use said pair of matched blocks to compare the vector of the prism way to the flat way and other prism way using a level and indicator - that gets the prism perfect and parallel to the other ways.

http://i785.photobucket.com/albums/yy132/michael01000/scraping/th_DSC_1650Large.jpg (http://s785.photobucket.com/albums/yy132/michael01000/scraping/?action=view&current=DSC_1650Large.jpg)

http://i785.photobucket.com/albums/yy132/michael01000/scraping/th_DSC_1641Large.jpg (http://s785.photobucket.com/albums/yy132/michael01000/scraping/?action=view&current=DSC_1641Large.jpg)

Well at long long last someone - you - have addressed prismatic (vee) way finishing without just using a camelback level (which the OP in the "other thread" has).

As I said - and as you seem to have also said indirectly - there is a lot of skill required and parts to be made and acquired than just scraping flat surfaces.

But as I was sticking to the OP's project ("other" thread) and not necessarily the general or any other case, it is for the OP to decide whether and how - and if - he wants to address the bed on his lathe.

I have no issue with anyone who has the need, tools and skills that you and others have undertaking that work and I admire you and those others for it.

But never-the-less some who may not have those skills may get caught up in the "hype" and hyperbole and think how "easy" it seems to be and get caught up in some-thing beyond their skill levels, time-frame and expenses. It can be disappointing at best and catastrophic at worst for them.

I've done my bit at this sort of stuff and frankly, I thought then as I do now that if there was a satisfactory alternative I'd use it.

My general approach is that I have to decide whether to use it or lose it.

If its not worth my while maintaining or "doing up" a tool or machine, I "lose" it. In short, it gets dismantled into its components parts and rendered unusable and irreparable and sent off to scrap. I never sell, and rarely give away, any tools.

If I need a replacement, I buy one - new - and get back to "square one" asap.

As long as the tools I have are "adequate and fit for purpose" for my needs they are good enough.

I respect others rights to be different to mine. I don't think its unreasonable for me to expect to be regarded similarly by others - whether they agree with me or not.

Nice job.

beckley23
12-01-2010, 08:27 PM
Don't rely on the SE as the sole determining factor of flatness or staightness. I made extensive use of the level as a cross check. I did a bit of small area scraping that still spotted with the straight edge, but the level showed a change.
Harry

oldtiffie
12-01-2010, 09:15 PM
'zactly as I said previously Harry - many thanks.

The CB level does very well on a surface that is wide enough to provide a substantial width for diagonal and cross/end/middle checking as its quite possible to get a small "wind/twist" and not detect it with a CB or other similar level. A precision "bubble" level will soon pick it up - if the surface the CB level is checking is also level end-to-end.