Hi guys,

I have a Taig DSLS 3000 that definitely needs a little backlash adjustment in the X and Y axis and today I thought I'd try to implement the backlash compensation as I was in the middle of job that was given me a bit of uneven cuts.

Well, it works great for large slow circular moves but when I get into quick moving operations like engraving letters that are about 3/16ths high and cutting at 4IPMs, it turns on the RESET and says I've hit a limit switch.

That obviously didn't really happen so I'm wondering since the DSLS controller has a feedback loop that is supposed to tell Mach when it loses steps, if this is actually mistakenly thinking the backlash is missed steps and throwing the RESET? The second I turn backlash compensation off, it runs smooth as ever with no issues.

I'm going to try this again tomorrow and run super slow to see if if maybe the controller encoders can "keep up" with the backlash.

I'd love to hear any thoughts you may have-

Thanks as always,
Dave

any chance that the quick moves are shaking your limits switches and temporarily breaking their contact? Try setting a highish debounce if you think that may be a possibility to check the theory out.
Hood

Nice thought, but no, that wouldn't be it as it's not that "fierce" of a move.

It's not like my machine is rattling about, but you have me thinking maybe I could just disable them for a moment, just to be sure.

Thanks Hood-

Dave

Not sure what the DSLS is but would have thought that if its some sort of servo drive then Mach will be sending step/Dir signals when doing the backlash comp so it shouldnt see that as lost steps.
Hood

That's pretty much what I was thinking but this is not a true servo setup as they'd like you to believe.

It's a stepper setup that just has an encoder piggybacked on each stepper and tells Mach where it is and if it gets too far out of range (not sure what that range is still to this point), then it tells Mach to go into RESET.

It does not "catch up" to the code after getting lost as I'm guessing true servo setups do, is this correct?

Dave

Servos kind of do that, mine get a command from Mach to move to a position and the drive will attempt to move it to that position, if there is resistance it will try harder to get it there but it has to do this within a certain time or it will fault. Mine is set up for 20 counts of the encoder (8000 count encoder) for 20ms, if its out by that then it will trip.

Hood

Yeah, that's a pretty slick setup  - 

Dave

Lets just study what happens, because if you have backlash on your system, you have it, whether or not you have backlash compensation turned on.

Mach seems to do three things if backlash is turned on. The pulses go to the axis (which may or may not move, depending on the amount of backlash you have. Mach then seems to realise that you have reversed direction, and stops. It then applies the amount of backlash compensation you have asked for (at the speed you have set it at) and then carries on with the move

For the first x number of pulses, the axis will not move, becasue of backlash, then it will and your encoder will start to move - but surely will always be the x number of pulses behind Mach. If compensation is NOT turned on, then Mach will stop and the encoder will stop, still behind - but seeing Mach has finished the move, then obviously does not register it. If compensation is turned ON, both will stop (in theory together)  - but this time registers a fault.

I think the only explanation is that you have a little too much backlash compensation on - and so when backlash is applied the encoder gets in front of Mach 3. This clearly does not matter when moving, the encoder is happy to be in advance, but if Mach stops the move and the encoder is still in front - then it must register a fault.
I can't see any other explanation.

Interesting, Jim, I will play around a little more today and lower the value as compared to what I actually have (only .002-.003). The problem with lowering the value is the engraved letters may still look odd which is the reason I tried it in the first place.

Bottom line is I'll have to remove the tables and adjust this a bit.

Thanks for your input-

Dave

A less elegant solution would be to turn off b/l comp and compensate in the code. Of course this is also a far more complicated solution.
Joe