RCD failure (the in-built test function that is)
 

Not been used for perhaps 8 years, tested numerous times previously in
use, all fine.
Go to use it again, push the test button and flash and a bang.
The 1/8 watt dropper that is switched in on test, connected to supply
side (240V here), partly burnt up and burned up thin traces.
That 1/8 W presumably MO resistor now measures about 1.8K, residul green
and brown bands, as 30mA trip, presumably was 510K.
Would there be metalisation creep associated with a MO resistor? or how
else could a MO resistor initially fail by "losing" ohms?
Obviously it would not have dropped to 1.8K in storage, just low enough
to overheat during half a second or so of powering, then runaway/sputtering.

RCD failure (the in-built test function that is)
 

If an instrument is "rode hard, put away wet", then there could be all sorts of failure modes from salts creep across some leads, corroded traces (especially across SMT chip connections) and similar. Where the failure showed up is not necessarily the first-cause.

I have seen situations (as recently as this last weekend) where an individual carried a tube tuner from his car to my clinic table in Kutztown for testing. What he did not understand was that it was a very hot and humid day, and he was dripping seriously through the cover into the guts. I did notice the sweat and cleaned it off before applying power. But had I not noticed - between the salts in the sweat and them corroding traces, can you say *POW*. Note that should there be some source of salts, these are naturally deliquescent - every time the humidity rises, the damage will go further.

Just a thought.

Peter Wieck
Melrose Park, PA

RCD failure (the in-built test function that is)
 

I agree with Peter. 240/30e-3 = 8K BTW

Metal Oxide resistors usually totally break in half under overload.

Maybe they should have raised it off the board,

They could have not taken into account the voltage rating: See http://www.hkresistors.com/-products...-resistor.html for an example.

240 * 1.4 = 336 V (0 to peak)

RCD failure (the in-built test function that is)
 

On 20/09/2017 06:32, Ron D. wrote:
I agree with Peter. 240/30e-3 = 8K BTW

Metal Oxide resistors usually totally break in half under overload.

Maybe they should have raised it off the board,

They could have not taken into account the voltage rating: See http://www.hkresistors.com/-products...-resistor.html for an example.

240 * 1.4 = 336 V (0 to peak)


I think that is the most likely, HV breakdown , perhaps 2x220K in series
would not have failed but 1x450K would.
The item was stored indoors , no damp involvement.
But it seems a bit odd that dozens of times tested in use and then fails
after an extended period of non-use.
Perhaps stressing while in use, cracking the casing , then a film of
mould or something organic can grow over the MO surface in storage.