Best practices Multi Power supply grounding and common mode noise

[Jester]

I'm curious to hear best grounding practices when connecting multiple electronic assemblies that are separated by short distances, say < 10'

Assume isolation (galvanic/optical) is not an option (for some signals) between these assemblies.

Real world example, CNC machine
 - Electronic control board powered with 5Vdc, has some isolated I/O and some non-isolated I/O. used for all control (servo and spindle) drives as well as limit and control switches, encoders etc.
 - Servo power supplies 36, 48Vdc
 - Spindle drive non-isolated MAINS based power supply 200VDC, with control circuits with significant common mode voltage.

Interesting related regarding common mode:  https://www.djerickson.com/dc-dc/

I did some quick measurements at my bench (connections in parenthesis)
Assume chassis = Earth.

Bench supply for reference Agilent E3615A
Vout = 5Vdc, 40mvp-p @60Hz, and 300mV p-p random high frequency (floating)
Vout = 5Vdc, 60Hz pretty much non-existent, and 150mV p-p random high frequency (10r gnd to chassis)

Meanwell RS-15-5 power supply 5V, 3A
Vout = 5Vdc, 65 Vp-p! @60Hz,  (floating)  <-- Wow!
Vout = 5Vdc, 60Hz about 100mV, and 2V p-p random high frequency (10r gnd to chassis)


Meanwell RS-15-5 power supply 5V, 3A with C or R-C connected with short test leads
Vout = 5Vdc,  18Vp-p@60Hz,   (0.01uF gnd to chassis)
Vout = 5Vdc,  18Vp-p @60Hz,  (0.01uF // 100k or larger gnd to chassis)
Vout = 5Vdc,    4Vp-p@60Hz,   (0.01uF // 10k gnd to chassis)
Vout = 5Vdc,  340mVp-p@60Hz, and 2Vp-p @ high frequency  (0.01uF // 1k gnd to chassis)

Meanwell RS-15-5 power supply 5V, 3A with C or R-C connected right at PS terminals
Vout = 5Vdc,  65mVp-p@60Hz, and 100mVp-p @ high frequency  (0.01uF // 100r gnd to chassis)

Thoughts on either hard grounding or R-C grounding the 5V return of the Meanwell supply, obviously any connections from the main control board to the spindle control board (that has no mains isolation), need to be via optos?

[tooki]

Meanwell RS-15-5 power supply 5V, 3A
Vout = 5Vdc, 65 Vp-p! @60Hz,  (floating)  <-- Wow!

Perfectly normal for a switching power supply that is totally floating, but uses noise suppression capacitors from input to output. It's at roughly 1/2 the input mains voltage. It goes away when either side of the output is grounded.

[Jester]

Meanwell RS-15-5 power supply 5V, 3A
Vout = 5Vdc, 65 Vp-p! @60Hz,  (floating)  <-- Wow!

Perfectly normal for a switching power supply that is totally floating, but uses noise suppression capacitors from input to output. It's at roughly 1/2 the input mains voltage. It goes away when either side of the output is grounded.

Assume the power supply is located about 1/2m from the control board.

Would it be best to hard ground (short) PS return to Earth at the power supply or the control board?
Any downside to hard ground and that point vs. a R-C circuit?

[mtwieg]

First of all IMO all power supplies should be mains isolated, just from a safety perspective. Wouldn't even bother worrying about EMC until that's addressed.

Second, you need to be clear about what you want to achieve. Minimize conducted/radiated emissions? Reduce noise on specific signals/cables? Usually can't do all at the same time...