Rod said:
(e-mail address removed) wrote
No evidence the the PSU has anything to do with the problem.
He's done that by trying another power supply.
Complete waste of time when he has already tried a different power supply.
Running the PSU out of the case is a more sensible test, and it'll be
done anyway. I would use the results of that one, since we're
unavoidably doing it anyway.
In practice, there are times when one makes assumptions that are quite
reasonable to make, in narrowing down a problem, and then you hit a
wall, and on reevaluating it from scratch with a fresh mind, you
realise one of the assumptions is wrong. It is very rare that a fair
assumption proves wrong, but it happens sometimes.
In his case, there's a slight chance that both PSUs have the same
problem of rattling inside from the wire hitting the fan. And when he
lifts the case it moves that cable a bit. And there may be unforeseen
issues too that cause one to question assumptions. So i'm not going to
jump into assumptions. It's good to double check, and choose your
tests carefully. And when looking at possible problems,solutions and
tests, it's a good idea to double check, especially if a double check
takes 0 seconds(it being patr of another necessary check).
I'll give you one of the rare examples where I've made an assumption -
or rather assigned a wrong probability (since I am willing to revaluate
from scratch with a fresh mind ignoring assumptions. Since one is only
human. ).
I had 2 computers hooked up to a DSL router/modem. One of them, I
never had any problems with the internet connection. The other one,
the internet connection kept going from time to time, then coming back.
Every time, I checked the other computer, saw that it was fine. The
router configuration is ok, it hasn't been changed. There's no in built
firewall in the "router" misconfigured.
So I figured that the router is fine(working), since one of the
computers gets to the internet with it.
I tried plugging the "computer that worked" into different ports, and
it still worked. So that double checked that the router is fine, in all
ports. It was really looking like it was this one computer that was
the issue.
To narrow things down fast with that one computer, I booted knoppix,
and got no internet connection (usually i would). So, it wasn't a
broken software firewall on the computer blocking the connection.
I tried a cat5e cable from the working computer, and it worked in the
other one, but then I lost th einternet connection again. Put it back
in the working one and the working one was fine.
This was narrowing down to the NIC in that "computer that wasn't
working".
I replaced the NIC, and got net access for 5 minutes, then failure
again.
I started getting bothered, because I could continue testing this, but
it is not a good possibility , infact the only possibility that I could
think of, and the most likely, that - and this is strange - that the
computer was assassinating my NICs. Could be a bad PSU, could be a bad
MBRD.. I don't want to keep trying new NICs in that computer and
potentially having them assassinated.
So I quite looking at the problem. After about a week, the "working
computer" suddenly failed to get an internet connection through the
router!! Then I got suspicious that the router was at fault!
I had an old router which I had declared broken but left it in the
shelf incase it had any life in it after being left around for years,
in the desparate event that i'd have to bother trying to use it. It
worked for about an hour!! And gave all my comps internet access. So
I went out and bought some new working routers - with warranties , from
my ISP!
So there you see a good example. Everything pointed to the router
being fine. My checks and double checks that I described. (though I was
limited, not having another working router to try). It really looked
like the NIC.. But that turned out wrong.
What you are doing, is making assumptions absolute.
I might find myself saying "I find it hard to believe that this is the
problem..." or "this is most likely, this is unlikely". And I do rule
things out.. But not permanently. And in deciding tests and possible
solutions, I look at how much time they'll take. If it's a great test
and it takes 0 seconds, i'm all for it. And i'm not against doing it
twice to be sure!
That is why I interleaved my comments,
so I can comment on particular bits of yours.
the method is snippnig and interleave. It applied here.
Nope, they have bleed resistors across them.
And out of interest, were these 'bleed resistors' also on all old AT
power supplies and All ATX PSUs?