[Amps] resonant filter tspa

[John T. M. Lyles]

This was all discussed here in the late 1990s, 
(see archive) but i will repeat as there are 
probably a few different hams reading and 
contributing since then.

Resonant elements in power supply filters were 
discussed in Terman's Radio Engineers Handbook. 
The point of using them was to allow using a 
higher bleeder resistor value (less heat), a 
smaller choke, and still prevent soaring when the 
power supply was unloaded. Normally the resonance 
only gets close at the low or zero current 
inductance, because at any load current, 
including zero signal idling current, the power 
supply has enough load.

Resonant choke HV filter circuit was used in the 
Harris HFL1000 amplifier, the Galaxy 300 SSB 
transceiver, the BE FM1.5A and FM3.5A broadcast 
transmitters, and probably a few other 
transmitters with high voltage supplies.

The Harris used 7 Hy in parallel with 0.25 uF. A 
single 6 uF shunt C followed this, and a 100k 
bleeder. It supplied 2.7 kVDC.

The Galaxy had 4 Hy in parallel with 0.1 uF, 
followed by a shunt string of three 
electrolytics, 200 uF each in series. They had a 
15 k bleeder across each, and it supplied 800 VDC.

The BE used a 5.06/3.5 Hy choke in parallel with 
a pair of series connected 0.97 uF 2500 VAC 
capacitors. This was in a 2 section filter, first 
a shunt 4 uF capacitor, then a series 3.5 Hy and 
a shunt 4 uF. Bleeder was 200 k. It supplied 4100 
VDC at up to an amp. The chokes were both in the 
negative lead.

I designed it for the BE transmitters in 1982. I 
tested many junkbox chokes using a setup with an 
audio oscillator, a DC supply, a General Radio 
1650 bridge, and an audio voltmeter. Feeding the 
choke in series with a decade box, I dialed in 
the value of R which caused the voltage division 
to be 1/2. This gave me the 'impedance' that the 
choke exhibited at the audio frequency, which 
resulted in XL. These days we can use a small 
meter. I also tried to bias the chokes with an 
Amp of DC, to measure how much they changed under 
current. At the time BE had a great collection of 
old chokes, so i tested a 1.9, 6.8, another 6.8 
and a 6 Hy choke, each with a 0.27 uF capacitor 
in parallel, to determine what sort of Q each 
resonant circuit had.

Several things became apparent to me, to build it 
at such high voltages. The capacitors could not 
be standard paper/oil similar to what Plastic 
Capacitors and many others make. I used 
polypropylene film dielectric capacitors in oil, 
lower loss at high frequencies. These were made 
for use in microwave oven (MO) power supplies, 
for the HV supply. I found hundreds of these at 
one company, for an excellent price, so we bought 
a large batch and tested all to get an average 
value of 0.97 uF. With two in series, or 0.48 uF, 
it worked with the choke i designed. The voltage 
rating of MO power capacitors was 2500 VAC, not 
enough for this 4.1 kV DC supply. So with two in 
series, it was adequate. I measured the voltage 
across the resonantor differentially, and found 
that at full load, the ripple viktage was 6 kV 
p-p. I still have a photo of the waveform. Not 
sinusoidal. At zero load, the ripple became a 
nice 120 Hz sinusoid, with 4 kV p-p.

For the choke, I had SNC transformers in Oshkosh 
design a custom unit with the two taps, one for 
US and one for 50 Hz power overseas. I sent them 
capacitors to use and they individually tested 
the chokes with them until they got it right. The 
late Carl Seivers at SNC did the work, he was a 
great man to work with and knew his stuff. At SNC 
he had an old guy who had designed a resonant 
choke circuit for some ham rig, maybe it was a 
prototype Heathkit or something, i remember the 
discussion about that.

BE used the same choke in the smaller 
transmitter, with a different power transformer 
for lower voltage.

I compared ripple, inrush current, transient 
response, and other measurements, and concluded 
that it was a decent circuit for a class C FM 
transmitter. Of couse, you would wonder why do it 
at all in a constant power transmitter? The 
reason was that the exciter can and would be 
removed, turned down, leading to no current, as 
the class C tube would be cutoff. At this point, 
the plate meters would peg unless we used a very 
huge L and a very hot bleeder. Stored energy in 
the power supply was also considerably less than 
the alternative 20 uF C version without the 
resonator. However, one thing to remember is that 
the resonator only notches out 120 Hz, and higher 
frequency harmonics of the power line, like 240, 
360....720 Hz are  still present. This was the 
reason for the second L and C, as a smoothing 
filter section, the transmitter had exceptionally 
low AM noise from ripple on the HV.

I never had them blow up, and it was a well 
behaved circuit. However, I would also recommend 
that hams consider other options, like a hefty C 
input filter with step start and high surge rated 
diodes these days, since the transformers are 
typically designed for that. As others have said, 
you can get some high circulating currents in the 
LC trap at 120 Hz, and this leads to capacitor 
failure if you don't account for it. Oil and 
paper caps will eventually fail if not careful.

73
John
K5PRO

>
>Message: 3
>Date: Fri, 11 May 2007 13:21:12 -0400
>From: "Jim Tonne" <tonne at comcast.net>
>Subject: Re: [Amps] Power factor and choke vs resonant-choke input
>	supplies
>To: <rbonner at qro.com>, "'Amps Amps'" <amps at contesting.com>
>Message-ID: <002101c793f0$c75ccb90$2602a8c0 at jim177093b3dd9>
>Content-Type: text/plain; format=flowed; charset="iso-8859-1";
>	reply-type=original
>
>
>This resonant-choke discussion is crying for someone
>to do a Spice analysis so we can see the various
>waveforms. 
>
>If someone will give me a schematic with relevant
>values I'll volunteer to give it a shot.