[time-nuts] HP 3586B Power Supply Failures

[Charles P. Steinmetz]

John wrote:

>Replacing known good filter capacitors is a zero-sum exercise at 
>best.  New aluminum electrolytics are relatively expensive, and 
>you'll be replacing parts that are probably near the bottom of their 
>bathtub-shaped reliability curves with parts that are definitely on 
>the left side of theirs.
>
>   *   *   *   After my first few encounters with high-ESR parts out 
> of the retail box, I stopped replacing good ones.
>
>I have no actual statistics to offer in support of either side of 
>the question, but in my own case, it certainly hasn't cost me 
>anything to leave good electrolytics alone.

I basically concur with John.  You can get good, reliable aluminum 
electrolytics, but you need the sort of Q/C knowledge that hobbyists 
rarely have.  HP/Agilent has a fine reputation for sourcing good 
parts, and very rarely has been caught out by batches of bad parts 
(unlike some other well-thought-of test equipment makers).  So the 
parts in your 3586 and 5370 are likely about as good as you can 
do.  I would NOT recommend shotgun replacement, absent evidence that 
your particular 3586 or 5370 was a victim of a batch of bad parts, 
and one failure -- regardless of how spectacular -- does not 
constitute such evidence.  For that matter, do you have actual 
forensic evidence that a capacitor failure caused the damage, or is 
that just a guess based on what you think is the likliest 
suspect?  Perhaps it was a diode (as in a spectacularly-smoked 5345 
of mine) or other component.

It is very rare for a failed electrolytic to cause the devastating 
damage that your 3586 suffered, so replacing them on an as-fail basis 
is a much better strategy.  Don't let the shock value of the 
magnitude of the damage blind you to a rational analysis of the 
probability and potential severity of future failures.

Finally, I would caution against wholesale replacement of tantalum 
caps with aluminum electrolytics.  Tantalums have a number of 
different properties compared to aluminum electrolytics, and it is 
not always obvious which difference led the designer to choose the 
tantalum for any particular location.  So, even if one decides to 
shotgun the electrolytic capacitors in a unit, it is probably best to 
replace tantalums with tantalums.  Even the best, low-ESR aluminum 
electrolytic may not do the job if the designer chose the tantalum 
for reasons other than low ESR.  Do you consider yourself to be a 
designer of sufficient talent and experience to second-guess an HP 
design team?  (I'm not suggesting that nobody is -- just asking if 
you are confident that you are.)

Best regards,

Charles