[vishnu]

We use Silastic J in our pulse forming networks at work, 22KV, works great... 😁

[jan-jaap]

If there are voltages in excess of 320V present then the sealant is really the least of my problems ;-)

From the types robespierre mentioned I can get DOWSIL 738 (but not before August) and 3145 (immediately, but it's €90,- for 90ml). MIL-spec always adds at least one '0' to the price...

I get get DOWSIL 732 for an acceptable price which according to the DOW 'Adhesives and sealants selection guide' is (1) general purpose rather than specifically for electronics and (2) faster to cure. The latter doesn't bother me since I wont be assembling parts but rather just putting a blob here and there.

Alternatively I can get Tecbond 7718 hotmelt from work. It's like hotglue but resistant to 135c instead of the 70c for regular hotglue. It hardens beyond rubbery, but not like 2-component glue which basically becomes glass. I've used it in the past to stabilize Dallas RTC hacks:

[robespierre]

You should make sure that the adhesive you select is not acetoxy curing, as that type has a corrosive byproduct. Many silicone caulks are unsuitable for that reason.
A rock-hard glue that is impossible to remove would also not be my first choice.

[vishnu]

When we use hot glue on our circuit boards, we use this beast:

https://www.3m.com/3M/en_US/p/d/b40066900/

With this adhesive:

https://www.3m.com/3M/en_US/p/d/b40065551/

[jan-jaap]

Rebuilt and revived a second broken PowerOne PSU :-)

Now that I have some measurements of a working PSU, repairing more of them becomes more structured.

In yet another one, leaky caps had done more damage:



The tin on the contact pads is gone and the copper corroded (but not etched away yet). I will not be able to solder new components until I fix this up.

[robespierre]

Those look like they could be cleaned up using a fiberglass pen (polishing to get to bright copper), activated rosin flux, and tinning with a cool-ish soldering iron (like 300° C) [lower temperatures keep the flux active longer before it burns up]. You might have to scrape the soldermask off with a scalpel to polish the whole copper pad. Also a cotton bud in 5% acetic acid followed by a water rinse can help with removing corrosive junk (although there are better solutions for ultrasonic cleaners).

[jan-jaap]

Those look like they could be cleaned up using a fiberglass pen (polishing to get to bright copper)

Excellent tip, thanks. This software guy is learning something new every day :-)

We happened to have one of these in the lab so I gave it a try. It's quite abrasive, I guess you don't want to overdo this. And as always when dealing with fiberglass you'll have itchy hands afterwards.

I'll do the rest at home, I have all the material there.

[robespierre]

Yeah, part of the skill is knowing how far to extend the fibers from the tool. The longer they are, the more gentle on the work (because they can flex), but the less strain they can take before breaking off. Downwards pressure also should be minimal: as with abrasive sanding, you're not trying to gouge the workpiece, just knocking off high spots.

[jan-jaap]

Another case of leaky capacitor damage:



This cleaned up nicely. You wouldn't be able to solder to the pad without cleaning it first. Most of the components with date stamps on them are from end of 1989 (week 45 ... 50), the custom PowerOne transformer is labeled 1/1990 so it's roughly 32 years old.

This PSU had been repaired by a previous owner. Again the bias supply inverter FET. In the majority of PowerOne SPM5 PSUs I've seen this part has failed. I think I'll try to fit it with a small TO220 heatsink to help keep it cool(er).

This power supply is from my 4D/380 VGX "Predator" rack. I had not run this system in a while and I was afraid to power it on after the carnage I found in other PowerSeries PSUs. It was originally owned by a German Fraunhofer Institute. It's this very system, the article dates it to the 2nd half of 1990. VGX (and 33MHz IP7 CPUs) were first introduced with IRIX 3.3 in 06/1990. I kept the name "DALI".



Also, I finally received the silicone adhesive to finalize the rebuild of the first two power supplies.

[vishnu]

As a power supply engineer myself I appreciate the difficulty and delicacy required of your efforts, that system is a treasure and all the credit to you for keeping it up and running! 😍