Proper grounding of 7i77 and cables

Hello

I have recently finished (almost) my wiring up a 7i80/7i77 setup on a manual lathe conversion project. I have tested the spindle and servo drives successfully, but I noticed that the analog signal seems to be a bit noisy. Turning on the low-pass filter on the input in the drives helps out a bit, but I'd like to correct this at the root cause. 

The 7i77 is powered by a 24V switched PSU. The GND (-) is connected directly to the mesa GND.
I have fixed the boards to the back-plate with nylon washers and screws so they are isolated from the casing, ie. not grounded that way.
The cable that carries the analog signal (along with the drive enable) is a separate 4x0.25mm2 shielded (non-twisted) cable. I followed the 7i77 manual and connected the shield at the GND terminal next to the AOUT (the brown wire). That wire is in turn connected at the Analog Signal Ground at the servo drive. I have no idea how those ground terminals are connected internally as the manual has some chinese heritages
The servo drives are Delta ASDA-A2 btw.

Do you see anything that could be the cause for the noisy signal? I started wondering whether the shield is actually connected to anything...
I assume the plating on the screw holes is connected to a ground plane on the board.
- Should I change to metal screws so they are grounded to the casing (PE)?
- Should I connect the shield at the "real" power ground, ie. (-)-terminal of the PSU?
- Or could I expose a bit of the shield near the end of the cable and clamp it to the back plate? This is something I did for the I/O cables - bundled them together and clamped the exposed shields using a metal clamp.
 

The 7I77 mounting holes are not grounded.

Do the drives have differential inputs?

I would make sure that the drives frame ground is connected
with a short wide connection to frame ground.

You want to make sure that high frequency noise
currents from the motor drive side do not travel
through the analog ground connections to the 7I77

- The drives do not, from what I can tell, have differential analog inputs. There is an analog in and an analog GND.
- The metal frame of the drive is bolted to the metal back plate of the cabinet and there is also a dedicated 2,5mm2 PE wire from the case grounding point to a common grounding point in the cabinet, which is also connected to mains PE.

Please see attached picture.

The drive has separate GND's for analog signal and power (24V). I believe they are not connected internally, since I first left pin 13 unconnected, it resulted in the analog output not working. Manual states monitor outputs are relative to GND, ie. Analog Signal Ground.
Having both analog GND and power GND connected together to the common PSU (-) means whatever noise is on the GND will also be in the analog GND, right? I suspect that might be part of the problem, but how to solve it? I need to have the analog GND connected to power GND in order for the Monitor output to work...
How about the shield connection? It is now connected to brown AGND wire at the 7i77 end. Having pin 13 connected to common GND (PSU-) at the drive end maybe could cause some strange behavior as the grounding of the shield goes through the cable from one end to the other and then is grounded through a tiny wire that is the 0,25mm2 power GND wire?

Btw. the drives are in a separate metal enclosure and cables routed far apart from any noisy ones, so I think that should not be cause for concern.

 

This is the analog signal measured with an oscilloscope across AOUT and AGND:
 

The noise is strongest on Z-axis (800mV PK-PK) and about half of that on X and spindle.
The signal is clean across VIN/VFIELD and GND and also across unconnected analog out's, so I guess the noise must be coming from either the drives or be picked up from the air by the cable. Strange though that none of the drives had 400V power turned on...
Also worth mentioning is that when I put the scope's GND clamp on the back plate and the probe on GND (PSU-) there was a similar signal but stronger and more noisy.

I would connect the shield only at the 7I77 end
A line noise filter may help also.

When you say the drives are in a separate metal enclosure,
is the 7I77 in the same enclosure? This would be best for
ground noise issues.

"I would connect the shield only at the 7I77 end"
- You mean at the AGND terminal, as they are now?

"When you say the drives are in a separate metal enclosure,
is the 7I77 in the same enclosure?"
- No, it is in a different compartment/box.

You are probably seeing ground differences from one enclosure to another.
It would be much better for the 7I77 and FPGA card to be located in the drive
enclosure or at least bond the enclosures together for a closer to equipotential
ground.