williamwright wrote:
On 14/11/2020 22:23, Roger Hayter wrote:
The only thing I can think of is an inductive load causing the
contacts to arc
and fuse. The monitors aren't ancient very large CRT ones I suppose?
No, new Samsung ones.
I mean, the weird thing is that some of the faulty switches have come
alright once they're removed from the panel. But I have to throw them
away, obviously. I have one in my hands now. It doesn't seem to be
caused by stress on the 1/4" connectors.
Bill
Inrush on a PC supply is 40A to 80A for a couple of 50Hz cycles.
OK, what kind of switch likes that.
I had a switch on a computer we designed at work, disintegrate
in my fingers one day while switching on. The switch was
destroyed by inrush. Switch contact rating was 10 amps continuous.
It took about 200 switch operations, to destroy it (one year, daily usage).
The senior manager in the group decided "he wanted this problem gone".
The senior manager and junior managers "owned" the chassis, because
there was a lot of interaction with the factory regarding the
computer chassis, so they owned that design and handled it themselves
without any help. None of the regular engineer talent was to
know what they were up to.
The switch was replaced with a switch plus a relay. The relay had
contacts rated at 40 amps. And it also happened to make a nice
quality "clunk" at startup. You could tell the machine meant
business, when you pushed the (ordinary) switch on the front
and you'd hear that "clunk" from inside the housing. The housing
for the computer was made from welded steel members. You could
drive a Ford F150 over it and not leave a mark.
I haven't seen a computer since, built like the tank that thing was.
I gather there'd been some complaints about the previous model
being "fragile" :-)
But, the electrical issue was fixed, and no more degraded
broken or fried switches were heard of.
The inrush limiter on PC power supplies is many things:
1) Protects rectifiers on front end.
2) Prevents I^2T issues with breaker panel at customer premises.
3) Does *not* solve all inrush issues!
Neat, eh. (3) is biting you on the ass. There isn't
enough inrush to kill the rectifiers, but there is
enough inrush to ruin your switch.
Now, don't ask me "how much switch is big enough switch".
Is it a 40 amp switch ? Is it a 10 amp switch ? Who knows.
You should do a post mortem on the switch and
look for carbon deposits (where the contacts close).
And now you have to tell the difference between
"normal" carbon deposits and "infrastructure destroying"
carbon deposits. Everything in the general area
is likely to have heat marks.
You can also alter the order of switch operation.
Turn off the PC at the back.
Turn off the "switch".
..... time passes
Turn on the "switch". No inrush, because PC is still switched off.
Turn on the PC at the back. (The PC switch eats the inrush.)
*******
NTCR1 is an example of an inrush limiter. This is an old design,
but the schematic nicely gets across some of the general principles
of ATX supply design.
http://www.pavouk.org/hw/en_atxps.html
The lower left is supervisor power. It provides power so the
"soft" power switch on the front of the PC works. It also
provides standby power for the RAM sticks in auto-refresh
while the PC sleeps. The upper right are the normal power
outputs, used in S0 ("Run") state. The fans spin, when the
upper right stuff is powered.
You might notice that design doesn't have a switch for mains
either. Unlike modern supplies, many of which have a switch
for mains. Some modern machines (like Apple) use a relay for
mains and sequence power to the supply. You hear a "click"
rather than a "clunk" when the power cord is plugged in.
Paul