I was recently making a part and had a problem that when traversing between two specific G81 cycle hole locations, the X axis would completely loose sync with the step pulses.  It would only occur on these two specific locations.  I simplified my code to G0 moves with the same coordinates during trouble shooting.  The following code will demonstrate the issue:

G00 G49 G40 G17 G80 G50 G90
G20
G0 x1.1983 y1.4473
G0 x1.9043 y1.4479
M5 M9
M30

Although I have my max speeds fairly conservative, I cut them in half and still had the problem.  I created a test loop of G81 cycles all over the table and ran a loop 50 times with no issue.  Through much troubleshooting, I have determined that the problem occurs when there is a very small Y movement required and backlash compensation is turned on.  If I increase the Y movement to .002, it does not happen.  I swapped cables, changed which pin drives which axis, and many other scenarios to isolate the issue.  I thought maybe it was mechanical at that location, so moved to a different location and zeroed the axis.  It still occurs.  If I turn off backlash, it does not occur.

So after much troubleshooting, the basic problem is that during a traverse, if the amount of Y movement is greater than 0 and less than .002 or so, and backlash compensation is turned on, most of the time, the X axis will loose sync with the step pulses.  If I do the same scenario to the X axis, it does not cause the Y to loose sync.

Can anybody shed any light on what might be going on?
If you run the above code, do you get the same result?

You do not explain what you mean by
'the X axis would completely loose sync with the step pulses'
WHAT step pulses dos the X axis lose sync with?

On a more academic level, if you have backlash compensation turned on for the Y axis to the tune of 0.002", and you try to move the Y axis by 0.001", just what do you expect to happen?

Cheers
Roger

Sorry I was unclear.  by "X axis loosing sync with the steps" I mean that it is as if you are asking it to move too fast so it stops moving and starts screaming or in other words, like you grab the hand wheel and stop its movement while it is still getting step commands.  I didn't mention in the previous post that I also dialed the acceleration down in case it was trying to start up too fast.  No help...

I actually have the Y backlash set at .0011.  If I asked it to move .002 in the opposite direction to the last move, I would expect it to rotate the screw enough to take the .0011 slack out plus the .002 movement.  What's even more interesting is that the Y does what is asked.....  but causes the X to break sync.

Odd.
Has not happened to me, but I am very deliberately not attempting to use backlash correction. I prefer to fix the mechanics instead.
You have not mentioned here what version of Mach you are using?

Cheers
Roger

I am using version 3.043.066.

I have a new Shopmaster MillTurn.  I have about .0022 backlash on X and about .0011.  I am not sure at this point what it would take to get that to zero on that price range machine.

The compensation code works fine on the X axis.  It's just that the Y axis compensation causes the querky problem with the X as described initially.

A lot of us won't use .066: it has some strange bugs. Instead most of us use and recommend .062.

My machine is not a ShopMaster, but it has some similarities.

I am not sure at this point what it would take to get that to zero on that price range machine.

IF the ball screws have single nuts and bearings, that could be a bit tricky. If they have double nuts and double ball races at the ends, then the backlash can be adjusted to well below 0.001". Getting there is mainly a matter of reading up on how to do it, then doing it. Takes a bit of time.

Cheesr
Roger

Thanks.  I'll see if I can find out what kind of nuts,... the Millturn has and also how to save parameters so I can go to that earlier version.  .066 came on the machine.

I would sure be interested if the code I posted screws up other machines or just mine.

Did you also reduce the acceleration value?

Just a thought....

RICH

Rich, yes, I tried turning the acceleration way down