What's new
What's new

Granite Flat Calibration and Lapping

Bruce Griffing

Titanium
Joined
Jan 1, 2003
Location
Temple, Texas
I recently read a post that made reference to a Planekator - a device for checking the flatness of a flat. The real live device has a calibrated granite straight edge and an indicator that will read 20 millionths of an inch. You can set the planekator up on a flat with its two supports and measure the space between the straightedge and the surface of the flat. By doing this in several orientations, you can build up a map of the surface height. There are some details in the math, in that all of the measurements are relative, but I am not worried about the mathematics of that. Having recently discovered an affordable electronic indicator that reads 50 millionths or 1 micron, I am encouraged to build my own Planekator equivalent.

http://www.shars.com/Online_catalog.htm?sec=126&lang=1033
(indicator on page 23 on this quirky catalog)

I believe that I can do this, but I have two questions. The first is building the bearing surface on the supports for the straight edge. My first idea is to use a hardened rod sitting in a vee way on a part with a scraped flat base and vee way. The straight edge would be supported by two of these parts. I also thought of a knife edge, but thought that that would damage the straight edge. I would be interested in other ideas for the supports. The second question relates to the lapping of the plate to remove high spots. I have done lapping in the past with a variety of abrasives. I assume this is done with diamond paste. But I sure would like to read a good reference on this process. Any thoughts on this would also be helpful.
 
I'm interested too Bruce; been having thoughts along the same lines. I'm curious as to what is used to lap the plate and how you clean it afterward. I guess a cast iron lap would be softer than the granite, thus the grit embeds in the lap and not the plate?

FWIW: Had the bright idea to find an old plate, and take it up to Starrett's place in Mt. Airy, NC. A phone call this past Friday to them revealed that they ain't there no more. Bought Tru-whatever and merged operations, sending the Starrett side of things to MI, or MN, or some damn where I'm not driving to.

Thus, my renewed interest in surface plate flattening.

Haven't found any leads on local companies to do this; would like to get some prices.

Rob
 
I've seen a few anf it's hard to beat them for elegant simplicity. But you do need a calibrated straight edge otherwise you're stabbling in the dark.

The risers can be 1-2-3 blocks in combination and the follower can be built on a 1-2-3 block. All you need is an indicator and a stable environment. At this scale of accuracy, the radiant heat of your body over time will affect the straight edge even if you don't touch it. So don't stand in one place too long.
 
Forrest -
The reason I suggested hardened rods vs a 123 block is to give a line of support to the straight edge rather than an area.

planekatorsupport.jpg


This is the cross section of the support I had in mind. Just what a kinematic design text would suggest. But using a simple scraped block would be much simpler. Do you think that would work as well?
As to the calibration of the straight edge - I am also working on that.
 
I am in Phoenix and we have Washington Calibration from here in town come in and do our granite plates every six months. The guy isnt too talkitive but he uses a cast iron plate with what looks like baby powder to me.
 
I'm sure it's not the same guy, but ours is as yours sounds.
From what I could see, cast iron plates, some kind of slurry, and Mahr gages with a camera "bulb".
 
Bruce, all those loaded tangencies are great in theory but less than handy when moving the set-up around on the surface plate as an assembly. Trip up on a smudge of hand grease and the whole thing would come down in a hell of a crash. A single fitted block for each Airey point would be the approach I would encourage. Line contact is preferred for geometrical perfection but area contact is acceptable of clean and even superior if highly loaded.

Some major manufacturer's tool div (was it Rahn or Federal?) had the whole schmeer for sale once in various sizes for only a small fortune. Their follower was a real metrological miracle but I've never handled it. As I understand, it used a reed hinged mechanism to acuate the indicator. I've never seen one in the flesh; I wish I'd paid more atttention when I had access to one.

There's a federal spec that covers surface plate calibration. I wush I still had access to a good technical library. Imagine just going down the hall to all those shelves and shelves of solid info.
 
The powder they use is diamond dust. I have a friend that owns a surface plate calibration co. They use the diamond dust and a cast iron lap to surface the plates. (He like to hire guys from the local gym to moose the laps around on the surface plates LOL)It's a bust ass job.

Raun used to make the Planeckator(spelling?)but they have been bought out I don't know if the buyers still make the calibration equip or not
 
Bruce and Company --

The Planekator was originally developed as a commercial product by the Rahn Granite Company, and has survived Rahn's merger into Tru Stone and Tru Stone's later merger into Starrett. A bit of a picture is on the Starrett Tru Stone website at http://www.tru-stone.com/pages/smp.asp

The Planekator straightedge support blocks are different, one is a fixed-height block with a couple of fences to help keep the straightedge in place, the other is an adjustable-height block. If I remember correctly, the actual adjutment is done by deflecting flexible elements in the support, sort of like squeezing a pair of parentheses-shaped springs together would make them narrower and taller.

The thought behind the adjustable-height support block is to allow the table-to-straightedge distance to be mechanically equallized at the ends of the to-be-calibrated line rather than having to analytically correct for straightedge rise-or-fall.

(Like Forrest, I wish I'd paid more attention to the adjustable support the last time a Planekator came into the plant.)

A homebrew version of the Planekator is very "do-able", but personally I'd hesitate on the digital gage. I suspect that better system-level accuracy can be achieved by interpolating needle position on a 0.0001 inch per division mechanical gage.

John
 
John-
I thought that I would not get enough resolution from a .0001 indicator. My reasoning is that I will be looking to get the whole flat to within about a tenth. If the digital indicator worked well, it would make taking the readings pretty routine. I will also be able to hook it up to my laptop, where the data will be analyzed anyway, and collect the data directly. Since it is a little more sensitive, I had planned to use the 1 micron scale and convert the whole map back to inches when done. But I am interested in your reluctance on the digital indicator - can you expand on that? BTW, I already have a .0001 mechanical indicator - so when all the parts are in I can also try that.
As to the adjustable support, I had planned to forgo it and make a linear correction to the data. I also thought that since I planned to scrape the two supports, I could get them pretty close to each other anyway. But that could turn out to be wrong.
 
Bruce,

One advantage of using a digital indicator with an SPC output is that you can do the data recording on either a little SPC terminal, or on a laptop.

I have a 50 millionths Mitutoyo dial indicate that I bought from someone here on PM and it has an SPC output. Now I just need to find a long, accurate straight-edge


Robert
 
Bruce --

The reason I'd hesitate to use a digital gage is that the more-or-less-typical uncertainty specification for digital equipment is something along the lines of "plus-or-minus (full gage displacement divided by some-big-number) plus-or-minus (least count of display)".

If the SHARS digital gage is spec'd that way, the uncertainty of the displayed value would be no smaller than 1 micrometer.

On the other hand, conventional mechanical gages will almost always repeat within a quarter of a dial division . . . and it's almost human nature to interpolate to a quarter division or so when using one of these gages.

The bugaboo of the conventional gage in this application is, of course, the potential of reading / recording error.

John
 
I can see it now - Bruce, the planekator rental man. Coin of the realm will be cast iron objects, granite acceptable with two forms of ID.

On indicators: I'd be a little worried about the mysteries and steps going on inside a digital. Somewhere Evan posted about how these tick I think. Personally, I have one very sensitive tenths DTI I would lean toward, and a small battery of Federal tenths DI's that with a little cleaning would also serve. I feel better when I can see the needle wiggle and know when I've twitched. SPI output is a bitch with a 50 YO indicator though. My solution is to install daughter on a stool with a pencil, and holler the numbers out.

Maybe I'll go all out and make my own surface plate. I live on top of a few billion tons of granite; see slabs going by all the time. Wonder what kind of a plate regular gray granite would make?

Rob
 
John-
On the digital indicator, I made the assumption that it would behave like most digital electronics I am familiar with. Specifically, that it would be linear and monotonic. In short, that it would have good relative accuracy over a short range. But that assumption does not derive from the specs and may in fact not turn out to be true. After I test it I'll report the results.
 
Re: pepo
I am in Phoenix and we have Washington Calibration from here in town come in and do our granite plates every six months. The guy isnt too talkitive but he uses a cast iron plate with what looks like baby powder to me.
We also use Washington Calibration where I work. They send Dean (or Dene?) Jones up here every September to check our granite plates and lap them if necessary. As has been mentioned, that "baby powder" is diamond dust. He doesn't come to visit, he comes to calibrate our surface plates. I found out he does enjoy talking about elk hunting, though..
 
I wouldn't worry too much about the absolute accuracy of the straightedge when testing a flat using the "Planekinator" method. You can calibrate the straightedge right on the surface plate under test but you do have to make some attachments, some exacting calibration runs, and manipulate the data collected.

Support the straight edge at its Airey points. Test the surface plate at precisely spaced intervals with the straight edge reference edge down. Record the readings to as high a resolution you can manage. Invert the straightedge holding the reference face against large stiff parallels. Place the parallels on tall standards so the straighedge is suspended under the parallels precisely over the track of the first test run. Run another set of readings at the same follower location.

On the second run, your follower should track the same point on the straight edge and if a different follower is requires its base should match the dimensions of the original. Ordinary surface gages, rods and snugs are inadequate for this work.

Position sensitive indicators over the Airy points and place the follower at the center of the straight edge. Jack carefully under the center of the straight edge and note the deflection at first movement of the Airey points. Plot a deflection curve on the same follower data points. Work sum and difference math to separate the straight edge error, deflection, and surface plate error.

Once you know the straight edge error and its deflection, an error map table can be constructed.

Be aware this is as sensitive a work as can be managed in a home shop. Pick your weather for minimal temperature swings, go over your equipment, make trial runs to ensure you can take repetitive readings consistantly. Especially make sure you can eliminate sources of heat differences both environmental and radiant. Wear long oven mitts or manipulate the apparatus remotely with wires etc. Even the heat from a 100 Watt light 15 feet away can throw error in this technique. Your body will radiate up to 300 watts of heat.

Whenever you deal with 0.0001" your equipment has to reliably repeat within 25 millionths. Closer would be better. 25 millionths over 36" equals one part in 1.4 million or so. NIST has trouble working to this degree of refinement in larger scales in all but their most closely controlled labs.
 
Forrest-
Thanks for this useful input. Let me see if I understand it. Let's ignore the correction for straight edge sag until I understand the basic setup. In the first run you support the reference edge facing the surface plate and measure the gap between the surface plate and the straight edge. I think I get that. In the second run you flip the straight edge over and reference its position from the top surface?? Is that it? If so, what gap do you then measure? I guess I don't yet have it.
 
Inverting the straight edge will double its error and differentiate it from the error in the surface plate. It's tough to explain without sketches or a demo but once you see it in action you'll holler "D'oh!!"

Don't forget to figure the indicator (or gage head) senseing from the same origin. Uncontrolled trips through a thicket of erroneously signed numbers can be frustrating.
 
Forrest-
I think I get the idea of inverting the straight edge. In principle, if I knew the distances between the straight edge reference surface and the surface plate surface in both orientations, I would have enough information to do the sums and differences as you suggest. What I don't see is how to construct the second setup and make the second measurement. I don't want the thickness of the straight edge or the geometry of the supports to enter in. I just can't visualize the second setup.
 








 
Back
Top