PDP 11 gear finally moved

[Tothwolf]

On Mon, 20 Jul 2015, Mark J. Blair wrote:

> Now on the topic of capacitors: The only component type that I replace 
> on sight at this point are the Rifa paper-dielectric EMI suppression 
> caps. Had one go incendiary on me so far, and I do a replace-on-sight 
> routine on them because my hypothesis of the failure mechanism(*) leads 
> me to believe that they're all likely to burn up once the plastic shell 
> has developed any cracks. They're easy to recognize: Rectangular, with 
> transparent yellow plastic housings, which are usually crazed with fine 
> cracks. Different caps which should not be subject to the same failure 
> mechanism are easily available.
>
> (*) Paper dielectric is said to absorb moisture from the atmosphere if 
> not sealed. So, I presume that once the yellow plastic shell cracks from 
> old age, moisture gets in, and then the caps break down under power. I 
> replace these with poly film safety-rated caps with suitable ratings, 
> since the poly film shouldn't absorb significant moisture even if the 
> housing seal fails.

Absolutely! I didn't mention those in my previous list since I was focused 
on aluminum electrolytics, but those yellow Rifa parts are an especially 
sore spot in older test gear. My own theory on these is that the swelling 
of the paper is what is causing them to crack. I'm not sure what chemical 
they treat the paper with, but it apparently doesn't hold up long term 
and/or this is just their failure mode as they age and wear out. While 
Rifa still makes these very same safety capacitors, I've been replacing 
them with MKP types from TDK which won't fail in the same way.

Oddly enough, those failing class X and class Y Rifa parts I see seem to 
be early 1980s to mid 1990s vintage gear, which puts them in that same 
20-30 year age I tend to use for replacing aluminum electrolytics.

Another capacitor type which I replace on sight are any wax paper 
capacitors such as you would find in tube (valve) based equipment. The wax 
coated tubular types are easy to spot, but the epoxy covered parts (black 
beauty, bumble bee, etc.) fail the same way. After replacing 100s, if not 
1000s of the things over the years, I've yet to find one, even NOS (which 
always seem to turn up in lots of parts from estates and such), which 
would pass a leak test.

I replace wax paper types with polyester (mylar), polystyrene or ceramic 
discs, depending on how they are used in the circuit (note however that 
for wound foil types, modern replacement parts do not mark the outside 
foil, which needs to be at ground potential in many tube circuits, 
otherwise the circuit can pick up noise and hum).