[SOLVED] Topic: 240Vac MAINS circuit in metal chassis with floating ground

[Jester]

I have been asked to modify an existing PCB that is presently in production and is supposedly UL approved for use in the USA as UL 508 (Industrial control equipment)

The grounding aspect of the board jumped out at me as possibly incorrect. So if your up on ground bonding and creepage please read on and let me know if If this is not quite right or I'm simply not interpreting the rules correctly.

I'm under a NDA so I can't share every detail, however I can paint a pretty clear picture without disclosing anything confidential.

Details:
1) This device is powered from a 240V branch circuit protected with a 20A breaker
2) The device is used to distribute 120/240V with taps at each point. Much like a standard power bar, however several of these are typically cascaded to allow tapping off of power at various locations while passing the power on to the next stage.
3) The device is in a metal chassis
4) The device uses flexible metal conduit for the power in and out connections; L1, L2, Neutral and safety Ground
5) Within the metal enclosure a PCB is used to route the power (L1, L2, Neutral and Ground)
6) A simple circuit on the PCB also taps into the 120/240V via a transformer to some user accessible low voltage circuitry (I don't see this as relevant, just noted for clarity)

Now the perceived issues:

a) The pass through ground connection on the PCB is not bonded to the chassis, the ground trace is treated the same as the other live and neutral connections and is essentially floating within the constraints of the PCB.

note that the ground from module to module is continuous in that the flexible conduit is connected to the chassis (via sheet metal screws on the tab of the conduit connector), and the safety ground conductor passes through the PCB on the traces, however their is no intentional connection between these two ground paths.

b) the creepage between the ground trace and the adjacent neutral is 3mm (119mil)
Now if the ground trace on the PCB was bonded to the chassis, I think the 3mm creepage would be fine (more than adequate), however if UL interpreted the PCB as an ungrounded device then I think the ground to neutral connection would need to be reinforced or 5mm

This is what the board looks like

 

[FvM]

I see this requirements (not UL specific, just plain IEC 1010)

If the chassis isn't connected to safety ground (through cables with sufficient cross section according to the power supply fuse), the circuit needs reinforced isolation against the chassis metal. Using "metal conduits" or simple isolated wires without an additional isolating jacket won't be acceptable.

Routing PE through PCB is only acceptable if the PE conductor cross section is sufficient for 20 A fuse, e.g. >= 1.5 mm².

Creepage between PE and other mains terminals must not necessarily comply with reinforced isolation, I think, but that's probably a lesser problem.

 

[Jester]

I see this requirements (not UL specific, just plain IEC 1010)

If the chassis isn't connected to safety ground (through cables with sufficient cross section according to the power supply fuse), the circuit needs reinforced isolation against the chassis metal. Using "metal conduits" or simple isolated wires without an additional isolating jacket won't be acceptable.

Routing PE through PCB is only acceptable if the PE conductor cross section is sufficient for 20 A fuse, e.g. >= 1.5 mm².

Creepage between PE and other mains terminals must not necessarily comply with reinforced isolation, I think, but that's probably a lesser problem.

It seems to me the simplest solution is to simply bond the safety earth trace on the board to the chassis, that way the chassis is at ground potential and the need for reinforced isolation is no longer needed. Does this make sense?

 

[FvM]

Yes, that would be also my suggestion. Sufficient cross section of PE connections between input cable/connector and chassis as well as outputs with PE must be however achieved. I guess the G1 trace doesn't have it.

 

[Jester]

Yes, that would be also my suggestion. Sufficient cross section of PE connections between input cable/connector and chassis as well as outputs with PE must be however achieved. I guess the G1 trace doesn't have it.

Although not obvious in the image, the through traces are 3.3mm wide, are on top and bottom and are 2 oz copper, so even at 20A the temperature rise should be less than 20 deg C

Thanks for commenting.

 

[FvM]

Yes, that should work. Also total melting integral of a 20 A fuse (around 1000 A²s) won't damage the trace.