JMC servo "jerk"

[Kjelt]

No it's not mechanical. Every motor makes these jumps at the same time when they occur.

All motors share the same PSU? In that case it is the PSU.

[serum]

2 psu's are used, psu's are swapped. (motors are swapped too) I personally would have ditched the PSU's in the first place, since they where unable to provide the power/voltage requirements needed for these motors in his situation. (either due to acceleration or de-acceleration, this is unclear to me) I still question the PSU's. If memory serves right i heard another user on youtube or a forum about these 600W meanwell psu's in combination with servo's being unable to accelerate and de-accelerate fast with his servo's. Not saying the PSU's are unable to deliver the power, the voltage might increase/decrease onder de-acceleration and acceleration. Perhaps the motor itself is acting up on a voltage peak/drop the PSU delivers. Seems to me the combination of the psu and the servo's are not a perfect match. I have my 48V PSU's setup at around 43V and i never measured a voltage above the 50V, and i can set my acceleration pretty wild. I thought the JMC's have a 55V hard cutoff, but i don't know for sure. Those JMC's have a weird error trigger; you can turn them quick half a turn by hand without any reported error. (they just jump back to their correct position) I don't know if a regular multimeter gives the reading you need. I would say an oscilloscope would give a better insight.

[Kjelt]

Elvis now also has an issue with one of his JMC servos.
Sometimes it just hicksup. He first thought it was the drivebelt slipping but that is not the case.
I still have to doublecheck connections shielding etc. I still find multiple servos on a single PSU with long cables without capacitor decoupling asking for power dips and irregular behaviour because if you look at the servo drivers there are no big capacitances there or am I mistaking ?

[serum]

Is Elvis by any chance using the meanwell 600W psu? You happen to know if these are the NEMA24 200/400W versions or the 180W nema23? V6 I presume?

[Kjelt]

No he uses a 2.5kW server power supply.

[mkuivamaki]

Started to investigate this again after few months. Meanwhile I have done these things:

- every cable and groundings triple checked and upgraded as good as possible
- JMC servo tuning back to autotune mode and servo stiffness adjusted to medium / easy settings

Still having random shocks One thing still to be highlighted: I have these shocks only on Z and Y axis motors. Those axis are the only ones which have EdingCNC crosscompensation affected. Feels like I'm having these shocks only during very low speeds and when there is very tiny crosscomp values ongoing

Example from my Y-crossCompensationTable, where the X is "master" (left value) and on the right is the amount to add or reduce from Y values:

0.0000 0.0000
87.0000 0.1100
151.0000 0.1600
220.0000 0.1700
335.0000 0.1600
470.0000 0.1100
625.0000 0.0600
775.0000 0.0300
870.0000 0.0500
970.0000 0.0900


And the Z-values (X is again the master axis):

0.0000 -0.0000
200.0000 -0.0500
340.0000 -0.0000
507.0000 0.0500
770.0000 0.1000
980.0000 0.1000

[hfjbuis]

Still having random shocks One thing still to be highlighted: I have these shocks only on Z and Y axis motors. Those axis are the only ones which have EdingCNC crosscompensation affected. Feels like I'm having these shocks only during very low speeds and when there is very tiny crosscomp values ongoing

Cross compensation is from what I have found on the internet, a feature to reduce oscillations due to fast deceleration and acceleration. It is only needed during fast deceleration and acceleration of the machine. What ever it should improve, turn it off and see if things get better.

edit:
Op basis van een video over eding, denk ik dat het hier om correctie in spindel posities gaat.

[mkuivamaki]

Yes. With crosscomp on EdingCNC you can fix small mechanical errors in your machine and put your machine on square by the software. I have heard that there has been issues using this feature.

[serum]

Still having random shocks One thing still to be highlighted: I have these shocks only on Z and Y axis motors. Those axis are the only ones which have EdingCNC crosscompensation affected. Feels like I'm having these shocks only during very low speeds and when there is very tiny crosscomp values ongoing

You have tested without crosscompensation? 1 minute test (that's the first thing I would do). I wonder what method you used to determine the correction needed. It's below 0,17mm over 980mm for the Z compensation. I doubt you'll ever have a part to made that size that requires that accuracy? Just calculate how much difference your gantry will have with 10 degree celsius temperature difference with the metal lineair rails strapped onto aluminium. You need a well controlled temperature in your room to minimize that error.

[mkuivamaki]

Yes, today tested again without crosscomp. No clear improvement.

[serum]

Replaced a psu yet for another brand/type?

[mkuivamaki]

Replaced a psu yet for another brand/type?

Not yet. I doubt that PSUs would be the problem because these PSP type Mean Well power supplies are used widely with JMCs. And these problems occurs only on very low power requirements.
But anyway I think I might buy one SDR type Mean Well PSU which is also another typical PSU for the JMCs and see if there is any difference.

[mkuivamaki]

By the way, which one you prefer for the pulse signal edge evaluation: rising edge or falling edge?

[Kars-cnc]

Basically that doesn't matter, as long as there is no race condition with the DIR signal (both changing at almost the same time). As far as I know the rising edge is most common, so I would go for that.

Best is of course to retrieve the timing relation of your motion controller for STEP and DIR, eighter by its specifications or by measuring it.

For your servo jerk problem, I doubt if this could be part of the problem because during a steady movement there should be no DIR signal change. Nevertheless, it doesn't harm to try the other edge and see what is happening...