Relay computerswhy so few 15

Hmm, interesting. Of course, it was a dirt cheap tube at the time.

Oh, yes. They faded with use. Something like a 2D21 or 2050 wouldn't fade as much, but would be bound by the need for the plate voltage to coss near or below the cathode, to shut off, though IIRC, a heavy blast of negative to the grid would cut them too, ultimately. Realistically, this demanded AC on the plate, but for low speed relay-solinoid driving, it wasn't bad, and even saved you the cost of a 5U4 (or 10:) for HVDC...

I never saw a 25L6 run relays. I *did* see, and I'm STILL amazed by it, a Seeburg 33 3-1 rpm auto speed unit for a Q100-160 once. This gem was basiclly a sine wave type of VFD (yep - no PWM), and drove a 1-100th hp Bodine induction motor. Basically, a low frequency oscillator driving an audio amp driving a motor. Impressively, this *worked*. The machine would load on 60hz, and once the mech kicked into play, it'd cut over to the VFD, which would run the motor somewhat slower to play a 7" 33 rpm record. It'd stay on 60hz line power for normal 45rpm operation. Even slicker, it could detect a failure and trip out a 33 immediately, instead of grinding to a halt and needing a manual reset.

Not surprisingly, the thing wasn't reliable, and an SCR based version came shortly after (I think the LPCs used that setup to get 45, and ran at 33 normally). The 7" 33 rpm craze in jukeboxes died by the early 60's anyway...

Wurlitzer, AMI, and Rock-Ola used a mechanical speed shifter, which was probbably just as unreliable....