Electrical – Eliminating ghost voltage and breaking ground loops in a system

accapacitivegroundloopslinearswitch-mode-power-supply

enter image description here
The above illustration is a setup I use where amplifier outputs go to a data acquisition system through BNC cables. The data acquisition system is earth-grounded. Mains system voltage Line to Neutral is 230VAC.

A1, A2, A3. A4 are force-transducer amplifiers(they output 0-10V DC signals) which are powered by an SMPS type 24VDC power supply(PSU in the illustration). As far as I found out there is a ghost voltage(when BNCs are not plugged/ when floating) between the BNC connector tips and the earth(marked as gray in the illustration) due to SMPS Y-capacitor leakage. This ghost voltage is 85VAC when the PSU is ON and 180VAC when it is OFF.

Even though this ghost/phantom voltage is safe due to microampere level currents, It causes a very unpleasant sensation. I want to eliminate or minimize it.

If I wire the DC GND of the PSU output to the earth the issue disappears, but then since the DAQ system is also earth grounded there will be ground loops.

My questions are:

1-) If I replace the SMPS with a linear supply, would the ghost voltage disappear? And if so, how can I be sure if the linear power supply I buy is 100% linear?

2-) In another forum, I read some suggestions which were wiring the DC output GND of the SMPS to the earth through a 100 ohm series resistor or using anti-parallel diodes. It was mentioned, one of these supposed to drain the leakage currents but break ground loops. Any idea about this work around and how to implement it?

Best Answer

That ghost voltage is caused by the switch mode power supply's internal EMI reduction capacitors AND it is likely that the SMPS you have used either doesn't require an earth connection for safety or EMC reasons.

If the SMPS did use earth then it would route EMI reduction capacitors directly to earth instead. This would kill-off the ghost voltage you are seeing.

As an alternative, try earthing the negative SMPS output lead.

You should also consider operating the DAQ with balanced inputs so that any common-mode earth voltage problem is much alleviated. I know this means halving the number of avaliable channels (in some DAQs) but you will get better measurements and in the collection of telemetry information most of the industry I know of goes down the differential route if they can (even if the signals to be measured are single ended).