https://control.com/technical-articles/comparing-voltage-when-to-ground-a-floating-power-supply/

 

"The most common reason for tying grounds together is to prevent dangerous electrical shocks. In general, if two pieces of equipment are within arm’s reach of each other, their grounds should be tied together. If they are not tied together, the two machines may have a voltage difference. If a worker touches each machine, and the potential difference is high enough, the worker could receive a dangerous electrical shock. " {Quoted from the article}

I remember in high school electronics class we had a signal generator and an oscilloscope (back in 1973) and were experimenting with different signals and how the scope displayed them. By chance I placed one hand on the scope and the other hand on the generator and got a pretty good shock. If memory serves, both units had two wire power cords. I thought I was being smart by flipping one of the two plugs over. Upon demonstrating to the teacher how I got the shock - I got another shock.

Mr. G was fun as heck. He had a healthy fear of electricity. Often whenever he'd plug a device in that he had been servicing I'd bang pliers down on the work top simulating the sound of an arc, which ALWAYS made him jump. Then swear using my name. He was the teacher I liked the most.

 

https://control.com/technical-articles/comparing-voltage-when-to-ground-a-floating-power-supply/

Good points. I don't automatically Earth (solid connection to the Earth grounded equipment frame) all AC/DC isolated power supplies on high power industrial equipment automation control systems because it can be dirty and inject noise into sensitive circuits signal common. You need to use supplies with very low AC supply leakage (no Y cap) to the DC terminals for critical floating (utility potentials) power applications.

 

My philosophy when wiring control cabinets has been to set up a star GND spot using a copper bus, and take the service earth conductor, & all PS commons, shields etc.
Also adopting most of the what the Siemens paper on equi-potential bonding suggest (prevent ground loops). Also with that in effect, ground both ends of shielded cables.
So far it has served me in good stead.

 

My philosophy when wiring control cabinets has been to set up a star GND spot using a copper bus, and take the service earth conductor, & all PS commons, shields etc.
Also adopting most of the what the Siemens paper on equi-potential bonding suggest (prevent ground loops). Also with that in effect, ground both ends of shielded cables.
So far it has served me in good stead.

That's what I do for motor control signals but lots of other things live in a world of floating potentials where things like Triaxial cable is needed. Signal, signal shield/ active guard and 'ground'.

https://en.wikipedia.org/wiki/Triaxial_cable

Another application for triaxial cables is for probes taking precision low-current measurements where the leakage current through the insulator between the core and shield would normally alter the measurements. The core (known as the force) and the inner shield (known as the guard) are kept at approximately the same electrical potential by a voltage buffer/follower, thus the leakage current between them is zero for all practical purposes, despite the imperfections of the insulation. Instead, the leakage current occurs between the inner and outer shields, which does not matter since that current will be supplied by the buffer circuit rather than the device under test and will not affect measurements. This technique can provide almost perfect elimination of leakage current but becomes less effective at very high frequencies as the buffer cannot follow the measured voltage accurately.

https://forum.allaboutcircuits.com/threads/ctmu-clarifications.115248/post-899468