[time-nuts] HP 5328 PSU nightmare... Or stupid engineer, you decide...

[Ed Palmer]

Hal Murray wrote:
> [variac]
>   
>> If you don't have one, you can wire a light bulb in series with the
>> power cord. Use 40,60,100 watt - whatever you need. 
>>     
>
> Neat.  Thanks.  That trick wasn't on my list.
>
> I think you can get lower wattage bulbs (at 120V).  I think I've seen 25W, 
> but I'm not sure.  Old style tungsten night-lights are/were 6W.
>   
Yes, and if your junk box goes back far enough, you may find some 120V 
Christmas tree bulbs!  Not sure what their wattage is.  Basically, use 
the smallest 120V (or 240V - depending on where you live) incandescent 
light bulb that gives you reasonable current / voltage in your sick 
circuit.  You can also use lower voltage bulbs on the secondary side of 
a power transformer if that works better.
>> The Variac is preferred because you may find that everything looks
>> good until you hit X volts. The light bulb obviously isn't that
>> selective.
>>     
>
> I'd expect a light bulb to work well.  If the DUT needs X volts to 
> demonstrate the problem, it will act as a high resistance at low voltage and 
> low resistance at high voltage.  Light bulbs are (very) non-linear in the 
> other direction.  If you connect the two in series, I'd expect it to be 
> stable at a balance point with some current, hopefully enough to debug things 
> without burning anything out.
>   
True, but with the light bulb you'd miss the fact that everything looks 
good below X volts.  That could be a significant piece of info.

> Here is another variation ...  I'm assuming you are chasing something like a 
> short on the power rail, probably a dead bypass cap.  Things get more 
> complicated if it's only sick rather than a solidly short.
>
> If you have localized the problem to one board, power that board from a power 
> supply with a current limit knob.  You may want to solder some wires onto the 
> board rather than using the normal connector.
>
> Crank the current up until you get enough voltage drop to be interesting but 
> nothing is smoking.  Poke around with a volt meter.  If you have traces 
> rather than a ground plane, a few amps and a reasonable meter will localize 
> things.  Even with a ground plane, it will get you pretty close.
>
>   
I've got a Jupiter GPS receiver that suddenly decided to start sucking 
silly amounts of current.  Never was able to find the short on that.  
I've got a better meter now, maybe I'll put it back on my list of things 
to fix.  Sigh.
> I think I remember dunking one board in a pan of Freon(?).  It was some 
> liquid that bubbled at the shorted cap.  That was ~30 years ago.  Maybe I'm 
> dreaming and confusing things with another story.  It does seem like a good 
> approach.  Are there any good Freon like chemicals available/legal these days?
>
>
>   
Freon was nice because it was non-conductive, evaporated cleanly, and 
wasn't flammable.  I'm drawing a blank on what would be a good substitute.

Ed