[time-nuts] HP Stories: An architectural view of the HP 5060/5061 and awkward oscillator adjustments.

[Poul-Henning Kamp]

--------
In message <CALiMYrs6cWfp2mNOTwHHMhZYFinZyE5U3MCCFuhj-K+Ubc6VcQ at mail.gmail.com>
, Adrian Godwin writes:

>I think I've had as many shorted-out tants as dried-out electrolytics.

I doubt Len worried about dry-out, he was worried about shorts.

A lot of people failed, and still fail, to realize that tube-voltage
electrolytics are an entirely different kettle of fish from
semiconductor-voltage electrolytics.

The crucial difference is that in an unpowered high-voltage 'lyt the
oxidelayer will degenerate over time, and then when you apply power
again, it shorts through and bad things happen.

This is where the "Ramp up the voltage with a variac" thing comes from,

As you raise the voltage slowly, the oxide layer has time to reform
to the required thickness for the the normal operating voltage[1].

Low voltage 'lyts do not have this problem, because the final layer
of oxide is not going anywhere in benign chemical environments.

Before that knowledge had rippled out through the EE community,
'lyts had got a reputation for being unreliable and prone to
"unprovoked" venting and/or explosion, in particular in the
"grew up with tubes" generation of EEs.

There is no precise voltage to separate "high" and "low" voltage,
it depends on the chemical formulation of the electrolyte, but as
a rule of thumb 'lyts exposed to less than 50V do not short when
powered up.

Any 'lyt with a higher operating voltage should be ramped up slowly,
if it has been without power for more than a year or two.

Where this gets tricky is badly designed switch-modes, where the
control circuit fails to start or latches up.

Best advice there is to use a dummy load rather than Unobtanium if
you can, or try find a way to disable the control circuit
until you have ramped the 'lyt up.

Drying out is an entirely different failure mode, caused mainly by
high temperatures, either from ambient (Tubes) or self heating due
to ripple-current in the ESR.

Poul-Henning

[1] I have tried to find out how slow one needs to ramp.
    People with chemical clue tell me that electrochemistry, is
    a matter of seconds.  I round up to minutes.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.