I had previously calibrated my machine’s belt-stretch using a dial indicator, but with 1" of travel (of the indicator) I wasn’t testing as much belt length as I would have liked.
So today I strapped a 12" digital caliper to my Shapeoko 3 and did some more testing. So now instead of checking 1" of belt, I’m checking up to approx. 10" (I lose a little due to the clamping consuming some of the usable length of the caliper.
So I was computing new values for $100 and $101, plugging those into the machine with the $100= and $101 commands, and retesting, rinse, repeat.
But then I wondered, is there a maximum # of times I should re-write $100 and $101? I know some sorts of non-volatile memory has a maximum # of write cycles, I don’t want to exceed that prematurely with my futzing around.
I gave up on belt calibration and here’s why. I checked my machine inch by inch in a 13" grid (Left, Center, and Right) on the Y (both plus and minus direction) and a 13" grid (Front, Middle and Back) on the X (both plus and minus direction) and what I found was that the belt and pulley timing isn’t that accurate. Sometimes it is short, then too long, then right on throughout the X and Y grid.
Meaning I can get it “spot on” at a certain distance (e.g. 10", but it will NOT be spot on at any OTHER distance (because the belt and pulley aren’t perfect).
So what I do, is if I need something to be spot on, I adjust my program to cut it spot on (by first making it small (or large) and then adjusting the program. Belts and pulleys have their limitations.
My Data: (I have the detail that was used to produce this data)
Thanks Rich, I had been told that the belts don’t always stretch evenly so I did want to go back and perform some more testing. I figured worst-case scenario, I could check the envelope I’m going to work within and get it close enough for the type of work I do (nothing as super-fine as the stuff you’re doing).
I had a chance to go back and test my new settings (that were established over approx. 9"), at 1" intervals. What I found was that I was able to keep my over/under travels to approx. -.0017" or better, often computing variances in the tenths and I don’t trust my calipers to that degree of precision.
Here are some results:
Machine = 25.992-mm, the caliper = 25.95-mm, for an under-travel of .0017" (I think in inches).
Machine = 50.01, caliper = 49.97, or .0016".
Machine = 75.002, caliper = 74.97, or .0013".
Machine = 99.995, caliper = 99.96, or .0014".
Machine = 125.012, caliper = 124.97, or .0017".
Machine = 150.005, caliper = 149.98, or .000984".
Machine = 174.998, caliper = 174.99, or .000315".
Machine = 199.99, caliper = 199.98, or .0004".
Machine = 225.008, caliper = 224.99, or .000709".
I was able to duplicate these results by repeating the test a couple of times. But now I’ll do some more testing by shifting my caliper to the right and rear of the machine and see what I get.
This is the X-only, need to test the Y next.
I don’t think I can adjust $100 any more, as my under-travels are pretty consistent. Adding more to $100 would likely toss me out of whack, the further I travel.
The last time I performed a calibration, I was using CM and a dial indicator with 1" of travel, and I think I was using inches instead of metric. Now I’m using UGCS and I’m comparing the machine’s coordinates with the caliper in metric. An interesting thing is that when zeroing the machine, UGCS often reports X or Y as .001 or .002-mm, often not just 0-mm. So that may be helping me achieve a little greater precision, knowing exactly where the machine believes it “is,” vs. the way I was attempting this with CM. It has been a long time, though, so I can’t remember all the details.
This degree of precision will be perfectly sufficient for the sorts of work I do, though.