First off, I should say what is left of my Sherline is the spindle and motor assembly. The rest is the full A2Z conversion.. along with the Kerk precision leadscrews, and anti-backlash nuts.

My current issue is backlash. I mount a dial indicator in the spindle, and set it against a flat object bolted to the mill. I then proceed to zero the indicator face, and jog the opposite way .0001 at a time. My indicator reads the X axis error to be .0025", and the Y axis .0035". Is this an acceptable value for the anti-backlash nuts/leadscrew from Kerk? I was expecting maybe less than half of those values.

The leadscrews are bolted down well, and I get zero runout while an indicator is measuring the backside of the leadscrew while moving it CW, and CCW.

These are recent upgrades to the mill, so I'm not sure what to expect. The mill is clean, lubed with way oil, and I can move the tables smoothly by hand if I remove the motors, so binding isn't an issue.

Any thoughts? The backlash seems to be pretty constant, and doesn't vary from the previous measurements.. just leave it as is, and allow the software to compensate for the error?

I just have the regular screws on a Sherline Mill. I tightened the nuts to about .002-.004" endplay no problems and with EMC (cnc) compensate for backlash with nut mid length of leadscrew down to 0.001" no problem

Trouble is the screws are small and stretch and compress a little so as nut is at far end of leadscrew it has more endplay and when nut is near end it has less endplay.

Even if ballscrew and nut had zero backlash the ballscrew itself stretches and compresses. On a CNC with closed loop feedback, it does not count turns of the screw but movement on a linear glass scale showing actual movement.

On an old Prototrak mill with Trav-A-Dial encoders there is a steel wheel with fine teeth in contact with the side of the mill table. If a chip get stuck on encoder wheel computer says table not moving and gives it full power til it does move. I have seen the table jump an inch when closed loop feedback not working. With open loop cnc you are counting turns of a screw. Even a ballscrew with no endplay will not tell you if 10 turns is giving 1" exactly everywhere on the leadscrew. Old machines often have leadscrew pitch variations.

Sorry I am not offering any solutions.

I have upgraded my sherline to the A2Z lead screws as well, I have .001 which is a far better then the .003 I was getting with the original screws, I did notice that the gibs have to be just right in order to get that backlash, to tight is no good, I would suggest loosening the gibs so they slop, check backlash, hopefully you get my numbers, then tighten the gibs so they just take out the slop and no more, it's a very small window of play, if you don't get my numbers when the gibs slop, it might be the springs on the nuts don't have enough tension. Hope this helps, but rest assured you will eventually surpass the backlash numbers that the original screws had.

I just upgraded X and Y on my round column mill using 20 micron nuts (20 microns = about 0.8 thou) Got less than 10 microns backlash on the static test. Well pleased.

To eliminate slop on the bearings I used pairs of axial races and preloaded them to about a quarter ton using Belleville washers. The X screw is held in tension, the Y is pinned at one end.

Cut first metal today on my new screws, came out around +55 microns in the Y and +35 microns in the X. Think it's just tool spring because it is very consistant, probably need to repeat the finishing cut.

Could your backlash be in the bearings rather than the nuts?