Question about flatness tolerance

richard newman · Dec 3, 2019

The flatness spec for some Blanchard ground plate is .002"/ft. So for a 4' long piece, does that mean .008" over the 4' length?

I ask because if you calculate the radius of a 12" long .002" deep arc, and then draw a 4' long arc of that radius, the total curvature is much greater than .008"

eKretz · Dec 3, 2019

You'd probably better get that clarified by the grind shop quoting the job. Most likely it means a plate ground on both sides will measure within .002"/foot of parallel. Not very inspiring accuracy BTW.

Red James · Dec 4, 2019

The radius has nothing to do with flatness. The spec means just what it says, nothing about radii. The best you can do is measure over a 4' length and expect </= 0.008" total deviation.

michiganbuck · Dec 4, 2019

Don't know about that but we Blanchard ground a 36" chuck full of inserts, TCT and blades to +- .001" and better

DMF_TomB · Dec 4, 2019

richard newman said:
The flatness spec for some Blanchard ground plate is .002"/ft. So for a 4' long piece, does that mean .008" over the 4' length?

I ask because if you calculate the radius of a 12" long .002" deep arc, and then draw a 4' long arc of that radius, the total curvature is much greater than .008"

.
often there is 2 tolerances the .002" per foot is a wavy tolerance and if any parallel passes the lap or height difference between the passes. then there might be an overall flatness tolerance across the whole part surface which could be much higher.
.
typically you might have a .0005" TIR wavy or lap tolerance per foot (like when checking with indicator going across the surface it might be going +/-.00015 or .0003" TIR) and a .0010" tolerance across the whole part surface.
.
wavy or lap error often from spindle tram error (leaning slightly), tool wear, or just localized hard spots that resisted cutting/grinding more than other spots. if you blue up surface and hand stone or lap it you will often see shiny high spots often of only .0002"

TeachMePlease · Dec 4, 2019

Flatness simply describes two planes between which the entire surface must lie.

So in any one foot section, the entire surface of the part must be within .002" of every other portion of the surface within that foot.

The second foot has no relation to the first foot.

The bar could be bent 90° at the intersection of the 1st foot and the 2nd foot, and still pass flatness spec.

Dan from Oakland · Dec 4, 2019

If Teach Me had stopped after the 2nd sentence above he would have been 100% correct.

sfriedberg · Dec 4, 2019

TeachMePlease said:
The bar could be bent 90° at the intersection of the 1st foot and the 2nd foot, and still pass flatness spec.

But crucially, no foot that includes the bend would pass.

eKretz · Dec 4, 2019

It would not pass flatness. It would pass parallelism.

jim rozen · Dec 4, 2019

TeachMePlease said:
Flatness simply describes two planes between which the entire surface must lie.
So in any one foot section, the entire surface of the part must be within .002" of every other portion of the surface within that foot.
The second foot has no relation to the first foot.
The bar could be bent 90° at the intersection of the 1st foot and the 2nd foot, and still pass flatness spec.

+1

Flatness discusses *one* surface only. By its nature it cannot say anything about parallelism of two surfaces.

eKretz · Dec 4, 2019

Nobody said that it did, nor argued otherwise! The only claim made was that a plate bent to 90° could potentially pass a parallelism spec of .002" but NOT a flatness spec of .002".

jim rozen · Dec 4, 2019

eKretz said:
Nobody said that it did, nor argued otherwise! The only claim made was that a plate bent to 90° could potentially pass a parallelism spec of .002" but NOT a flatness spec of .002".

Could NOT pass a parallelism spec because the underside the plate is all wavy. But if the spec it written such that the one test area is in one
section for a foot, and another section for the second foot, then it would pass. Make the test 0.002 flatness over the entire piece, then no
it would indeed fail.

gbent · Dec 4, 2019

richard newman said:
The flatness spec for some Blanchard ground plate is .002"/ft. So for a 4' long piece, does that mean .008" over the 4' length?

I ask because if you calculate the radius of a 12" long .002" deep arc, and then draw a 4' long arc of that radius, the total curvature is much greater than .008"

It could, but it would be unusual. If you need better, call it out and be prepared to pay for it.

It sounds like you are assuming the error in flatness will be of uniform geometry. That is far from the result you will receive. You haven't told us the thickness of the plate you are getting ground. For a Blanchard, it matters. A Blanchard has a powerful magnet, it could probably pull 1/32 of bow out of a 2' x 4' plate 2" thick. The end result depends a lot on the flatness of the original stock for thin parts.

Assuming a good machine and a good operator, your part will be flat to better than .001 overall while still on the machine. The result on a surface plate will depend a great deal on the skill of the operator, as well as the requested tolerance. Your plate should be able to be within .002 overall, but you will be paying for a lot of extra machine time for shimming, testing, and flipping.

9100 · Dec 4, 2019

I don't run machines like that but I work on them and watch others run them a lot. Anything but a completely killed piece, meaning multiple heat and chill cycles, is going to warp. They usually rough them down on a Camut grinder, think of one with a Blanchard stone but a linear table like most surface grinders. It has no Y travel because the stone is as big as the table is wide. Then they turn the magnet off and slide shims under the areas that have bowed up. They grind both sides as straight as they can, then move them to a 96" X 48" Mattison, shimming as needed during the process. Next the really picky pieces go on a Chevalier for a final carefully nursed grind. The customer gets a very flat part and pays for it. Those parts are used in a very exacting application and the customer seems to think it is worth it.

Bill

CarbideBob · Dec 4, 2019

michiganbuck said:
Don't know about that but we Blanchard ground a 36" chuck full of inserts, TCT and blades to +- .001" and better

Good lord I would hope so and hold the batch of 5000 pieces to 1/4 or better of this.
Top and bottom is the first op and you hold it very tight as it will puck with the others that follow.
If you need microns down the line you had sure as shit better have nailed this one or you will be chasing Blanchard loads.
We segregate and mark thickness loads just for this adjustment so that we know when it hits us downstream.
The first op is often so way open but it cascades down the process stream.

Get the first few flat, square, Tir or whatever very tight and all the others that follow are easier.
For sure these features do not need to be way inside print but maybe you do to make life not so complicated.
Bob

richard newman · Dec 4, 2019

gbent said:
It could, but it would be unusual. If you need better, call it out and be prepared to pay for it.

It sounds like you are assuming the error in flatness will be of uniform geometry. That is far from the result you will receive. You haven't told us the thickness of the plate you are getting ground. For a Blanchard, it matters. A Blanchard has a powerful magnet, it could probably pull 1/32 of bow out of a 2' x 4' plate 2" thick. The end result depends a lot on the flatness of the original stock for thin parts.

Assuming a good machine and a good operator, your part will be flat to better than .001 overall while still on the machine. The result on a surface plate will depend a great deal on the skill of the operator, as well as the requested tolerance. Your plate should be able to be within .002 overall, but you will be paying for a lot of extra machine time for shimming, testing, and flipping.

The spec I quoted was from Nifty Bar for their stock plates 1/2" thru 1-1/2" thick. I think they use hot rolled, not sure if it's stress relieved.
https://niftybar.com/wp-content/uploads/2019/05/PriceListLO_9000_1_2017.pdf

I'd gotten a quote for touching up a 1" x 30" x 48" plate that had once been ground, but then welded onto and distorted a bit, and was told could expect flatness of .010" to .015" across the diagonals.

Thinking about it made me wonder about the .002/ft and what that really meant.

triumph406 · Dec 4, 2019

richard newman said:
The spec I quoted was from Nifty Bar for their stock plates 1/2" thru 1-1/2" thick. I think they use hot rolled, not sure if it's stress relieved.

My local Blanchard grinder really hates cold rolled bar to the extent he'll send it out for stress relieving before grinding. If it goes out with other plates then he'll do it for free.

He much prefers hot rolled plate to start with. Then so do I, cos it's cheaper.

jim rozen · Dec 4, 2019

We were sourcing silicon plates years ago, the tops had to be flat within a micron over the entire surface. These were about 6 inch by
3 inch rectangles, about 3/8 thick.

There was a company called Bond Optics that did these for us. They were checked at our end with a special tool called a Zygo interfermometer.
Basically gave you a fringe pattern over the entire surface. HeNe laser red fringes, about 600 nm, or .6 micron. They made these within
a fringe or two on the top surface. I was impressed.

Bond Optics would build ring laser gyros, all from one solid piece of sapphire. Metalized mirrors and all.

And they're still in business apparently:

Bond Optics | Home

9100 · Dec 5, 2019

triumph406 said:
My local Blanchard grinder really hates cold rolled bar to the extent he'll send it out for stress relieving before grinding. If it goes out with other plates then he'll do it for free.

He much prefers hot rolled plate to start with. Then so do I, cos it's cheaper.

A friend wanted a rectangular bar that was an improved fit in the table slots of his Craftsman table saw, the kind of job you would decline if it was a walk in stranger. I started with a 3/8" X 5/8" cold rolled bar and wasted almost an entire morning, grinding one side, then flipping it to send the warp back the other way. The damned thing almost rolled up in a ball.

Bill

metlcutr55 · Dec 9, 2019

have seen many jobs spec'd at .00x flatness per foot. some of them quite long, or quite thin. some of these jobs have had the words "in the restrained condition" on the drawing, which means parallelism not flatness is the real issue. many have not, and I have spoken with folk who did the engineering and been told, yes, in the restrained condition is fine. taking the time to make the call has gained me the opportunity to make many of these parts. if I had not made those calls id have had to no quote. and yes, I have had to make calls to get the customer to have his qc dept talk to engineering.

Question about flatness tolerance

Titanium

Diamond; Mod Squad

Stainless

Diamond

Diamond

Diamond

Titanium

Diamond

Diamond; Mod Squad

Diamond

Diamond; Mod Squad

Diamond

Diamond

Diamond

Diamond

Titanium

Diamond

Diamond

Diamond

Cast Iron

Similar threads