Measuring appliance earth leakage

[Ron Lowe]

Hi, All.

I'm having some RCD problems. We never had an RCD before the current building work. Our house has been moved from a single-RCD to a dual-RCD following my complaints ( ongoing building work ).

( In all liklihood, I will rip the CU out and replace it with individual RCBOs once they have finished and gone away! )

My computer room seems to be a big problem.

There are 5 PCs, an APC UPS, filtered trailing multi-way sockets, 2 large laser printers ( one a big colour one ), 3 LCD monitors, 3 big network switches, a firewall, 2 routers, a stack of cisco power injectors for the IP phones and aironet wireless access points, and other small junk.

Sounds like an earth leakage nightmare.

What's the easiest way to measure the actual L-N imbalance for individual devices?

There's a fluke clamp-round one, but you'd need to seperate out the phase and neutral and pass them through the clamp without the earth, I think. This emulates closely what the RCD sees. It's also close to £300!

Is there an easier / cheaper way?

I could make a test jig with a mA meter in the earth conductor, but I'm leery of that idea for reasons that I don't know...

[Andrew Gabriel]

I wouldn't expect that lot to run off an RCD (at least, not a single RCD). Data centres don't use RCD's, for exactly that reason.

By using an expensive PAT tester with that function built-in.

[John Rumm]

I probably have a similar quantity of this type of kit - and that seems happy sharing the same RCD as the other power circuits in the house. So you may find the leakage is not as large as you expect (assuming nothing is actually faulty).

You could make up a test extension lead with live and neutral broken out into a couple of coils that you could clamp round, you would then be able to make a similar measurement with a cheaper more basic clamp meter.

(I have a similar lead that I used for assessing the loading of IT kit when sizing for UPS specification - this only has ten turns of each conductor - enough to add a digit of accuracy to the clamp meter)

The main risk there would be in the event of a genuine fault to earth - the mA meter may not be able to offer a low earth loop impedance that would survive long enough to clear the fault. You could include a RCD in the test set to mitigate this risk.

Another approach would be to build a leakage injection box - basically some known resistance that can be switched across L/E to add a certain amount of leakage. That way you could ramp up additional leakage and see at what threshold the RCD trips. This would give an indication of the headroom available on the RCD.

[Graham.]

The IT gear doesn't strike me as the prime suspect for nuisance trips, do you have any reason for suspecting it is? If it does turn out to be something in there, then paradoxically perhaps, my first suspect would be the UPS as it will have filter caps between line and neutral to earth. PC PSUs would be second on my list for the same reason.

So I assume it is nuisance trips that we are discussing here? How frequent are they, and what else is on that same RCD

[Ron Lowe]

Yes. Engaging the MCB for the computer room alone usually trips the RCD. With the computer room off, I get no nuisance tripping.

Indeed, and engaging the UPS often trips the RCD.

Indeed, in addition to other loads, plugging in several computers often trips the RCD.

Indeed.

Several times a day;

A couple of other ring main circuits and lighting circuits. About 50% of the house. The house is now split about 50/50 between 2 30mA RCDs. Only when the computer room MCB is on do I have nuisance trips. The leakage in there obviously adds to any leakage elsewhere. The configuration is marginal, and I suspect no actual fault, just cululative leakages.

That's why I'd like to make *actual measurements* rather than guess.

I suspect I'm going to:

1) Ditch the existing CU and install individual RCBOs per circuit; 2) Drop RCD protection in favour of MCB-only protection on the computer room circuit. ( and to hell with the regs! ) if necessary.

[Dave Plowman (News)]

Many high power SMPS?

[Peter Parry]

You need to read Section 607 of BS 7671: 2001 relating to high protective conductor current which has been incorporated into Section

543 of the wiring regulations 17th edn.

[Andrew Gabriel]

It is. Each piece of Class I (earthed) IT equipment can leak 0.75mA. Maximum design leakage for an RCD is 25% of the trip rating, i.e.

7.5mA for a 30mA RCD. That's only 10 pieces of Class I IT equipment, i.e. 5 PCs and their monitors.

Often you'll get away with more as not every device will leak the maximum it's allowed to, and/or you might only get a trip if all is switched on at the same instant, but this is one of the reasons data centre kit isn't run through RCDs.

All Class I (earthed) equipment.

[meow2222]

The way to test appliances is with a megger. If you dont want to fork out for a megger, you can simply use mains voltage to test them. Apply live to L&N of the appliance, and apply mains E to the appliance earth. Put ammeter in live feed. Add resistor too to limit fault current to something the ammeter will suvive.

Using mains antisurge filtering does of course add earth leakage, as well as reducing reliability, increasing fire risk and wasting money. The protection these things give is piffling compared to the circuitry already built into IT equipment.

NT

[pcb1962]

I have a similar situation with 2 racks of cisco gear and servers with UPSs in the garage. My solution was to put the garage on its own ring with an RCBO on the non-RCD side of the split load consumer unit, which has helped matters enormously. I still get occasional nuisance trips when simultaneously switching on a bunch of computer equipment (eg plugged into a 6 way extension) on the domestic rings, it would be nice to find a solution to this.

[Ron Lowe]

I have access to plenty meggers.

Is it normally safe enough to megger L-E and N-E without damaging the equipment? My meggers have 250, 500, and 100V ranges: I'll use the 250v setting.

But don't meggers apply DC? That wouldn't see the AC current path through filter caps to earth that the RCD sees, surely?

[Ron Lowe]

"pcb1962" wrote in message news:67169c1c-863a-442e-be99- I have a similar situation with 2 racks of cisco gear and servers with UPSs in the garage. My solution was to put the garage on its own ring with an RCBO on the non-RCD side of the split load consumer unit, which has helped matters enormously. I still get occasional nuisance trips when simultaneously switching on a bunch of computer equipment (eg plugged into a 6 way extension) on the domestic rings, it would be nice to find a solution to this.

[Ron Lowe]

[blah blah]

Do commercial premises come under different regulations in regard to RCD requirements?

I'm guessing they must: our new-build at work was apparently done to 17th edition , but there's not an RCD in sight.

[Andrew Gabriel]

No. That was bad advice. You should check that the equipment conforms to some part of BS EN 60950 before you perform an insulation test, and then it's 500VDC max. Otherwise you perform a leakage test as I described earlier.

Correct. It's the wrong tool for the job.

[Dave Osborne]

Not sure what you're saying here. I've never heard of "Maximum design leakage" and it makes no obvious sense to me. The test specification for an RCD is that it does not trip at 50% of its nominal trip current and always trips at 100% of its nominal trip current.

A 30mA RCD should therefore never trip with a 15mA residual current and always trip with a 30mA current.

Can't see at all where the 25% comes in. Can you elaborate, please?

IME, 30mA RCDs typically have a trip current between 22 and 28mA. I have seen them as sensitive as 17mA.

If you are having a problem with nuisance tripping it is worth getting a ramp test done on your RCD to see what current it trips at. If it trips at a low current (i.e just above 15mA), try replacing it for one that's less sensitive (i.e. just below 30mA).

I know for sure that one well know manufacturer of portable power distribution equipment (for TVland, theatre and outdoor events, where earth leakage is quite likely and nuisance tripping can be a right pain) routinely buys RCD's by the hundred and individually tests them. Any (30mA nominal) RCD that trips at less than 25mA residual is sent back to the supplier!

[Andrew Gabriel]

The leakage you are expecting in the circuit. I described how you work out the contribution from Class I IT equipment.

That's correct.

So you make sure you design your circuit to leak no more than 7.5mA (if using a 30mA RCD) to avoid nuisance trips.

You should find out where the 15mA leakage is coming from.

They could tell them to sling their hook...

You can buy adjustable ones, but they might not go as low as 30mA (more usually up in the Amps range).

[meow2222]

You connect L and N together, and apply the HT from those to eath pin.

Really? Any mains appliance that cant handle mains plus a bit L to E would be hard to qualify as safe for use.

You said use a pat tester, which afaik simply uses a megger function. Where's the difference?

usually. There are ac ones too.

right. Hence the mains test option. Both options have their issues, the mains one tells you what the in-use leakage is more than what voltage the insulation can withstand.

NT

[meow2222]

This is merely a way to stop the RCD working. If you want to stop leakage, better to earth the secondary centre tap.

Another possible is to remove the filtering in the appliances and run the lot via one mains filter.

NT

[Ron Lowe]

Yes, that's right. Which is what the previous poster asked. It's creating a new final circuit that's not RCD protected.

However, instead of shoving a lump of ironwork in the way, [1] why not have the ring final non-RCD at the CU and have done with it. Same end result.

That's not easy. Goto [1].

[meow2222]

Ok on supplier earthed systems, but entirely unsafe on TT systems.

I guess it depends if you know some basic electronics. For those that do its easy enough. Not as convenient and warranty friendly as would be ideal though, and no use to some folk. Hence the 120-0-120 supply option - afaics this covers all bases. On a TT system you would add an RCD on the output side of the transformer.

NT