Check to make sure your x rail is squared to your machine. Bring the rail to within a inch or two of the very front and use calipers to make sure both sides of the x rail are the same distance from the Y plates.
There’s a few ways to do this. I don’t recall any videos.
I use some spare 20x20 extrusion that are the same lengths, put them between each of the Y plates and the axis with the machine unpowered and slowly pull the axis tight again the extrusion and then turn on and lock the steppers. I do this for every power on.
If you find it out of alignment a lot, hen your belt tensions on each Y might be uneven.
So far I have calibrated my stepper motors and squared my x axis. I am still waiting for my scale so I can test the belt tension.
I am still getting undersized and lumpy circles. Interestingly, whether I make a peg or a hole, they are both undersized. You would think that since they are routing on opposite sides, that one would be undersized and the other would be oversized.
Uneven stiffness between X/Y axis so the amount of system deflection vary during the transition from X/Y and Y/X. Provided the actual bit diameter is true and there is no excessive run-out.
May I suggest the following:
Do a 1" and 4" circle (path outside) with your current feed rates etc.
Do the same 1" and 4" circle (path outside) but after running the carve once, run it again at the same place, with feed rate at 25% of previous speed and in one, full depth, carve.
If your deviation is the same amount between the 1" and 4" your calibration is probably pretty decent, but tool deflection / transitional play is your culprit.
If your deviation is 4x larger for the 4" version vs the 1" version you may need to recalibrate.
(If a 20" jog equal precisely 20" travel then calibration is okay, defect is due to actual tool diameter/runout/tool deflection.
FINALLY I saw improvement with loosening the v wheels. Rotatable with one finger is too loose, and rotatable with two fingers is just right (although difficult to do on the x carriage). Why is this not recommended in the initial calibration instructions?
My results got rid of the lump at the 2 o’clock position on the bottom circle; although the 12:00 position and the 3:00 are the values that changed. The top circle is the new repeatable cut. Is this the best I can expect for roundness in the end? On smaller circles about .375 it more of an issue than with larger ones.