I’m quite surprised that the values would change that much in a few hours (and that they would be below 39 steps/mm after the first calibration already) because I hardly ever tighten my belts, and keep getting relatively precise dimensions.
Just to check, when you say you checked at 125mm after the calibration and it was good, did you measure that using the same caliper process, and without having moved the caliper since the calibration measurements?
If so, a better check could be to move the caliper elsewhere, and recheck. A possibility I can imagine is if the axis of your caliper during the calibration was not 100% aligned with the natural axis of the machine, then you would measure a slightly longer or slightly shorter distance than what the machine has actually travelled. And you would still read a perfect 125mm after calibration, because…the calibration factor would have included that skew. From experience, it’s not so easy to attach the caliper such that it is perfectly aligned with the machine axis, so who knows…this is just a guess.
You could also try to measure the calibration factor using a different method, and see if you get the same values? For example, cut three pockets far apart but aligned along X and Y, and measure (with a ruler) the actual distance edge to edge:
Yes, I attached a pic of a setup I use based on one of our members setup using a dial caliper.
The caliper is aligned using the holes on the waste board. The holes were verified to be running parallel to the X axis and Y axis.
The caliper did not change and is at a fix position to avoid the uncertainties of cosign errors.
I use the touch probe to show me when the inverted router bit touches the caliper jaws. Also the Z axis is dropped down to the same depth using the rapid position plus 6mm of manually moving the Z down.
Mmh. I have no clue then, these values and their variation seem weird to me, but I don’t have an XXL either so maybe other XXL owners who have calibrated their belt stretch could comment whether they have such low (38-ish) $100 params…
Did you have a fresh battery in the caliper ?
Unless someone has a better idea, I would advise to do the three-pockets method, see what value comes out, in the (unlikely) event that something in the caliper setup is wrong
So 125 mm at default value of 40 mm resulted in a measured distance of 131.26 mm.
With a new calibrated value of 38.09234 mm resulted in a measured distance of 125.45 mm.
As in my previous calibration, not being satisfied with 125.45 mm, I tweaked the coefficient value by trial and error till I got 125 mm. Only this time I will not till I can find the root cause of why the default values have a deviation of 5.81 mm.
Found the culprit … that explains my entire issues on this thread.
I readjusted the eccentric nuts having found them to have loosen during the break in period a while back.
I checked the friction on the right rail and found one of the v-wheels a bit too tight. Re-adjusted and the left side rail NW to SW is tracking.
Back to my calibration verification … Whew … so much moving parts … reminds me of my RC Helicopter … a hundred parts all working together to make the heli fly. In this case, make the Shapeoko put out accurate projects.
These numbers sound like you may be measuring to opposite sides of the bit; a 6 mm bit included in a 125 mm measurement would be about a 5% error which would agree with your initial 37.9 steps/mm factor.
DAH, it was a tight v-groove wheel on the right Y rail and also using a dial caliper at 5 inches to perform the belt tension calibration created an error over the overall length.