Can I make a joke along the lines of: if it's Federal, it's probably not good for you? I guess I just did. On that note, did you know that the archaic form of the word federal is foederal? I suppose that label was deemed a bit too transparent (cue laugh track). But, I digress, I'm not interested in talking about commonly accepted popularity contests or congregations of self important assbags, I'm more interested in rectifiers at the moment.
Here's a close-up of the selenium sandwich stack in a Navy ME-6D/U electronic multimeter (which will likely be covered in a future installation). Ease of identification is selenium's greatest attribute, it can typically be spotted in well under a minute of having cracked the case. The component pictured above is a less common version without the oversize heatsink fins. My philosophy is to change selenium out when I see it. You can read about the failure modes elsewhere on the net, here's a fine write-up: http://yarchive.net/electr/selenium_rectifiers.html. This is the rectifier block in schematic:
And, extrapolating from that, the electrical orientation of diodes in the stack:
In terms of out with the old and in with the new, this cannot be easier to wire; the main concern is going to be introducing some sort of framework that will avoid a free form rat's nest sculpture that's rectifying the B+ voltage (I'll be sure and photograph this section when I post an update on this unit).
In terms of electrical characteristics, selenium and silicon are not identical, and depending on the installation, a low value high wattage resistor might be beneficial in terms of nudging your B+ down a bit. Also keep in mind that average wall voltage has crept up over the past 50 years (I think evidence of selenium in the power supply typically dates the unit to that age at least) so the combination of higher AC mains and reduced internal resistance at the power supply might place undue stress on the circuit. A bucking transformer at the power cord is generally a good idea (even with tube recitification). http://www.philcoradio.com/phpBB2/viewtopic.php?f=11&t=2247
Regarding the selenium pull jobs, I have the sick notion of dropping them into ring modulators. Owing to the greater forward voltage drop (read as lower efficiency) I expect I'll need to build boosters on all three legs of the circuit. We'll see what shakes loose on that front.