Yves wrote:
"Patrick Turner" a écrit dans le message de news:
...
[ . . . ]
Maybe a triac could also be used.
Patrick Turner.
Oh! Not a such noisy device 
What about a source follower with an HV MOSFET between the rectifier
and the first smoothing cap ?
With a properly adjusted RC to feed to gate, may be made so that HV rise
slowly, at the same rate that toobs start to conduct.
Could also help in filtering or even regulating HV !!
Yves.
The circuit I just posted is for a 35 watt SE amp.
The amp has Si rectifiers throughout.
It sent out a "whoop", or short burst of audio oscillation during warm up, and
I discovered the
whoop noise began just as the 4 output tubes just began to conduct, and then
stopped
after 1/2 a second during the charge up of the cathode caps.
This happened even with the grids all grounded, so it was a phenomena
that generated in the output stage only, which has a CFB winding on the OPT.
I tried all sorts of slowed down screen voltage rise, but nothing helped,
and I am not sure why I got the noise burst, and it could be a string of RF
pulses.
With the heaters on, and able to conduct, the problem went away.
I the 4 screens are fed from a shunt regged 272v supply, with a 3k feed R,
and I increased the C across the zener regs from 100 uF to 470 uF,
and that slowed the screen voltage rise when the HT winding turned on,
and along with the other 701 uF in the B+ supply, the rise in
B+ from HT turn on to +380v takes about 5 seconds, which is slow enough,
and does not generate any unwanted LF transients at the output of the amps
in excess of about 1/3 of a volt.
So I don't need that HV mosfet source follower, although its a nice idea.
My experience has not always been positive with HV regulators, sooner or later
a
voltage spike gets in there somewhere, and poof!, no more reg.
I'd rather use heaps of capacitance, with enough series R to slow a B+ rise,
and have the amp turn its HT off when it gets ill in some way.
Even better is the use of the small slave tranny, which turns off the mains,
but space in the SE35 is a bit cramped.
In my amp, which has a fixed voltage applied to the screens,
It'd be best to have a slow rise HV mosfet SF or HV emitter follower applied to
the screen voltage,
because its the screen voltage which controls the DC flow more than the anode
supply,
and the currents and voltages involved are less with the G2 supply than with
the anode supply,
so a BU208/MJE340 darlington pair would be a nice way to slow the rise of the
anode current.
The warm up oscillation probably won't occur in a PP amp, since its a balanced
circuit.
Its only in this amp I noticed the problem.
Once warmed up, and ready for use the SE35 with 4 x 6CA7 have excellent
stability, with
BW from 16 Hz to 65 kHz at 30 watts, 5 ohms, and able to
power any value of capacitance without any R at the output.
BW is from 7 Hz to 65 kHz 5 ohms at 2vo.
With 4 uF as the load at 2vo, the response rises only 1 dB at 20 kHz.
There is not the slightest sign of any RF instabilities.
The other advantage of a delayed B+ turn on is that
the inrush current to cold heaters does not coincide with
the inrush current to the B+ rails, because of all that capacitance I use.
So the mains fuse value can be smaller, so the amp has better
fuse protection than usual.
Patrick Turner.