Rebuilding the Sarkes-Tarzian #50 Selenium Rectifier in the Collins 516F-2 Power Supply

This procedure posted by Dave Tumey W5DT to the Collins mail list, April 28, 2011, and posted here along with Dave's pictures, with his permission.

I have had the chance to rebuild several of these now and have refined my technique a little - the results are now so good, my XYL can't tell the difference between old and new!
 
I had a S/T #50 Selenium rectifier (neg. bias supply) go south in a 516F-2, so I decided to rebuild it using a Silicon diode substitute; I wanted to post the process I used to the list in case any others have the same problem (which I understand we all will as the Selenium rectifiers don't age well).

1. Remove the protective cage from the three large "bleeder" resistors.
2. Remove the nut and lock washer that secure the S/T #50 bias rectifier.
3. Using solder braid, desolder and remove the leads from the components on both legs of the module.
4. Remove the rectifier from the power supply.
5. Using a Dremel tool and ball-mill, remove the potting material that covers the top of the Selenium stack until the stack can be clearly seen.
6. With a smaller ball-mill, clean out all the "potting" material along both sides of the Selenium stack (the sides next to the connecting plates) to at least 1/2 the depth of the case.
7. Mount the rectifier by threading it into a secure support (I screwed it into nut held in a vice).
8. Heat the rectifier stack with a heat gun - heat it up real good - the potting material will soften and the side conducting plates can be removed along with the stack by gripping and pulling with needle-nose pliers. Wiggle the stack back and forth until it comes free - be careful not to crack the Bakelite case - if you do....super glue.

9. Clean everything up as best as you can.

10. Bend the Anode lead of a Silicon diode (I used a 1KV 2.5A job from Shady-o-Rack) up in a "U" shape. Clip the lead about 1/4" after the bend.
11. Solder this lead to the right side of the front of one connecting plate.
12. Bend one lead of a 100 ohm 1W resistor (the new ones are about the same size as the old 1/2W carbon resistors) in a "U" shape. Clip the lead about 1/4" after the bend.
13. Solder this lead to the left side of the front of the other connecting plate.
14. Position both plates back into the S/T #50 case. The plate with the resistor should be against the "+" side of the case.

15. Clip the leads of the resistor and diode about 3/16" from the end of the part. Bend these leads together and solder.
16. Remove the assembly from the S/T case. Pull on both plates so you will need to squeeze the assembly slightly to get it back in the case. The parts create some spring tension.

17. Fill the S/T case about 3/4 full with a non-conductive 2-stage 5 min. epoxy. Allow epoxy to cure.

18. Remount the "rebuilt" rectifier in the 516F-2 and reconnect the leads.

19. Fill remaining 1/4 of case with fiberglass resin (Advance Auto) colored gray (by mixing in about 1/2 tsp. pastel (I got my pastels at Michael's and used sand paper to powderize it).  I mix the resin, hardener and colorant in a plastic teaspoon then poor into the case.  I used a straw to transfer the resin from the can to the spoon and a toothpick to stir.  Two drops of hardener was all that was needed.

20. After the resin cures, replace the protective screen and secure to the top of the power resistors.

Total time for rebuild about 1 hour (not including adhesive waiting times).

As a side note, it might be a good idea to install a small .001uf capacitor @ 250V across the diode to prevent the fast switching from producing RF hash. The original Seleniums did not have this problem. I did not use the capacitor and there was no noticeable increase in noise - probably because this is not a high-current application.