Y axis problem

If you are using a dti to measure travel, you should be able to measure backlash, if any, and eliminate that as a cause. Set up the dti so that it is at mid-travel, then move the axis a couple of mm in one direction and return to starting position. Note dti. Then do the same in the opposite direction. Difference in dti readings is a reasonable estimate of backlash. Could be as simple as badly-adjusted ballscrew angular contact bearings.

If the error were rather smaller (of the order of 0.005mm) but increased by the same amount each time you repeated the movement, I would suspect a step pulse polarity problem. I'm a bit sensitive to this one as it bit me in the bum on my machine, and it's a bit subtle to detect as it doesn't show up very easily except after many repeated direction changes. However, your error is many times larger - which is why I asked exactly how you were doing your test. Over what distance, at what speed, does the error show up after one movement or after many repeats, etc. If you can fill in the details, it might let someone spot a likely problem and/or eliminate red herrings.

Do you have backlash compensation turned on and if so is it set to the correct value?

The llead screw is a ball screw,
And I haven't got backlash elimination on. As for the test it's a square 30 x 30mm. Set machine to the centre 0,0 0.2 deep, cycle start, matches ok
Cycle start = error of 0.3
Cycle start = error of 0.6
And so on
Hope that makes sense

Can anyone shed any light on my problem. I know it's not the program because the square shape machines perfectly on my machine at work

An accumulative error, sounds like fun. Off the top of my head I have 3 guesses... excessive acceleration/deceleration causing a motor to stall/overrun, excessive acceleration/decelleration causing a loose mechanical connection to slip on a shaft, excessive microstepping so you don't reach the minimum up/down times on the driver step pin.

Thanks Robin, I see where your coming from, but the feed is only 200 mm/m and the rapid to position is 1500 mm/m. Could it be a driver fault ?

You need to mark the motor shaft, do the cut, return to the start position, see if the mark has moved. That will tell you whether the fault is before or after the motor.

Hi Robin as you suggested mark the motor shaft do a cut return to start position. the motor shaft is about 1/8th of a turn out of position when cutting a circle 40mm dia

Quote:

---

Originally Posted by Kev2960

Hi Robin as you suggested mark the motor shaft do a cut return to start position. the motor shaft is about 1/8th of a turn out of position when cutting a circle 40mm dia

---

There you go then loctite the pulley or coupling on then

Edit. My bad I thought he meant shaft had moved in the coupling

If the motor shaft is out of position at the end, doesn't this sound more like an electrical problem?

1/8 turn is about 0.6mm error - is that about right?

I think you said 320 steps per mm, which would be x8 microstepping with a 5mm lead ballscrew - sounds about right. So, 1600 steps per rev. 1/8 turn is 200 steps - that's a lot! Does this happen with air cutting? If so, can you slow the max speed and acceleration right down to remove the chance of it being missed steps? Is the error always exactly the same each time you test it, or does it vary quite a lot? Does it happen with the spindle running and spindle stopped?

Out of curiosity, where does the gcode come from? As I mentioned earlier, I had a similar problem when I first started setting up my own router, although in my case the error was much smaller (although still cumulative) and I'm trying to think through a logical sequence to try to narrow down the problem.