Upgrade? Chinese CNC Machine

[Jonathan]

I dislike just scaling the step/mm values to make it work as it can mask more fundamental issues, and generally isn't as accurate. You should be able to measure the ballscrew pitch and work it out. The comment earlier saying the 'the Y however seemed to jerk' implies it could be stalling as the motion should be very smooth. If that is the case you need to find what's causing that, else the distance readings for that axis are meaningless.

Instead of just measuring a small distance along the screw, measure say 20 turns and divide the result by 20. Assuming you can access the motors, you shouldn't have to dismantle anything. Just measure the distance between the axis and a convenient point, turn it 'n' times by hand and measure the distance again. Divide the difference between those distances by 'n' and you have the pitch. If you've got a digital caliper then measure it over about 150mm to get the most accurate reading.

Doing a test cut is not the best way either, since spindle runout and tool deflection can introduce small errors and it's not possible to measure in wood that accurately anyway. Instead measure how far the axis moves. Measuring the distance between two parallel lines would be easier, and a bit more accurate, than measuring the length of the line, since you're measuring between straight edges not arcs.

Assuming the '3 turns per cm', i.e. 3.3mm pitch is reasonably accurate, the step/mm values we've ended up with don't make much sense. The motors are run from TB6560 ICs, which only support microstepping settings which are powers of 2 up to 2^4 (16). Assuming (as there's a slight chance it may not be) the motors full step resolution is 200 step/rev, that means your motors can be set to 200*2^n where 0<n<4. We need to find steps per mm, so divide by the pitch-> 200*2^n/3.3 = 60.6*2^n, where 0<n<4. That means the correct step/mm value could be one of the following (rounded to 2DP); 60.61, 121.21, 242.42, 484.85, 969.70. None of those seem anywhere near the 408.1 and 512.8 you tried... so I reckon there's something else causing problems here.

You could measure the voltages at the M1 and M2 pins of the TB6560 ICs on your driver board to determine what the microstep setting actually is. The datasheet here shows which pins to probe.