Grounding concept for an electrical machine test rig

Question

I'm tasked wich building a test rig for electrical machines which consists of

• Machines mounted on top of a large ground plate which is connected to PE through a thick cable

• A torque transducer: The torque signal GND is internally connected to the chassis and thus also to PE

• An oscilloscope which records the signals from the torque transducer.

• A resolver which also internally has the signal ground directly connected to the chassis and thus also to PE

• A digital controller (e.g. dSpace/RT Box). From Datasheet: "PE and signal GND are coupled internally by means of a high impedance RC network, consisting of a 1 MΩ resistor, a 1 µF capacitor and a 5V zener diode (SMAJ5.0CA), all connected in parallel". The resolver signals are connected to the signal GND of the controller and thus the signal GND is pulled to PE with two 3 meter long coax cables.

• Oscilloscopes, controller and PC communicate via Ethernet: The ethernet connection also pulls everything to PE through the Ethernet socket in the wall.

  

Now I'm wondering about the ground/PE concept. I want to have everything as failure proof as possible, because much hardware and software used in the test rig is self-made. I see 3 problems:

• There are multiple parallel PE loops. Supposing a high voltage hits the ground plate until the RCP fires.
  • There is the desired low impedance connection from the ground plate to PE through the grounding cable
  • There is an undesired current path from the ground plate to torque transducer signal ground(=PE) to the oscilloscope to PE
  • There is an undesired current path from the resolver (signal GND=PE) to the controller through the "high impedance RC network" (which I assume will block the fault current) to PE
  • There is an undesired current path from resolver/torque transducer to controller/oscilloscope and then through all Ethernet devices to the Ethernet socket in the wall which connects to PE

My questions:

1. Are my concerns valid? I'm concerned about a fault working through the entire test rig and destroying everything.
2. How is such a test rig usually built fail-safe (e.g. in industry)?
3. What do you think about (1) putting all Digital electronics behind an isolation transformer which breaks the direct PE connection and (2) using a WLAN stick to break the PE connection via the Ethernet socket? Then the only connection that remains is through the grounding cable

I apologize if the description isn't clear enough. I'm not a trained electrician, but more of a "theory idiot". Thank you!

Answer 1

For best noise immunity, use common PE connections with overvoltage and current limit protection and/or CM choke filter on signal lines.

Use STP cables for signals with source ground only to shield.

It is good to realize impedance paths for ground faults and use a GFCI and follow installation guides in manuals.

Industry standards were defined for immunity inhouse and gradually found their way into IEEE and CE standards.

PE is a standard practice for all equipment with PE plugs. Noise immunity depends on sources, cables, orientation and coupling and desired levels of signal/noise ratio.