One thing I would NOT do with the single-pass-transistor power supplies is set the voltage lower to make the Icom just a tad bit cooler. Why?
That single transistor (basically acting as an electronically controlled resistor), is dumping even more current, getting even hotter than when at the nominal 13.8v. If you wanted to make the transistor run cooler, you'd actually UP the voltage to say 14.0v or so.
So chicken-and-egg dilemna. The problem is that if you DO lose that pass transistor due to running it hotter at a lower regulated voltage, with the smaller supplies having no "crowbar" to protect against the unregulated 18-24v now passing, you could smoke the 8600. And if your Icom is living in a shack over 77F temperature, getting even more closer.
So the first option being the RS-7A with the crowbar - great, at least if you lose the pass transistor, hopefully the crowbar will catch it. But it's still a single transistor unit.
So I guess if one is going to drop voltages, you'd want to go with a dual-transistor setup to dissipate more heat. Although it seems overkill for most, I see why N9EWO likes the 12A model. A possible equivalent to that might be the Tripp-Lite PR15 (the first in the lineup to have the crowbar).
I can see why this might be mind-numbing overkill, but I suppose if one wants to play it so safe with so much riding on the end of the supply, it might make sense.
What can also kill pass transistors? Cleaning up the shack's rat nest of wiring, and scraping a coax shell across the transistor case and the heat sink. Yeah, did that. Interestingly I see the Tripp-Lite PR15 seems to have the transistors mounted with them inside, but still dissipating heat through a rear sink. First time I've seen that. Maybe too many people cleaning up their shack and shorting the transistors.
