Table flatness & general precision

[cnc_works]

Hope this doesn't turn out too long for you guys.

I have had some years of experience with a really tight, precise, smaller cnc router, generally .001 repeatability or real close to it.

I bought a used PRT96 with a super heavy table earlier this year and have finally got it together with a 1/2" X 48" X 96" sheet of aluminum shimmed pretty flat, then two sheets of 3/4" MDF with the top sheet machined flat, then immediately coated with a couple coats of shellac to slow moisture in & out.

Tonight I put a chunk of aluminum (a vacuum plate from a process camera) 1 1/4" X 30" X 40" on the MDF and machined the top flat to use it for a specialty vacuum plate for myself. Pattern showed pretty good perpendicular on the Z, but when I finished and put a machinists 4' straight edge on it, I had about .004" to .005" hollow in the middle of the plate. Can't for the life of me figure how, but there it was.

My question is, would you be happy with that tolerance? I wouldn't be so happy with that tolerance on my small CNC, but is this what I should expect for the bot? I mean, I wouldn't be too disappointed if this was IT, but I'd like a little closer. And any hints on how to get closer would be appreciated, given that I spent a lot of somewhat intelligent effort getting to this point.

Last point, like many others, this forum is what gave me the confidence I could probably get what I wanted out of the bot. Thanks!

Donn
www.cncworks.biz (http://www.cncworks.biz)

[gerald_d]

Donn, just something to consider, and I am not saying this is what happened to you....

If your gantry (y-rails) were sagging hollow in the middle, maybe due to the weight of the car+router, your table will also be hollow after using that car+router to machine it "flat". Your 1.25" thick alu is too stiff to flatten onto a hollow table - even with vacuum, the the table will flex to meet your stiff plate.

Also, your alu table bolted to (presumably) steel bearers will be temperature sensitive. When hot, it will bow up / when cold, it will sag down. Alu expands and contracts a lot more than steel with temperature differences, and you effectively have a bi-metallic table. However, with a 1.25" thick slab of a workpiece, the table flatness is not an issue any more - your issue is probably rail straightness and stiffness.

Having said all that, I think that you are doing pretty well to only have a .005" hollow over 40"!!

[kerrazy]

Silly question...But?
Did you seal both sheets of MDF, or just the surface sheet. I realize they are ussually pretty good from the mill, but it is something to consider, and with your tolerances, could you have a build up of shellac in the centre of the table?

food for thought
Dale

[cnc_works]

Gerald, my machinists straight edge says there is no sag in the rails with the car off to the side and I haven't figured out how to measure flatness with the car in the middle. Two things about this: One, the plate was in the 0,0 quandrant of the table, oriented with the long side of the rectangle the 96" way. So, the low spot was in the middle of the plate, but the middle of the traverse of the car where the weight would make it sag the worst was at the edge of the plate. Secondly, if that is the issue, what in the world do the guys with a Columbo and a router do? I'm only running one of the new lightweight PC890's so do they have to do some extra bracing or shimming to make up for the extra weight on the rails? Pertinent to me because I want to install as a second Z another PC890 router.

Did I create a problem for myself with the 1/2" aluminum plate for a bottom table support? Our inside temps only vary from about 65deg to 85deg winter to summer with pretty gradual changes. Would I be better off with nothing more than MDF or MDF & plywood?

Dale, as they say, no such thing as a silly question. I'm in the computer business and one I often ask (and occasionally get a negative reply to) is, "is it plugged in?"

I sprayed two coats of shellac on each side and the edges of both sheets before I put them on the aluminum base, screwed them into pre-threaded holes in the aluminum with nylon bolts countersunk into the MDF, then surfaced the top sheet and immediately sprayed two more coats on it. I've had a little practice spraying, so I kinda doubt that much of a build-up difference.

I've probably got this table set up decently, but I sure would like to chase down a few more thousandths.

Thanks for the responses.
Still mentally malnourished.

Donn
www.cncworks.biz (http://www.cncworks.biz)

[bjenkins]

Hi again, Donn,

I have some personal history learning the lesson Gerald points out. Dissimilar metal expansion rates can have dramatic effects over your plate dimensions. In my experience, even lighting the top of the aluminum plate can cause some temperature delta that causes the top of the plate to expand versus the bottom. I know it is hard to believe, but it adds up quickly.

Also, as mentioned, try checking the straightness of your rails. If you can see an orientation to the bow (x versus y) that could be a clue as to which rails may have contributed. The way you load the hold-down bearings in Y can also be very reactive to bow in the Y rails. I have found that even after shimming my X rails to be very flat, there is a slight bit of deflection locally as the y-car moves between the support points. I've considered adding shims at intermediate points, but so far I'm getting all the accuracy I need in the stuff I've done.

One last thought, the sequence of your cut could have resulted in some heating of the plate and expansion. Think about the orientation of the bow and the cutting direction and sequence.

[gerald_d]

Donn, mount a dial gauge under the y-rail (stand the gauge on the table and let the "feeler point" press under the gantry. It is quite interesting to see how it responds when you push/pull the router nose simulating cutting forces, or push up/down on the car simulating weight and/or helical cutter forces.

Was your plate (after cutting) equally thick right around all the edges, or was it wedge-shaped, bowl-shaped, saddle-shaped, etc?

[bjenkins]

By the time I made the last post you had a new one out there, Donn. You answered some of my questions...

Just to calibrate you on the impact of thermal expansion, aluminum alloys expand in the 11-13 millionths of an inch per degree F. So in 30 inches you can see .0004" with one degree and .004" with ten degrees. Of course, the differential expansion is the issue Gerald brings up and that is about 4 millionths of an inch per degree F. That gets complicated because you have to analyze the structure to determine how differential expansion will translate into surface deflections.

[gerald_d]

For seeing temp. effects, leave the dial gauge under the y-rail during different temps.

[spektr]

Before you freak out, it is probably(most likely) the material itself. Platestock always moves when skim cuts are made. Thats why they are usually fly cut twice, once on each side. Scott

[cnc_works]

Bud, it seems to me that the only thing I can do about the aluminum is to be aware of it and thanks for pointing that out.

Gerald, I did put the dial indicator under the middle of the Z car, then moved the Y car from edge to middle. It sagged a little over .004", which would lead me believe it was part of the hollow plate problem. But, given the plate was in the 0,0 quadrant, logic would tell me that it should have had the low spot more toward the edge than right in the middle. Despite this, the interesting part of that exercise was learning about the sag in the X car from the weight of the Y car. It increases my concerns about adding the second Z. Makes me wonder if I should be looking at something stiffer than the light box beam to support the X car rails when I do add the Z.

The other interesting thing that I noticed as I flattened the plate was that rastering in the X direction produced a cut that was not nearly as smooth as rastering in the Y direction. Acceptable with X raster, but the ridges along the edge of the cut and irregularity of the cut in general were much more pronounced.

Scott, I sure did surface the back before I surfaced the front.

I guess I'm just going to be pragmatic and use it for awhile and see how the tolerances play out in the sign engraving and parts production that I want to do. I suspect that for most things, the tolerances I've achieved will work just fine.

Lastly, has anyone experimented much with the ramping numbers in the new software. If I use the defaults, the machine sure seems a little jerky in engraving 3d centerline stuff, especially on smaller letters. I have to think that a set of ramping numbers for engraving would vary pretty dramatically from someone cutting cabinet sides.

Regards,
Donn
www.cncworks.biz (http://www.cncworks.biz)

[ron brown]

"Lastly, has anyone experimented much with the ramping numbers in the new software. If I use the defaults, the machine sure seems a little jerky in engraving 3d centerline stuff, especially on smaller letters. I have to think that a set of ramping numbers for engraving would vary pretty dramatically from someone cutting cabinet sides."

One might want to experiemnt with the "Circle Segment Resolution" in the VU command and "Movement Blocks" in the SC command... I think one can write these commands in the parts file to keep from clogging up your memory on larger, faster moving files. Don't forget to reset the file as you exit the parts program... I have known folks who have forgotten to reset parameters and ruin the next "blank". I think it was - and probably will be in the future, a mistake I have made.

Ron

[K.Simmerer (Unregistered Guest)]

Ok, lets get our dial indicator back out and mount it to the z axis. Plunge to the same depth in a grid across the plate and record the readings. If the readings correspond pretty much with your straight edge findings then you are dealing with a deflection problem. If they check out reasonably consistant then you are machining a matching sag or bow in the rails, a ride up problem with the rollers or some other setup related problem.

Keith

[signman8]

I have nothing to ad to this hitech conversation. All I need is a little help to remember how I flattened the first piece of MDF! I am obviously doing something wrong, it won't go past the first cut. What did I forget? The manual is very vague and I remembered I had to call ShopBot the first time to get it fixed. I know you are saying at this point, Why didn't you write it down? Well, I must have had a brain sputter.
Thanks for your help. Whoever it might be.

[Dan B. (Unregistered Guest)]

Did you set the pocket function? Outside to inside.