Hey Dave,
As you get down in the tolerances  the whole system comes into play. If It's big variations then maybe there is  one culprit to be found causing a problem. But  when you start cutting hairs sometimes it's difficult to refine the machine
 / system and frankly may be wasting your time.
Here is an example:
My old mill, which I CNC'd, will produce an egged hole when milling a circle to hold a bearing. The gain in the lead screw is 0.00025" / inch and that gain varies over a 12", but there a few sweet spots in it's travel, the ball screw is within spec. I can use the sweet spot, even adjust the steps, and do an extra step or two and produce a very close tolerenced hole which is not egg shaped if i want to take the time to do so.

My engraving machine is different in that the X & Y have phenomenal accuracy and the Z movement is great also.
So no need to even think about tweaks, but, the tool runout can become limiting. 
So there's were i come from in my response and logical to me ( FWTW ). 
RICH
 
 

Hi Rich,

It is true that as the tolerances get tighter more things come into play. But do you really think that 0.05mm can be referred as a tight tolerance in the above sense? I would think that it is 0.01mm and finer tolerances when things get tricky.

What does your gcode look like when you make those adjustments to achieve an accurate circle? I mean, if it's oval then tweaking the radius in the gcode won't do. I would think of using scale factors instead....?

What's your mill's backlash? And what kind of tolerance for the circles can you get?

Daniel

Daniel,
What you have is what you need to work with and do the best you can with it.
Comparing diffrent machines is useless.
No code changes are done.
RICH
 

IS the error the same no matter what the size of the rectangle is or does it grow the larger the size?

(;-) TP

Rich,

Wasn't trying to compare machines Wanted to get an idea of what kind of tolerance you can get on your machine with known backlash and gain if you work appropriately, and then work out from there of what can I get if I learn to work correctly with my machine.

So what do you mean by "adjusting steps" and "adding steps"?

Daniel

Hi Terry,

Yes, the error seems to be the same no matter what's the size of the rectangle. Were you thinking it was wrong steps per in motor tuning?

Daniel

Sorry I mentioned the "adjusting steps " since if I get into this kind of discussion folks will think it is good practice not realizing that something was done for a very specific understood reason and it's not normal to do so.
 
You calculate and use the steps per unit and check / reafirm as noted in the manual and NEVER change them.

There are books about maching techniques and suggest you review that information.
BTW, a few swipes with a piece of sand paper can go a long way in adjusting something to dimension. 
RICH   

HI daniel, Yes that came to mind(;-).  You need to first do some dedicated testing to confirm the steps/per is correct.

SEEMS to does not count(;-). Test and confirm.   You start with the straight line test and then move to  square/round  cuts and confirm the diamensions..  

RUle is do the math for the steps value IF possible.  THIS should be the correct value. BUT you may find variances in the screw itself so you test across the range and confirm.

IF the math cannot be done (unknown values) then test and tune is appropriate. DO the mach calibration test and TEST TEST TEST until you have a repeatable value to work with. IT must repeat itself in ALL conditions to a very close +/- toerance. The key is repeat itself(;-).

With all that out of the way you are down to machine design / wear tolerances. SOme machines just cannot get tight tolerances do to design and or wear. Cutting flex CAN effect you as well. Sharp bit VS dull bit, there is a host of things to effect accuracy.

SO start at the top and work you r way down throught the list untill you find the culprit. THERE IS ALWYAS A REASON(;-)  Then repair it IF possible and then live with the rest as best you can.

Just a thought, (;-) TP