I'm helping my friend's kid with a school project where their team has to produce a small ROV. Most of their design I was able to review and help make corrections with them but I'm stumped at one junction, because it's not something I've had to work with much... Ground Faulting. They need to detect a ground fault to the metal hull/frame of the ROV.

They currently have no design to accommodate this, and I had suggested they start by looking at current on the supply compared to return, but that's where I got stuck because that only accounts for the power use and will always have less current on the return due to energy converted to heat.

I've heard of people using a negative voltage on the hull/frame and monitoring that for ground fault detection. Does anyone have any experience with this and can suggest a design or point to a resource for this application?

Edit: There is no battery connection to the frame, but some control boards are mounted to it. The system is supposed to be fully isolated. The thruster motors are also mounted to the frame. They are home made thrusters, so the core laminations may not be isolated from the frame, despite being encapsulated.

Update - I probably should have said this the first time. There's an isolated power system, and a non isolated battery bus. I cannot supply schematics - they're not mine to supply. Does one use only one ground fault detection circuit or does it not matter if the impedance is large enough?

  • 3
    \$\begingroup\$ "that doesn't account for the power use and will always have less current on the return" - er, no, it won't. Current out will always equal current in. In fact, it's hard to see how current could get lost to the frame unless it's connected to the supply. \$\endgroup\$
    – Finbarr
    Jan 9 at 13:49
  • 1
    \$\begingroup\$ Is there a battery connection to the hull/frame (like the battery negative connection to a car chassis) or is the power supply supposed to be fully isolated? Please edit to put the details in the question (not in the comments). \$\endgroup\$
    – Transistor
    Jan 9 at 13:57
  • 2
    \$\begingroup\$ Hi ... what's an ROV? \$\endgroup\$
    – jonathanjo
    Jan 9 at 14:03
  • 1
    \$\begingroup\$ @jonathanjo Remotely Operated Vehicle \$\endgroup\$
    – Finbarr
    Jan 9 at 14:04
  • \$\begingroup\$ What is the dry-end power source supplying the ROV? What is the purpose of the GF detection, safety or corrosion? Is the communications to and from the ROV isolated? \$\endgroup\$
    – qrk
    Jan 9 at 19:25

3 Answers 3


To detect a grounding fault it is, perhaps counter-intuitively, necessary to create a deliberate connection to ground.


simulate this circuit – Schematic created using CircuitLab

Figure 1. Ground fault indication circuit. The lamps form a weak "pull to centre" arrangment so a meter would read 6 V between each terminal and the chassis.

How it works:

  • In normal operation there is no connection between battery positive or negative to ground.
  • Both lamps (which are rated for the full supply voltage) will only have half the supply voltage on each so they will glow dimly.
  • If the negative rail gets connected to the chassis then L2 will be short-circuited and turn off. Meanwhile L1 will have full voltage across it and glow at rated brightness.
  • Similarly if the negative reail gets connected to the chassis then L1 will turn off and L2 will glow at rated brightness.

This may satisfy the requirements of the project as you didn't mention that the supply needs to switch off in the event of a fault.

  • \$\begingroup\$ So how would something like this work if you have an isolated bus with your motors on the battery bus? Do you need this on both the isolated AND non-isolated sides? Seems to me that would cause a problem... \$\endgroup\$ Jan 11 at 13:46
  • \$\begingroup\$ This is the first mention of isolated and non-isolated busses. I think you need to update the question with a schematic of the power system. \$\endgroup\$
    – Transistor
    Jan 11 at 13:52

This is a variation of the other answer, but uses strings of a red LED, 7.5 V zener diode and series resistor, instead of lamps. This also doesn't switch the supply off in the event of a fault.

The idea is to have the LEDs off in the normal condition, rather than lamps which are dim.

The volt meters, ammeters and switches are just to allow circuit lab to test what happens in normal and fault conditions.

The following shows the normal condition, and there is only 650 fA through D1 and D2 so will be off:


simulate this circuit – Schematic created using CircuitLab

If manually try the switches to inject faults then get the following current through the LEDs which indicate grounding faults, and so should be bright. The series resistors R1 and R2 could be adjusted to change the brightness:

Switch closed Current through D1 Current through D2
SW1 8 mA 0 A
SW2 0 A 8 mA
  • \$\begingroup\$ The injection switches are a nice idea but I would use circuit breakers so that safety is assured if an inection test is made with a hard fault on one of the output lines. \$\endgroup\$
    – Transistor
    Jan 9 at 20:27

It appears that some commercial ROVs have a mechanism to measure resistance between circuit common and the frame of the ROV by the reports I get about leakage current problems. Perhaps you could rig up some sort of ohmmeter to perform this function.

There are commercial Ground Fault Protectors (GFP) for DC which see common usage in solar panel arrays. These are for equipment protection and the trip currents I have seen are not suitable for human safety.

Otherwise, you would need to build your own using a flux gate sensor or perhaps a hall-effect sensor (hall-effect may be prone to falsing due to drift and noise) and a pair of side-by-side parallel wires. If the current on the positive lead matches the return current on the negative lead the combined magnetic field of the parallel wires is zero. If there's a mismatch in the return current, the magnetic fields from the parallel wires would be mismatched and cause a net combined magnetic field that you may be able to measure. The circuit would do something like warn and/or disconnect. A flux gate magnetometer is susceptible to the Earth's magnetic field, potentially causing false alarms as the vehicle moves about.

Your comment

and will always have less current on the return due to energy converted to heat the outgoing current versus return current

does not make sense.

Consider a light bulb connected to a battery.
On the positive lead there is 1 A of current flowing.
What is the current on the negative (return) lead if the bulb is 50% efficient?
The answer should be 1 A unless there is another path for the return current.
This is the whole premise of current flow analysis in a circuit - what goes in must come out.

  • \$\begingroup\$ I agree with you - it makes no sense, but I've experimented with motors, power supplies, etc and seen a different return current from supply current with a device on a plastic frame and no ground connection. Again, my knowledge of ground faulting is minimal, thus reaching out to this community. I can only report what I see, and an ammeter confirmed what I read on my processor board. \$\endgroup\$ Jan 13 at 3:38
  • \$\begingroup\$ @JediEngineer Make sure your communication lines aren't carrying return current. The comms should be fully isolated and differential. \$\endgroup\$
    – qrk
    Jan 13 at 4:10
  • \$\begingroup\$ this isn't my project, I'm overseeing some of it for a friend. But they're using full duplex CAN for comms. \$\endgroup\$ Jan 18 at 13:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.