Re: When it rains, it pours...

[Michael J. Mahon <mjmahon@aol.com>]

Alex Freed <alex_news@alexfreed.com> wrote:
> On 1/11/2012 4:09 AM, Steven Hirsch wrote:
>> 
>> I'm going to start by adding a couple of electrolytic caps to the
>> LittlePower and perhaps ceramic bypass as well. I think PC motherboards
>> have better on-board decoupling than the IIgs and are probably more
>> tolerant.
> 
> 
> Steve,
> 
> there are two things that go wrong with the electrolytic caps with time:
> they loose capacitance and become leaky. Not always in terms of the
> liquid spilling out, but in terms of starting to conduct unwanted
> current. A cap in parallel won't cure that. My advice: bite the bullet
> and replace all the electrolytics.

I agree with those failure mechanisms, but his problem could not result
from a capacitor becoming leaky. Loss of capacitance (or the equivalent for
a filter capacitor--increased equivalent series resistance) is the only
failure mode for a filter capacitor that would produce his symptoms. 

For this situation, bridging is exactly the appropriate diagnostic
technique. 

If it were a coupling capacitor, leakage would be a failure mode as
well--and bridging would not undo the effects of the installed capacitor
leaking. 

Voltage measurements will usually reveal the presence of leakage current,
but disconnecting the "downstream" side of the coupling capacitor is
sometimes necessary (and *always* confirmatory). 

Leakage in a filter capacitor only becomes a failure mode if the leakage
current becomes large enough to overload the supply, which usually results
in the sudden, odorous destruction of the capacitor from overheating and
runaway leakage (both electrical and liquid ;-). 

As to replacing all the electrolytics instead of only the failing one, this
raises the "if it ain't broke, don't fix it" issue. 

Generally, capacitors of the same manufacture and age, exposed to the same
conditions, could be expected to have similar lifetimes. But if not all
factors are the same, very different outcomes are possible. Local heating
is a big factor, as are ripple current, voltage derating, etc. 

Since it is not particularly difficult to gain access to the caps, my
preference would be to replace only the failed one. Time will tell whether
that was enough for the next few years. And replacing any component on a PC
board entails some risk of further damage. 

I'm not a big fan of "shotgun" part replacement, since you never find out
which component was the actual problem or whether the replacement of the
other components was really worth the trouble and risk. 

-michael - NadaNet 3.1 and AppleCrate II: http://home.comcast.net/~mjmahon