[Amps] Amplifier frustrations

[Peter Chadwick]

Colin said:
>So, what is the problem?<

This particular device uses a variable L in the pi network. that unrolls a silver plated phosphor bronze tape from a an insulating drum to big brass drum, and the plate tuning caps are ceramic tubs, rated for a quite a few kVa. The loading cap is a 375pF 3kv variable, with a switching arrangement that adds 250oF at a time up to around 3000pF. As it seems OK on 20, my feeling is that on the LF bands, some of the switched in ceramic caps are playing up under load, and because they take the PA so far off resonance, it doesn't like it. I feel they're underated anyway. It's a bit peculiar - originally designed as a 350 watt output tx using a pair of QY4-125 (very similar to 4-125), the Army wanted a higher power set (700 watts PEP, 900 CW, 500 RTTY - I've got a conference paper describing this) So they fitted a pair of QY4-250, a bigger filament xfmr and a bigger HV xfmr, and bingo! it was done cheaply. They didn't fit any extra cooling for the filament seals, so the socket problem we've all heard of on 3-500Zs appears rapidly in the original. And the ceramic tubular tuning caps get hot, especially at full whack on CW. But it was a cheap modification, so doubtless management approved..........Anyhow, now, trying to tune into a dummy load on the LF bands produces lots of sizzling, the odd spark, and the HV fuse dying, or the bias collapsing and tripping the interlock.
On other funny on the original was the driver - 5894/QQV06-40. That has its anode volts supplied from the PA screen supply, said supply is fused. The 5894 screen is from a different supply, so if you lose the 600volts on its anodes, the screen lights up like light bulb - but not for very long! However, that's not a problem for me.

I'm thinking that with parts from stock, it should be possible to convert it to a conventional pi, if not a pi-L, and by getting rid of the suspect ceramic caps,get some more sense from it.

It worked this morning, although I didn't get the KH8. The EU zoo with the deliberate QRM didn't help, and a spread from 14027 to 14058 seems rather too wide to my mind. Hopefully, there's tomorrow.
73
Peter G3RZP