[Amps] Henry 2k-4 HV inductor - Tuned choke input filter

[John Lyles]

Hi Jim, I see you remembered this! I learned of the tuned choke input 
idea from the 1964 Collins Radio SSB book by Pappenfus, and also looking 
at the Henry circuit, a 1950s edition of Terman Radio Engineering, one 
of the Galaxy transceivers, and a few other commercial circuits that 
used it. I had also talked to a designer at SNC transformers in Oshkosh 
who knew the idea and was challenged and interested in making the 
chokes. I had liberty to do it at Broadcast Electronics, as we could try 
ideas that weren't conventional;we were trying to introduce products 
that would change the industry at the time. The standard for FM 
broadcast rigs was an L-C-L-C choke for single phase and just an L-C for 
three phase power supplies. With FM being a constant power mode, it made 
sense. The hitch was that if the exciter drive was removed, with a class 
C final, then the load shifted to near zero current. And then, any 
reasonable L input filter would cause the HV to soar tremendously. 
Typical response was for the HV meter to peg the scale and scare the 
engineers! FCC R&R had strong mandates that the final voltage and 
current meters operate in the upper 2/3 of scale, not at half scale, for 
the rated output of the transmitter. So when voltage soared because of 
not enough load, the meter went above 100%. The only way around it was 
to put in a huge bleeder that wasted hundreds of watts, otherwise. This 
was to get enough load that a reasonable choke like 10 Hy would meet the 
critical inductance criteria.

At the time (1980s) no one would have put a huge capacitor filter in the 
transmitter, due to cost and space. Voltage for a single phase 3500 watt 
transmitter was 4500 VDC. Series electrolytics did not meet the long 
term 24/7 reliability goals. With a resonant choke, this was simplified. 
It didn't need to do anything most of the time, and it was only going 
into resonance with the parallel cap at zero load. That prevented the 
voltage from soaring and pegging the HV meter when it happened. FDesign 
of the choke then was simplified, as it had to have the exact value of 
L, measured with low DC current. When loaded with more current, choke 
inductance drops. Resonant freq rises above 120 Hz. (or 100 Hz for the 
overseas model tap on the choke). And the voltage stays reasonable as 
the load current is high enough to ensure critical inductance in the 
filter.

As for exploding capacitors, I found quickly that the normal oil filled 
paper/mylar filter capacitors were not appropriate for a resonant choke 
circuit. I measured the peak voltage across the cap when it was resonant 
and when it was off resonance, and indeed it was a large swing. Hence I 
went to two HV caps in series with plenty of margin. It helped that my 
cap manufacturer had a lot of microwave oven caps made with 
polypropylene and low loss tangent, for a very low cost. Rarely if ever 
did one pop. I designed it so that with no current drawn, it ran right 
at resonance but never on the low side.

Many years later, (in the past decade) I have talked to some of the 
owners of those transmitters still running them. They didn't understand 
the circuit, ended up having a cap leaking oil eventually, and just left 
them out or worse. So the thing had lousy regulation no load to load. 
Thats the worst case problem once the cap fails. The chokes didn't fail, 
SNC did a great job of building them for this circuit.

One other point, I learned that the resonant choke doesn't provide 
adequate filtering for harmonics of the rectification frequency. 240, 
360, 480, etc. They just roar through the L as it has a shunt C 
bypassing it for the harmonics. So a second L-C section had to be added 
to the design, for the higher order filtering. It was very small, not 10 
Hy or anything like that.

All in all, it was an interesting project, but I would never do it again 
today, well who would use a tube for FM anymore? It is probably not an 
appropriate design for SSB or AM where the load is varying significantly 
all the time.

73
John
K5PRO


From: "Jim Thomson" <jim.thom at telus.net>
To: <amps at contesting.com>
Subject: [Amps] Henry 2k-4 HV inductor

<   On 12/13/2017 01:15 PM, Jim Thomson wrote:

 > ##  Although the  tuned choke concept works, it really is a throwback 
to the
 > 1950s.  IF the choke and parallel resonating cap just happen to 
resonate at 120 hz,
 > the peak V across that parallel tuned choke will skyrocket, and both 
the cap and choke
 > will explode.   Typ the choke is resonated just a bit higher than 120 
hz, like 123 to 124 hz.
 > When u start sucking loads of plate current, the inductance of the 
choke will DECREASE a bit,
 > and the resonance of the choke + resonating cap combo will  INCREASE 
some more, like now
 > up to 124-130 hz.
   /* snip */

##  Nope, the small resonating cap is wired  directly in parallel with 
the choke.  That forms a
parallel tuned circuit, killing the 120 hz component. Any cap after 
that, that is wired from B+  to
B- just kills any residual 120 hz components, plus harmonics of 120 hz. 
HV filter caps are not wired
from B+ to chassis, they are wired directly between the B+ and B-.   Any 
HV meter is also wired
directly between B+ and  B-.  Old ARRL handbooks will depict HV meters 
wired between  B+  and chassis,
which is incorrect..and more fubar.

##  John lyles designed a HV supply for a FM broadcast TX.  For the 
resonating cap, he used the
.9 uf at 5 kv rated small oil cap  you will typ see in any microwave 
oven.    He used 2 of em in series,
so  .45 uf at 10 kv.   Both caps sent to the choke maker..who builds the 
resonant choke around the
.45 uf combo.   On any single phase setup, the component you are trying 
to get rid of is always 2F,
or in north america, 2 x 60 hz  = 120 hz.    In japan, UK, etc, its  2 x 
50 hz  = 100 hz.   For 3 phase
setups, it now becomes 6 F, or  300 hz in the UK and  360 hz  in north 
america.

##  resonate the choke at 120 hz, and the caps will explode every time, 
it has to resonate just above
120 hz.   Ask anybody who has tried it.