What's new
What's new

diy surface plate lapping

Demon69

Titanium
Joined
Mar 3, 2013
Location
Area 69.
Well the s/h 3' x 2' granite plate i bought a couple of months turned out not to be quite as flat as id hoped! :o. Sooo i thought id have at it with some of 3M's finest to see if i could improve things. First signs (dare i say) seem pretty good but im gona have to wait until tomorrow to check it further, was certainly a workout.
Will try and upload some pics in mo, Its all a bit Heath Robinson so dont say you havent been warned! :D
 
So heres what i started out with. The 2' camel back (which it has to be said was being scraped from this plate) prints along its length when blued against the plates long side, but only prints edges taken across the short side. This suggested to me that theres a belly running along the plates length. Adding this to previous problems i had trying to scrape a 5' se across its diagonal and the belly seemingly being confirmed with a level, i started to look up threads re lapping etc.

IMAG1660.jpg IMAG1664.jpg
 
Last edited:
Reading up on the subject i came across a couple of posts by Richard King, not really to do with lapping as such but how you can check a plate for flatness using just a level. l liked the idea straight away! Thinking if i can check it with a level, i can probably make improvements with a level!

A quick trip to my local paint factor sorted the 120 and 320 grit 3M wet and dry that I hoped would work. This was after id had a quick go with the 800 grit i already had laying around. A definite improvement!
IMAG1666.jpg
 
Next i sat down to try and get my arms around the numbers. The level i used is an ebay oldie. The numbers on the front note 1 Div = 0.05mm per 1m. From that i worked out (hopefully correctly) over 230mm (the centers of level feet) 1 Div = around 11microns. There were several point to point areas on the edge of the plate that where in excess of 1 div
I wasnt too bothered about numbers as such but more about concerned with being able to measure and sand the edges of the plate parallel. Also being able to read the level well enough to call it good. Reading glasses helped! ;)

IMAG1671.jpg IMAG1673.jpg IMAG1674.jpg
 
Most of the sanding took place along the long edges of the plate as youd expect, constantly checking the plane of the worked area and the level of the plate. I had to re adjust many times due to movement (from sanding effort i guess). I used a wet and dry covered stone (the one I use to debur) as a block, mainly to take down the highest high points, and the camel back wrapped in the same to smooth things up and go over the whole plate with. Couldnt bring myself to take a pic of that!! :o

So this is what i was left with.

IMAG1689.jpg

Be a crime not to blue it up and have a quick check tho eh! ;)

At first, disappointment! When i checked the short span of the plate, it printed better (showed trace in the middle this time) but still very heavy on the ends.
IMAG1691.jpg

Then confusion! It printed the same heavy on the ends pattern along the length of the plate! Which had been much better before!
IMAG1690.jpg
 
Last edited:
So what can i conclude from that? I think (hope like hell) that heat from handling has distorted the SE to print concave, cant be sure till I get out there tmrw and have a look. It did come as a bit of a surprise though as in general i was very happy with how the level worked out.
Will soon see! :).

Time for bed! Gimme scraping any day! This lapping malarky is hard graft!

Cheers D
 
Lapping a surface plate? You are working to the limits of accuracy attainablle in an open shop. Also you need a better straightness reference than even the best scraped 2 ft straight edge. I men you can calibrate a 2ft x 3 ft granite straight edge using only a 2ft straight edge but the probability for error is high and the process even if perfectly executed would take a month of evenngs.

As a minimum equipment expedient I suggest a pair of 10 arc second levels. One level will be the reference which stays in one place on the surface plate and the other will be the working level. The levels are used only in line or parallel with each other. Short (8") cheap import levels have an advantage for this application. Long 15" and 18" levels span too much of the surface bridging short pitch undulations.

The first step is to calibrate the levels together in the same spot on the surface plate so they both agree. The next step is to lay out a British flag on the plate surface, that is lines parallel to the edges but 2" in, crossed center lines, and diagonals connecting opposite corners.

The following process is comparative. You don't really care about actual level readings so long as the bubbles are on range of the levels' graduation. Your are lookin for defferences: differentail readings the essense of comparative technique

Having verified both levels are in agreement and the surface plate is laid out in a grid with diagonals, you are ready to start. Plant the reference level in one corner and align it on the long edge. Set the working level right in front of it and note the difference in vial readings between reference and working levels. Thus you are comparing readings not recording absolute readings. Record the differentail reading on the map you've drawn showing the diagonaled grid. Procede down the length holding the working level in line with the reference level in 4" increments. Note differential readings as you go. Rotate the reference level along the short side, the center lines the other margin and diagonals etc ensuring both reference and working level are eighter in line or parallel.

If the plate is truely flat the differential readings will all be zero within your ability to interpolate between vial graduations.

Secondarily, mount a 0.0001" graduated dial test indicator on a surface gage and extend it out a few inches to register local undulations. Zero the dial at some arbitrary point on the surface plate and while minimixing heat input to the apparatus and returning frequently to the 'zeroing site" scan the whole surface and note localized departures from zero reading, focusing especially on the plate margins.

Be very careful of heat. Use fluffy kitchen mitts or doubled towels to manipulate apparatus. The heat of incandescent lamps, nearby warm masses like water heaters, sunshine from a window, a warm wall, furnaces, and your own warm body will have an effect on your readings. You are working to 0.0001" over three feet. That's one part in 2.8 million

This is a very tedious and time consuming process but unless you gain access to a Federal electronic differential level system, an optical collumator and right angle mirror, a laser interferometry alignment system, your home shop lapping job is doomed to be of uncertain quality without some means of verification. Two short cheap import 10 arc second levels used with a comparative technique is the best expedient I can work out to support home lapping of your granite surface plate.

Is there a calibration service handy to your location? If you drop the plate at ther facility a re-lap and calibration is not that expensive. Last figure I heard was $120 for a 2 x 3 ft plate. Someone correct me. Shipping is the killer.
 
Thanks for the input Forrest. Points noted, especially re the length of the level and and union jack style map technique. I found this on youtube that shows the general process of measuring a plate.
https://www.youtube.com/watch?v=NWObTpn6dTk

Re the level. I think the one I have is 10 arc seconds (sorry, im not up to speed on the angle stuff yet). 1 Div (graduation) is = 0.0005" over 10". It is 12" though which id agree is far too long. Id expect a lot more resolution with a shorter level of same accuracy.It did cross my mind to fashion some sort of table with feet at 4" centers that the level could sit on. What i did try to do was place the levels feet in the same place, ie place the following foot where the leading foot had been, just so i could concentrate on the overall plane rather than the local undulations. The aim being to get the same reading from point to point.
Im not sure i fully understand the need for two levels though. I mean if you start in one place and measure the rise and fall from point to point, you could in theory at least, map the whole plate. Providing the plate doesnt move.

I did try to lash up a repeat O meter style contraption, but i dont have dti sensitive enough atm. Lets just say that my setup lacked rigidity too! lol.

Re getting it calibrated and lapped. Im thinking it will be a kings ransom around these parts, but i might get a quote, just to see. :)

Your points about heat transfer bore out well last night! I took some prints this morning and can happily report full length (well almost) blue on the SE again. Anyone in any doubt about heat messing with a SE ability to stay straight should compare the pics to last nights results! Thats what ya get when you use as SE for a sanding block! :D

Along the plates length
IMAG1695.jpg View attachment 77705

Along its width
IMAG1696.jpg View attachment 77704
 
Just to gloat, I have been thinking of doing the same thing myself....... I live in Montana and just to acquire a decent sized surface plate implies criminal shipping charges. I just got my Davidson D600 comparison autocollimator. Measures to 1/10 of a second of arc. Just as you described using 10 second levels, this autocollimator compares two mirrors, on a stationary reference and the other the moving measurement.
So, since I love to keep getting more and more projects started at once (they are all times to be finished before I die), I was pondering buying a bigass chunk of granite locally and lapping it in.
Flame away people.........
 
As you've noted the level is a bit longer then convenient for checking the plate, you can use this to your advantage. The linear sensitivity of a level is inversely proportional to the length of the measurement base, which means that at the 10" ref length of the level one dev = .0005 if you were to put some "feet" on the base of the level the same level will move 2 dev for .0005 rise if the distance between the "feet" is 5.0" if the "feet are 2.5" apart .0004 rise will move the bubble 3 dev. by reducing the apparent length of the level base in your level will allow you to map the plate with finer discrimination, but not necessarily greater precision. the proper lapping of a plate would require a cast iron plate about 12x18 for the bigger areas, a 6 x 6 cast iron block for local high spots and Silium (sp) abrasive and lots n lots of elbow grease. as Gernoff noted an autocollimator provides a high order of precision but its still alot of work and calculations. I use electronic differential levels and a mapping program on my laptop. The "British flag " pattern Forrest described is referred to as a Moddy flatness check or an 8 line calibration pattern in Mill spec GGG-p-463c
Cheers Don
 
gernoff
Just to gloat, I have been thinking of doing the same thing myself....... I live in Montana and just to acquire a decent sized surface plate implies criminal shipping charges. I just got my Davidson D600 comparison autocollimator. Measures to 1/10 of a second of arc. Just as you described using 10 second levels, this autocollimator compares two mirrors, on a stationary reference and the other the moving measurement.
So, since I love to keep getting more and more projects started at once (they are all times to be finished before I die), I was pondering buying a bigass chunk of granite locally and lapping it in.
Flame away people.........

Thats the spirit! ;). Just be aware that it took me a fair effort to remove a guesstimated 0.0003" over one 6" x 20" area! I dont know what you mean by 'bigass', but if the chunk o granite wasnt formerly a surface plate, you might have to bring out the big guns. Like an angle grinder maybe! :D

Pete F
My first question, having seen the first pictures, is how is your plate being supported?

Its being supported on a stand that came with the plate. It provision for both four and three point mounting. Im using four points on the floor (which is a concrete slab) and three points to support the plate.
IMAG1707.jpg

bebop
As you've noted the level is a bit longer then convenient for checking the plate, you can use this to your advantage. The linear sensitivity of a level is inversely proportional to the length of the measurement base, which means that at the 10" ref length of the level one dev = .0005 if you were to put some "feet" on the base of the level the same level will move 2 dev for .0005 rise if the distance between the "feet" is 5.0" if the "feet are 2.5" apart .0004 rise will move the bubble 3 dev. by reducing the apparent length of the level base in your level will allow you to map the plate with finer discrimination, but not necessarily greater precision. the proper lapping of a plate would require a cast iron plate about 12x18 for the bigger areas, a 6 x 6 cast iron block for local high spots and Silium (sp) abrasive and lots n lots of elbow grease. as Gernoff noted an autocollimator provides a high order of precision but its still alot of work and calculations. I use electronic differential levels and a mapping program on my laptop. The "British flag " pattern Forrest described is referred to as a Moddy flatness check or an 8 line calibration pattern in Mill spec GGG-p-463c
Cheers Don

Re the level. If ive got you correctly about the feet, then its along the same lines as the 'table' i mentioned in my reply to Forest. I understand what you mean about the change in sensitivity if you were to move the feet closer together. It would effectively change the angle that the vial sees for a set change in height. Unfortunately my level already has feet per say which makes things a little more awkward, hence the table (with feet) idea.
IMAG1706.jpg

Re the lapping. It sounds like youve done this before :). Id be interested to read more on the abrasives used, do you have any links? I tried to search Silium but didnt have much luck.
Can i ask. When youre lapping. How do you protect the lap from the heat from your hands whilst rubbing? It was an eye opener to see just how much my camel back bent when using that! Silly when i think of it now, when scrapping it, id use rigger gloves to transfer it from bench to plate and back again, yet i didnt think that using it as a rubbing block (with bare hands!) might be a problem! :o


Will have another go (using the union jack style method) at mapping the plate today, staying in from edges as Forest suggested (along with oven gloves or some other such measure to try and keep heat out of the level). Ill aim to keep the level as point to point as possible and try and put some numbers to the readings.

Thanks all for the input.
 
Its being supported on a stand that came with the plate. It provision for both four and three point mounting. Im using four points on the floor (which is a concrete slab) and three points to support the plate.
You might check out GGG-P-463c for proper support of the plate.

Airy Points and their relative position to the plate, to be specific.
 
Hmm. The general gist of mounting a plate from what ive seen so far suggests a 3 point mounting method. These mounts should be positioned around 20 - 25% in from the edges of the plate. From memory mine are in around 10%. How important is this? Will be a chore to alter the stand! :(

http://www.practicalmachinist.com/vb/general/supporting-surface-plate-196513/

I should add that the plate is dated as 1990 and has no visible locations for mounting points.
 
Could someone who has done this before comment on filling scratches and dings with mica filled epoxy? Isn't there a special product just for this purpose?

Thanx
<jbc>
.
 
Its being supported on a stand that came with the plate. It provision for both four and three point mounting. Im using four points on the floor (which is a concrete slab) and three points to support the plate.


Thanks all for the input.

You scraped your straight edge from this plate right? I'd suggest the issue is the plate is not being properly supported
 








 
Back
Top