Digipot variability

Hi All, 

Not a major issue, but my spindle speed is varying significantly under contant load. Sometimes during a cut and sometimes when it's just sitting and spinning up under no load and should be at constant speed. I'm not sure how much it varies because if I try to probe the pin the variance disappears, but audibly it slows down as if I ploughed it into a slightly too heavy cut that wasn't enough to stall the motor but enough to bog it down noticably. 
My VFD (AT4 Aliexpress job) is driving a 3 Phase induction motor, 2.2kW, 2 pole. 

The VFD has a 24v output but takes a 10v analog input, so I've just wired the 24v out to a simple voltage divider resistor pair (<5k total resistance) and I'm using that to feed the high side of the digipot on my 7i76e with 7v or therabouts. I tried using my 24v rail that supplies the limit switch, e-stop etc  originally but the spindle speed was varying a lot (under constant load) so I assumed the VFD reference voltage and 24v rail might be too de-coupled and switched to using the VFD output thinking that would be a more consistent Vref for the VFD, but the spindle speed is still varying a lot.

My .ini file has got the SPINDLE_SCALE set to 8000 (It's an 8k RPM spindle), so I used M3 S8000 and then adjusted the vfd paramaters (P73) to get 60Hz, then checked at M3 S4000 and got roughtly 30Hz showing on the VFD screen so it seems to be working OK at max rpm and  linear to within a few hundred RPM at 50%. 

If I try to probe the voltage on the pin the impedance of the multimeter is enough to pull the VFD back to max RPM as soon as I probe it so I can't tell what the VFD vout or analog signal from the 7i76 is when a dip occurrs. My instinct is that the multimeter has some huge resistance in voltage mode and that a weak pullup / pulldown (1Meg?) would help, but I'm not sure so want to check to see if this is typical behaviour, I'm surprised that it doesn't 'just work'

On a separate (but potentially related?) note, I've got a significant AC voltage leakage to the machine frame ground, I'm not sure if this is the VFD itself or the SMPS which feed 72v to the stepper drivers, I need to do some un-plugging and checking. It's enough that you get a tingle if you touch the frame and you're grounded. I'm getting no DC voltage from machine frame to protective earth in the garage, but 20-60v AC showing on the multimeter. If I use current mode and check leakage current from machine frame to ground it immediately trips the RCD, so >>30mA. Not sure if this is an independant thing or if this leakage could be messing with the analog voltage signals. 

Thoughs? Next steps? Independant things or related? What would you do next....

My guess is you have a large amount of HF noise so may need a line
filter on the VFD input, common mode choke on the VFD output + better
(lower impedance ) grounding.

The (typically 10M Ohm) load from the voltmeter should have no significant
effect on the output voltage, but in a high EMI environment, the added
capacitance of the test leads may make a noticeable change.
 

I'll order some filters and stick them on.

I've got some big ferrites in a box here too so I'll stick them on but looking online all of the motor chokes I can see are bloody huge, bigger than the VFDs supplying them so I suspect my ferrites are in the wrong league.

Thinking of it now I believe I get the same behaviour with the VFD not acutally running, so analog voltage being supplied but with no direction signal it will still display the commanded frequency on the screen, I'll test this afternoon