Inaccurate circles/squares ...thoughts ?

Do you know what else there is to adjust to do a better job?

As pointed out many factors effect the outcome of a cut. Backlash, rigidity, flex etc. Try taking light cuts, pay attention to spindle speed and feed rates.
Also conventional vs climb cuts as the stress on the machine is different for both.

Backlash comp can help improve your cuts in some cases.
If the machine is pushing the tool into the work backlash comp may help you get the correct length of cut, however if as in climb milling the tool is pulling itself into the cut it is likely to overshoot.

You can also mount a Sharpie in the spindle and with the spindle turned off try running your pattern on a piece of paper and see if the results vary from what you get when actually milling.

Rick G

The leadscrews on the sherline are precision versions of allthread.
Not the greatest, but not much worse than chinese rolled ballscrews.
The backlash is adjustable on sherlines, they recommend 0.003" - 0.005" but depending on tolerances, you could get less (the backlash is adjustable in steps, not infinitely adjustable)
make sure the gibs are acceptably tight also.

I have machined aluminum no problem on a sherline, but it wasn't the super flexy version pictured above.
It looked like this

I have never tried backlash compensation, and I never plan on it.
Just remove as much as possible.

Thank you for the info. I want to note that I mill aluminum on my sherline all day long, and everything I make comes out looking beautifully. The only time I ever notice anything is when I want to make, specifically a circle, or rectangle using a small bit. Which, as we know annunciates the backlash. I'm going to try adjusting my backlash nuts again (of course I already did that once). And then you say make sure your gibs are tight . What does that mean, because I thought you would only lock the gibs when you don't want that axis to move ?

Thank you again.

Well on a manual machine, you would lock the gibs on the axes that are not moving.
On a cnc, they could all potentially be moving, so no need to lock them.
You do want to make sure they are adjusted so everything is tight and smooth through the range of travel.
If you can grab the table and rotate or tilt it then that is potential deflection causing your problems.

Ok, yeah that makes sense. I don't feel any movement I don't have the mill affixed to anything though, which I was beginning to think is very important.

On the sherline mill (the 5xxx version not the 2xxxx "flexy" version) I noticed that I could not seemingly spot drill, then drill thru without seeing the drill noticeably bend. It looked like the y position had moved. What it actually turned out to be, was that the Z axis gib was too loose, letting the spindle deflect towards me when spot drilling, and hence the spot was not in the same place as when the drill came down.

Moral - get the gibs tight, but not so tight the steppers stall when cutting!

As for backlash, I also had issues with things out of round - in my case radius on the end of levers. I fixed it to a satisfactory level of accuracy for what I was doing as follows.

Minimize the mechanical backlash using the mechanical adjustments. Measure using a dial gauge what it is. (at several locations if you care enough)
Use the electronic backlash compensation in LinuxCNC. That's pretty well the best you can do for a low cost hobby machine like the sherline.

You can do useful stuff with the sherline in Aly, Brass, Steel and even Ti, provided you are not in too much of a hurry.
See below a handwheel for my 45mm loco] made using a 0.8mm ball mill. This has had only a cursory de-burr. Sorry about crappy soldering!


Also a miniature aircraft flap track done as a trophy for a work collegue. It's 1:20th full size and sitting on a similar "real" one. Now the real one takes some serious effort to machine!

Edit: Attached my config for comparison (note it is an inch machine!)

Thank you so much for all the info. I will try tightening things up and see how it goes.

Thanks.