From: email@example.com (Peter Gutmann)
Subject: Re: Strange failure mode in switchmode power supplies: Followup
Date: Fri, 27 Sep 2002 05:04:19 -0000
Organization: Posted via Supernews, http://www.supernews.com
References: <3D8F74CA.9DC367D8@usa.net> <3D931FC5.3C97466A@boeing.com>
Sender: firstname.lastname@example.org (Peter Gutmann)
Paul Hovnanian writes:
>John Woodgate wrote:
>>I read in sci.electronics.design that Peter Gutmann
>> wrote (in >>) about 'Strange failure mode in switchmode power supplies: Followup',
>>on Wed, 25 Sep 2002:
>>> This may be a clue (different ground
>>>potential in different rooms? Most odd
>>If you have a voltage drop in your earth system, you have current
>>flowing in it, and that should not be so, if it's more than a few
>>milliamps due to leakage and EMC filters. You should thoroughly
>>investigate - with care!.
>Yep. I'd check your outlet wiring for open grounds, swapped
>ground/neutrals, etc.Those little neon testers are OK, but
>They will glow on an open ground due to distributed capacitance.
>I'd check the grounds with a small load, like a 60W bulb
>connected between hot and ground. This isn't really safe if
>someone happens to be hanging on to the plumbing, so do
>it when nobody is home and unplug expensive equipment.
>Also, make sure you have a good neutral connection between
>the utility transformer and your panel. A bad neutral will
>usually manifest itself by lights going both dim and brighter
>as large 120V loads are switched on and off.
After having taken readings from every point where a voltage (or lack thereof)
might conceivably occur, everything looks normal. Speculating wildly, I'm
guessing that as the ferroresonant transformer produces a phase shift in the
line voltage, the end effect was one of running two pieces of gear connected
by a ground loop off different phases. How that would affect a low-
voltage/grounded piece of circuitry is unclear, but given the value of the
equipment involved I'm reluctant to try tracing the current path.
Since it only occurs when the ferroresonant transformer is used, and only
between the one computer and monitor, and I'd done extensive checking ages ago
when the problem first cropped up and somewhat less extensive checking now,
I've gone with the "don't do that, then" cure (the ferro wasn't making any
difference anyway, and made rather a lot of noise).
>Steinbach's Guideline for Systems Programming
> Never test for an error condition you don't know how to handle.
If only that worked for SMPS problems :-).