Date: Sun, 2 Nov 2025 07:42:11 -0500
From: Ron W4BIN <ka4inm@gmail.com>
To: amps@contesting.com
> As a retired broadcast engineer I don't understand the brute force kind
>of power filter used in many power supplies. Most huge amplifiers that
>must not contain a detectable hum component use the "high pass" kind of
>"filter network" DC power filter. Aided with a six phase rectifier an
>oil filled 4.?Fd capacitor (at 50. k Volts) is all that is used.
>At 3.6 k Volts (@ ?.7 Ampere) two 10. ?Fd at 4. k Volts with an inductor
>(filter choke) between them resulted in no detectable hum in the 1 kW
>(peak of sync) NTSC output using oil filled capacitors.
> The important aspect of this kind of power supply is when there is a
>tube arc there is not enough energy stored to cause permanent tube damage.
> A 2. kW Tetrode cost over $4,000. (UHF TV translators)
> Ron W4BIN
It's a actually a low pass filter network that is used, with a defined
lower cut off freq. LC filters attenuate at the rate of 12 db per octave.
RC filters attenuate at the rate of just 6 db per octave. Too much V
drop across the RC setups, depending on current drawn.
Don't confuse harmonics of the main ripple freq ( 120 hz if on single
phase in NA) with octaves.
The raw DC from a FWB is 67% on 60 hz single phase.
The raw DC from a FWB is just 5.2% on 60 hz 3 phase.
That's a HUGE difference in favour of single phase. The single phase FWB
uses 4 x legs of diodes. The 3 phase FWB uses 6 x legs of diodes.
Ripple freq on single phase 60 hz is 2F or 120 hz.
Ripple freq on 3 phase 60 hz is 6F or 360 hz.
The caps get charged up or 'topped up' 60 times per second on
single phase... but 360 times per second on 3 phase. That's a massive
difference.
With 3 phase, the raw DC is very little, so WAY less filtering required.
On 3 phase, the caps are being topped up 3 x times more often.... hence
less filter C required, like only a 1/3 required.
I modelled your 3.6 kv ( @ .7 amp) supply ( assuming it's 60 hz single
phase) and with your 10 uf - choke - 10 uf C-L-C config, and it works out
to -40db if 1H used for the choke...and -55 db if 4 H used for the choke.
I can achieve -40.5 db on my drake Supply (2.5 kv loaded to 800 ma) with
just 2 x harbach boards used in parallel. Each board has 8 x tiny 330 uf
lytics. And just -31 db if one board is used. -31 db is
unacceptable imo....( 2.9% ripple). I can clearly see the ripple on any
monitor scope, driving me nuts. Ripple % is inversely proportional to
filter C. Double the filter C, and ripple is now in half....down 6 db.
This is voltage and current, so it's 20 log (rms ripple V / loaded B+)
For SSB, we want excellent dynamic regulation. With your FM/TV
broadcasting, everything is in a steady state, so V regulation is a moot
point. Just tweak the taps on the plate xfmr for aprx required, loaded
B+.
For SSB use, and good dynamic regulation, and low ripple, we need a low
esr plate xfmr, and plenty of filter C. Chokes are getting to be non
existent these days, esp big ones required for this job. (The 4 H @ 3 amp
CCS dahl choke I have is 127 lbs...and uses the same 10 kva hypersil core
as the Dahl 10 kva plate xfmrs). Oil caps are also getting to be non
existent.
Both oil caps and chokes are heavy items, which makes shipping costs
prohibitive.
And 3 phase power is NOT an option for residential use here in NA. I
have caseloads of those 2500 uf lytics....and also 2200uf. They weighed
nothing, bought NIB surplus, so shipped him 20 of em.
Your megabuck tetodes for TV use is a different ballgame. Typ those
supplies will incorporate a high speed (microseconds) energy diverter
cro-bar circuit, typ from Eimac's AB - #17
" They employ large high speed diverting elements such as thyratrons,
ignitrons, triggered spark gaps, or in some cases, a high-speed vacuum
switch which may be required to handle thousands of amperes peak current."
With our simple high MU GG triodes, and ample 225 watt grid, it's less
of a concern. 7500 vdc / 50 ohm glitch resistor = 150 amps of fault
current. It's 150 amps of fault current regardless of filter C. 150 amps
of fault current will open off the BUSS HVU-3 fast HV fuse in 1-2 msecs
everytime. (tested several times)... event over.
Jim VE7RF
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|