[Amps] Tuned choke input filter for PS

[John Lyles]

Three phase power supplies can get away with smaller L and C, higher 
ripple frequency too. They are great for high power RF transmitters.
But most hams and a lot of small broadcasters don't have three phase AC 
service.

Single phase high voltage power suplies can benefit from having enough L 
to sustain constant current in the inductor instead of discontinuous 
pulses charging the capacitor. By adding enough load resistance (lower 
bleeder R) lower value inductors can be made to work. So the trade off 
is power consumption/dissipation in this resistor. Selecting a good 
inductor ('choke') also requires consideration of the DC resistance, as 
this can cause poor regulation as well.

L/C filters suffer from a low frequency resonance, which causes poor 
transient response during turn on/off  (overshoot) and poor regulation 
during keying or low freq audio modulation.This is covered in the 
earlier (1964) Pappenfuss/Collins Radio SSB book, and maybe even in the 
Terman textbooks. But L/C filters can get by with rectifiers that have 
lower peak to average current ratings, such as mercury vapor tubes.

Tuned choke filters work well, if carefully designed. Components need to 
be rated for the high 120 Hz voltage that will be across the 'trap' and 
the inductor needs to be designed to be the desired L at some DC 
current. This is not a swinging choke as was already mentioned here. But 
it does need to 'swing' enough that it will go into resonance with the 
tuning-capacitor at the desired load, typically when the power supply is 
only loaded by the bleeder resistor. Then, when current goes up with RF 
power, the resonant choke goes out of resonance and becomes just a choke 
input filter.

It is true that the higher frequency ripple components will come through 
the filter with less attenuation, unless a second section of L and C are 
added. This is because the value of L required for a resonant choke is 
much smaller than needed for straight L/C filter.

I spent months developing one of the last commercial tuned choke power 
supplies, in 1982 or 83 for Broadcast Electronics Inc. for their single 
phase FM transmitters up to 3.7 kW. Tube FM amplifiers are easy, being 
class C, with uniform current during operation. But during grid tuning, 
and if the driver is switched off, the plate current drops to nothing. 
In the 'olde' days the driver was typically a 4CX250B stage, and it was 
hard-connected to the final. Around 1980, FM drivers starting to switch 
to solid state, 50 ohm stages that had a coaxial interconnect. It became 
possible to disconnect the driver and test it into a load. With a 
conventional L/C filter, the power supply typically soared to a much 
higher voltage. Designers didn't want to compromise the design with very 
expensive 20 H chokes or low value bleeders that dissipated hundreds of 
watts. Oil/paper capacitors would need to be rated at an even higher 
voltage compared to normal operating voltage, like 1.4X. Enter the tuned 
choke.... It worked very well in this application, the HV supply would 
not soar anymore when unloaded, and the L/C components were relatively 
small. I kept detailed notes of the experiments at BE, and later wrote 
up my experiences here, can be found in the amps at contesting.com archives 
in several places. It never had dangerous explosions or disasters, 
because we chose components that could handle the voltage in the tuned 
trap. Picking off the shelf L or C and trying to force it to work in 
this topology is a compromise that can end in failure. I started this 
way, and quickly realized that I would need a custom choke as well as a 
custom capacitor. Probably a thousand or more of these rigs were sold to 
broadcasters, and the only failure mode I ever heard was when they quit 
regulating, starting to soar again when unloaded. This appeared to be 
from capacitance changing, long term drift or an internal change.I never 
found out what caused that, but replacement capacitor usually fixed it. 
Or maybe it was wandering of the power line frequency at night? One more 
thing, the choke had a tap on it for 60 and for 50 Hz operation, as the 
transmitters were sold internationally.

SSB amplifiers, by nature, have idling current in the PA, even without 
drive, so the choke size can be selected along with bleeder to keep the 
voltage from soaring. Tuned chokes may be unnecessary.

Nowadays, with silicon rectifiers that are able to handle huge inrush 
currents and charging peak currents, the trend is to go to larger 
capacitors alone, adding the complication of inrush limiting, stored 
energy precautions and needing to have transformer iron designed for 
C-input configuration. These filters have eliminated the leak-through 
rippled from powerline harmonics, and since the power supply voltage is 
set by peak charging voltage on the capacitor, will give higher voltage 
and pretty decent load/noload regulation without needing a big bleeder 
to accommodate the lightly loaded condition.

73
John
K5PRO