4
\$\begingroup\$

I am developing a mobile robot that has a battery with a nominal voltage of 24 VDC. According to a relevant norm, there shall be no electrical connection to the frame of the robot, though there are some exceptions. One exception are circuits with a nominal voltage not greater than 60 VDC, which are galvanically separated from their energy source. This is not the case in our current design. The battery is not galvanically separated, e.g. through a transformer, from the circuit it supplies.

My current plan is to connect GND of the 24 VDC circuit with a large resistor to the frame of the robot. I haven't decided the value of this resistor yet, but I would argue that by selecting a large enough resistor, with sufficiently high voltage rating, we would still be norm compliant. Let's call this plan A.

The benefit of plan A is that there should be no static charge build up on the frame that would cause a large voltage with respect to the electric circuit, as the large resistor would still allow very small currents to flow, thereby keeping the frame and GND roughly at the same potential.

Not norm-compliant, but a commonly used alternative would be a TN system, where GND is connected to the frame at one single point, e.g. in the electrical cabinet. The disadvantage that I see is with respect to plan A is that a GND-fault (GND connected to the frame somewhere else) may cause EMI issues due to the GND loop. This would not be an issue with plan A as the high value of the resistor would block ground loop currents. With plan A checking for such an issue is very easy. Measuring the resistance between the frame and GND must be roughly the value of the resistor. If it's much smaller, there must be a fault. With a TN system, this could only be checked by detaching GND from the frame and then measuring the resistance between the two.

The disadvantage of plan A that I see w.r.t. to a TN system is what happens in case of a live-fault (live wire connected to the frame). In case of the TN system it would trip a fuse, immediately causing the system to stop. With plan A, the potential of the frame would be driven by the live wire, again potentially causing EMI issues. Measuring such a fault would be relatively straight forward as the voltage from GND to the frame should be zero. If it's not, something is wrong.

A third option is a to connect a current measuring device between GND and the frame. If the current is too high, something is off. I didn't look into this alternative so much yet due to its higher complexity.

Are there important considerations, safety-wise, EMI-wise, or other, that I missed and that would favor one solution over any other?

\$\endgroup\$
5
  • \$\begingroup\$ Are there antennas present on the chassis and if so how do you plan to ground them? \$\endgroup\$ Commented Sep 23 at 7:53
  • \$\begingroup\$ @Lundin There are three devices with WiFi capability that are powered by the 24 VDC battery. I'm currently not aware of any specific requirements this could pose. Please explain if you see a need. \$\endgroup\$ Commented Sep 23 at 8:16
  • \$\begingroup\$ The antenna ground shouldn't have a wildly different potential from signal ground. Meaning that if you ground the antennas to chassis and the chassis has some floating potential, that will be the same thing as not having an antenna ground, leading to very poor performance. \$\endgroup\$ Commented Sep 23 at 8:30
  • \$\begingroup\$ Is this a stationary or moving robot? \$\endgroup\$ Commented Sep 23 at 12:58
  • \$\begingroup\$ @Jeroen3 it's a mobile robot without any permanent connection to a fixed structure. \$\endgroup\$ Commented Sep 23 at 13:39

1 Answer 1

Reset to default
2
\$\begingroup\$

Standards like IEC/EN 60204-1 and application standards (e.g., EN/BS EN 1175 for battery trucks) generally require no electrical connection to the frame, with exceptions such as circuits ≤60 V DC that are galvanically isolated, charger earth terminals, cable screens, suppression capacitors, etc.

Your idea of tying GND to the chassis through a large resistor (plan A) does have some practical advantages: it avoids large ground loop currents, keeps chassis and GND roughly at the same potential, and makes fault detection easy by simply checking resistance between them. The catch, though, is safety. If a live conductor touches the chassis, the resistor won’t allow enough current to flow to trip a fuse or breaker. That could leave the chassis sitting at a dangerous potential without the system shutting down, which is the main reason why this approach is hard to justify safety-wise.

The TN-style connection (solid GND-to-chassis bond) solves that problem neatly, because if there’s a live-to-chassis fault, protection devices will trip quickly and shut the system down. The downside is that you risk ground loops and EMI, especially if multiple unintended connections to the chassis show up.

The third option you mentioned (current monitoring between GND and chassis) is actually quite elegant. It lets you catch faults that don’t generate enough current to blow a fuse. But it’s more complex to implement, since it requires sensors, threshold logic, and a reliable hardware shutdown path.

In practice, the cleanest solution is galvanic isolation with isolated DC-DC converters or PSUs. If that’s not feasible, then a solid bond with proper protective devices and possibly added monitoring is the safer path. The resistor trick is clever for handling ground loops, but it’s not enough on its own to guarantee safety or compliance. If you cannot isolate, connect the GND to the chassis but design fault protection (fuses, circuit breakers, isolation monitoring) and plan cable routing to minimise ground loops.

\$\endgroup\$

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.