Checking Tolerances on a CNC Router
When parts are coming out slightly undersized, start with a few basic troubleshooting checks. October 20, 2013
Question
(WOODWEB Member):
We have a Multi-cam router running with Alpha-cam software. The router is cutting .015 smaller on a geometry than it should be. If I have a 12"X12" square and rout with a new 3/8 bit, my result in both X and Y is .03+- smaller in dimension. The pinion gears in X have been re-seated with no effect. Since I get the same variation in both X and Y I am thinking this is not wear and tear on our 12 year old machine. I have been told that it may be a "resolution problem." Any similar problems out there? What was the fix?
Forum Responses
(CNC Forum)
From contributor D:
Silly question, but what is the diameter of the bit? If it measures 0.375, take a single straight cut and measure the width of the cut. That will at least eliminate tooling. Unlikely I know, but you'll feel really stupid if you chase this for 3 days and find out it's the router bit. Don't ask how I know...
From contributor N:
I have done that with both re-sharpened bits and new bits, checking for proper diameter and possible run out. Bits check out with correct diameter and little to no run out. Still at a loss. My poor engineer is having to compensate all parts by .015 now.
From contributor B:
Just a shot here - are you conventional cutting (counter clockwise) around the part? If so, try cutting your test square in the opposite direction (clockwise - climb cut) and see if the dimensions grow by 0.030" instead.
From contributor V:
Contributor B is correct. The first thing to check after confirming the bit diameter is deflection in the system. If you are conventional cutting and there is some Z-axis play due to wear and tear, that would account for the undersized part. If it is indeed a deflection issue, your part should come out .03" oversize on a climb cut.
From contributor D:
Also, check if the error is different on different size parts, i.e. 0.030 on a 24" part and 0.0075 on a 6" part.
From contributor M:
In addition to the above, validate the output Gcode.
From the original questioner:
In doing test cuts, I routed two 12" X 12" squares, one in each direction. The square climb cutting seemed to be just slightly larger than the normal forward cut. I have not checked the g code but will do so. Thanks for all the input on this.
