Testing Power Supplies! Re: Norsk Data Nord-10/S restoration

From: Tony Duell <ard_at_p850ug1.demon.co.uk>
Date: Mon Nov 15 18:06:52 2004

>
> >>>>> "Joe" == Joe R <rigdonj_at_cfl.rr.com> writes:
>
> Joe> Testing power supplies before powering up a system is just plain
> Joe> paranoid! (And I'm trying to be polite here!) There are a
> Joe> thousand parts in modern computers that can cause just as much
> Joe> damage!!!!!!!!
>
> I disagree -- at least in the case of linear regulator supplies, which
> is what you'll normally find in a classic computer.
>
> A very simple and very nasty failure in a linear regulator is a short
> in the pass transistor. If that happens, you get a much higher output

Worse are the non-isolated switching regulators, like those DEC PSU
bricks. They have a higher input-output drop (typically about 30V DC on
the input side for a 5V output) and if the chopper transistor shorts, it
all appears across the output terminals!

> voltage, and there is no current limiting. It's a fair bet that this
> will fry a large fraction of the semiconductors in your system.
>
> If the supply has a "crowbar overvoltage protection" circuit in it,
> that's a different matter. But testing a linear supply is trivial.
> So why not do it?

I'd rather not trust a protection device if I don't have to. Sure I like
crowbar circuits, and they have saved my chips a few times, but I don't
like to depend on them.

-tony
Received on Mon Nov 15 2004 - 18:06:52 GMT