I’ve cut a simple 190mm square with 145mm diameter circle in the middle. The square is perfect 190mm but the circle is 142.7mm. This is using both Fusion360 and Easel. Cutting is with a 6mm End Mill.
This has also happened with 20mm holes, 17.8mm is cut.
The belts have been tested.
The machine has been calibrated for $100,$101,$102
VWheels are firm and roll well.
The mill bit is the 6mm diameter.
The frame is square, tight and ridged.
Can anybody please help?
Cheers
John
You’ll probably find that the circle is also an oval. I find I get better results with slower feed rate and smaller doc. Deflection of the bit whilst going around the curve seems to be the culprit.
If that is the case then the deviation is due to different flex Y vs X axis.
If the outside is spot on and the inside pocket is undersized - and assuming calibration is ok, is odd.
What deviation do you have if you test carve a 20 / 40 / 60 / 80 circle?
Thanks, but it is a perfect circle, inside the perfect square with the square being the correct size and the circle being evenly in error.
Cheers. I’ll try the range of circles with both internal and external cuts.
Also see if it gets closer to desired size with slower feed rate.
Provided the calibration is correct, any backlash/system deflection should show up with oversized outside cuts and undersized inside cuts. It is reported that only the inside is undersized which puzzles me 
But wouldn’t deflection only effect the curved cuts? Only the inside cut was curved, the outside cut was straight.
Good point, worth considering 
Thanks people. Some more info. And I’ll run some more tests.
Info:
The circle is a circle with no deflections, just the wrong size.
The diagonals of the square are matching.
When I cut from Fusion360 the circle and the square are done with the same pass.
I’ll test:
Slower feed rate
A test cut with internal and external circle and square.
Double check the drive system, the bit (with micrometer), and the settings once more. Then triple check.
4 Tests later! All good
Test out put using Easel. The 150mm line is done with and attached pencil. Now I’m really confused. There is no play in the wheels or frame.
Just to be clear, there is NO machine that do not have some sort of play 
The high dollar ones there is barely any, but always some. Thats why there is compensation settings present in such systems.
The reason I suggested circles with different diameter was to see if the deviation was the same throughout or increased with larger circles. If they were consistently cut say 1mm short on all, then you have a 0.5mm system play going on.
That would be a combination of rigidity (lack of), tool deflection and back lash.
This image suggest that you do have some play, circled in red is the entry/retraction point of your router, when it plunge/lift there is “no” sideway forces given, but when the router start to cut X/Y the bit will want to drag behind.
BTW - if your end mill is actually turning dark like the photo suggest => it is heating up => going to slow / rpm to high.
