Three phase earth bonding

[Theo]

Suppose I have a variety of equipment which is variously powered from one of the phases of a three phase supply. I'm concered about the potential difference between two boxes on different phases. The boxes have low voltage electrical connections between them (similar to USB ports). Each box has a PC-style switching PSU. Leakage from the mains side into the DC side could cause a high potential difference between boxes on different phases, which would be enough to zap sensitive inputs on the interconnections.

Suppose I connect the ground of the low voltage side of the PSU to that box's chassis, and then bond all the chassis together and join that to earth. Will that be sufficient to prevent any potential difference between phases arising that could cause trouble with sensitive inputs? Are there any issues this might cause (such as EMI, or tripping the RCD)?

Is there a common way to detect automatically that this earth bond has come adrift (ie not by waitng for the annual PAT test)? I suppose checking for a voltage between ground and earth, or putting a voltage across them and checking the current hits some limit. But do people ever do this sort of thing?

Thanks Theo

[John Rumm]

I would have thought any decent SMPSU would have good isolation between its mains and ELV sides anyway.

Depends a bit on what you mean the "low voltage side" of the PSU? Its casework etc I would already expect to be connected to the chassis electrically. The "ground" / "-ve" etc of the output would not necessarily be connected directly to mains earth.

It might also be sufficient to introduce lots of mains (or local earthing system) bourne interference into your signal ground. So I don't think I would want to join them.

In multi phase environments with Class I devices, then its important to make sure they do have a working earth connection right up to the appliance. Generally the earth connection in the supply cable for an appliance is also considered an adequate EQ bonding conductor.

Not the kind of thing a PAT would find...

[Theo]

I would hope that there would be very little current flow. However it's voltage that kills electronics, and a capacitive connection might be enough to zap. I've felt leakage through supposedly good quality Apple PSUs when using a non-earthed mains cable (to a two-pin non-UK wall scoket).

I meant connecting the case of the equipment to GND on the PSU's output.

Do you mean the metal case of the equipment would be grounded with respect to the LV PSU output, but not necessarily mains earthed? I'd have thought it should be earthed to prevent live-to-case faults?

I'm confused. Bond the case to the mains earth, but not to the LV ground? In which case, would you have a separate low impedance ground connection between equipment on different phases? Would that avoid mains interference?

(Bearing in mind that there will be a connection anyway via data cabling, although not necessarily low impedance).

Theo

[John Rumm]

That's down to the mains input filters typically rather than the PSU design itself. As you say, on a properly earthed supply the problem goes away.

Yup, I don't think I would be keen to do that personally.

The case of the PSU will be connected to the case of the equipment, thence mains earth. However the DC 0V line out of the PSU might not be - and I can't see any advantage to adding one if its not there already.

Data cabling either has data ground and a separate and non connected shield. Or just a data ground in many cases (depending on what kind of connection it is)

If a shield is present that will likely couple to the casework on metal clad machines, and hence mains earth.

[Theo]

Indeed, although such grounds are not always low impedance (think the teensy thin wires on USB cables).

So do you reckon it's better to have a low impedance ground bond (connect all the GND lines of different boxes together, not relying on data grounds) and separately earth bond the chassis together?

Theo

[Rob Morley]

Sounds like a job for opto-isolators.

[John Rumm]

They are there to facilitate comms, not provide equipotential bonding... i.e. they are functional earths and not protective ones.

In the general case, I think you might be over thinking this a bit.

In a multi phase environment then its important that any Class I kit has a proper mains earth connection. That way any chassis can not be dragged to half line volts by its filter caps (and where those line volts may not all be the same to due to different phases being used on different circuits). The earth connection in the mains lead should also provide an adequate EQ bond in the vast majority of cases.

Data connections may have a shield - which will hence bridge one chassis to the next. The data ground however may or may not be tied to mains ground. If it is, you don't need to worry since they are all properly earthed, ans so should be seeing pretty much the same potential. If they are not, the they are floating wrt to "true" ground, and hence again nothing to worry about.

(the exceptions here would be when discussing interconnected systems where lightening strikes are more likely).