Big linear bench PSU trips RCD at turn on or off

[Herald1360]

Not cheap, unless you already have one, but running it from a big isolation transformer will sort it! (Though you may then have inrush current trip problems from that!)

Could be worth measuring the earth leakage current of the beast (temporary 1K resistor in earth line monitored by DVM across resistor is the usual way). 1V per mA of leakage. If the leakage ties up neatly with the Y cap values, then a judicious reduction in C may be in order.

[wave solder]

I had this problem once in a PSU, I linked the Neutral on the switch so only the Line was switched. That solved the problem

[PJL]

Suddenly stop supply of current to an inductor and you get flyback. There doesn't need to be much capacitance from the transformer winding to earth to create earth current. Do that many times in quick succession as the switch contacts arcs and you have a problem.

[G6Tanuki]

Had a similar issue a while back with the Diamond GSV3000 PSU I use as a general 12V 'office'-supply and to run things like my Yaesu FT897 and the occasional "Larkspur" military radio.

My fix was to follow theatrical-lighting practice and install a separate 16A-breakered radial [wired in 6mm T&E and terminated in a pair of 15A BS546 sockets] to the 'office', so it was independent of the RCD-controlled ring-main that fed the rest of the house.

No more "Hit transmit and hear the click-of-doom from the PDU-cupboard" nonsense.

[Biggles]

I would definitely agree with the double pole switch causing problems. I have seen this kind of thing happen when a double pole mains isolator switch was switched off with a heavy load connected. It was a mains supply to an outside shed and someone left an electric fire plugged in permanently, and thought it would be a good idea to use the switch in the house to isolate power. It didn't always happen, but was enough to be a nuisance. The switch poles obviously had a slightly different time constant.
Alan.

[MotorBikeLes]

A few years ago I designed a single to three phase rotary converter. Many big caps (100UFd at 450v plus others) and of course three lots of three phase motor coils all in there.Once I had the circuit sorted on paper, I realised some of the supply lines could be connected directly, thus saving the number of contacts required in the relays.
I found I was getting maybe 50% of my switch-ons causing my local 30mA RCD and the house's 100mA RCD to both trip out, causing howls from SWMBO. You can easliy reason out that with capacitance on one side and inductance on the other, unbalance throws the trips. Later I ensured all L and N feeds were made and broken together, with absolute removal of the trip problem. One relay needed six contacts (3 main plus 3 auxiliary), but it achieved the desired result.
Les.

[mhennessy]

I had this on my electronics bench. This has an RCD followed by a double-pole isolator switch. Occasionally, the RCD would trip when powering up the bench, even with a relatively small number of items being powered at the time.

The clue here was that the house RCD never tripped (the two are in series* - and the only time a fault genuinely tripped the RCD, both the bench and the house RCD trips went together!).

A replacement RCD in the bench fixed it completely. I've no idea if the old RCD was faulty, or just not very well designed, but the replacement was a second-hand model made by MK, a bit newer in design than the one I originally used.

So based on that experience, I'd consider replacing the RCD first of all.



*: When I put an RCD on the bench, I was well-aware that there's no real way to provide discrimination with two identically rated RCDs in series, but I eventually plan to re-do some aspects of the house wiring - long story. In the meantime, I do my best to avoid tripping either. Just one "event" in the number of years I've been here is pretty good going

[wave solder]

If you have two identical RCDs in series say one in the house and one in the workshop, the one in the house is probably going to already have a slight leakage flowing through it from other circuits so the one in the house would be more likely to trip than the one in the workshop.

[mhennessy]

In theory that's an understandable and reasonable assumption, and I initially thought the same, but in my experience I've found that both tend to go together.

I suspect that's in part because the leakage current doesn't gently rise until the first one trips - in a fault scenario the leakage is usually much larger than the threshold so both are triggered, and because they don't disconnect instantaneously, they both have power for long enough to operate the release coil.

An exception to that would be when you are gradually increasing the leakage current by adding more and more devices. Then you might well expect a more reliable discrimination, depending on what else is present.

RCDs are a relatively rare thing where I work, for that reason. A bay full of gear can have a surprisingly high leakage current - a typical SMPSU might leak half a milliamp or more, and you can get a lot of those into a bay. But they are present at every electronic workbench...

[Chris Wilson]

My best guess based on all this discussion is the there is sometimes a slight timing mismatch as the two poles of the PSU panel switch open or close. Hence why switching with gusto appears to minimise the tripping. There was much to be said for simple distribution boards with wired fuses and none of this safety stuff

[Chris Wilson]

I removed the chassis ground connection from the two caps to L and n and the tripping ceased, so i have to assume the 2 pole switch doesn't precisley change over both poles simulataenously. I am a bit loathe to leave it like that due to the SCR noise from the supplie's primary control and the nature of the Class D amp it is running, so will now install new Y caps which i have bought. thanks for the great replies!

Out of curiosity are you, "MotorBikeLes" the same gentleman that's on the Harrison Lathe forum??


Thanks again everyone, will report back if new caps sort it, or if it's something I have to live with unless i put a higher current trip RCD in the shack panel. Cheers.

[cmjones01]

Is there a way you could wire the class Y capacitors before the front panel switch, so that they're always in circuit? That way you'll get the benefit of them but without the risk of tripping when operating the switch.

Chris

[Ambientnoise]

I am not sure how switch timing (live relative to neutral) can affect this. Assuming no actual fault and normal uk mains, the cap from neutral will have little voltage and therefore little leakage. Leakage must all be from live so try a smaller cap there.
Rgds

Ken

[mhennessy]

Although my problems were cured by a different RCD - and by "different" I'm not necessarily implying that the first was faulty - the usual cause of nuisance-tripping at switch-on and switch-off with double-pole switching is capacitance to ground...

I had another look at the diagram in post #1 - are those Y caps really 100nF? I didn't spot that earlier. If that's the case, then that's the answer right there - you'd expect a few nanofarads in that role, not hundreds. Each of those might pass around 10mA peak, which doesn't leave much margin when you consider all the other devices scattered around the house. I found one reference that suggested that just the house wiring alone might account for 4mA in a typical domestic property...

IIRC, for class 1 devices, the limit is 0.75mA, which means no more than 10nF, give or take. Whether it's practical to reduce them is another question. A monster isolation transformer might end up being the solution, assuming you don't want to re-design the house installation...

[Radio Wrangler]

The turn-off tripping may be the give-away to the mechanism. I suspect the PSU is creating rather large transients on the mains, and any imbalance in the impedance to ground looks to the RCD as a transient leakage current, and the current/time profile may be enough to trip it.

I have a couple of large tool transformers which can do this as well, both on being switched on and off.

David

[See_Mos]

I just installed a 1Kw toroidal isolation transformer. The kick on that sometimes trips the house breaker. The solution was to use RS 903 2986 zero switching triac. The main terminals are permanently in series with the live to the transformer. The control side of the triac is connected one side to neutral and the other side is switched to live. The switch only has to handle a few milliamps AC at mains voltage.

[Herald1360]

A zero switching triac is bad news on a transformer with inrush type load. It will ensure maximum inrush peaks when the conditions for inrush are met. A peak switch on is (perhaps counterintuitively) much better.

[Chris Wilson]

Just to close this, I replaced the caps with new ones of the correct rating. No change. I didn't want to remove any caps as I use it to power a 1kW TX on 136kHz, and there's a real risk of noise. So I got hold of a 100mA trip RCD to replace the 30mA one, and fitted it yesterday. Problem ceased immediately. Many thanks for all help and discussion.

[Boater Sam]

You compromise your safety with 100mA rcd, I would advise you use an additional 30mA trip if working directly on a chassis with mains supply.
To give you some idea, on the farm even a 30mA trip rcd will kill a cow before it trips, 4 legs on the floor makes it worse.

[GMB]

I would try changing the capacitors to have a larger X class across L-N and a smaller Y class from N-E, so loosing the L-E direct one. The L-E capacitor looks a bit of a high value to me anyway.

It is also generally worth noting that when RCD tripping is a problem it can be an indication that some other appliance is leaking at a sub-trip level. If something is dumping 29mA into earth then you effectively have a 1mA RCD for the rest.