[AMRadio] dynamic load on power supplies

John Lyles jtml at losalamos.com
Fri Feb 16 11:10:44 EST 2007

I agree, nice summary Don on your 1T experience. The dynamic power supply regulation is a problem to contend with, for CW. Jim's suggestion of a resonant input L has merit but is very tricky to implement. 25 years ago I designed this into a single phase line of commercial transmitters which sold to FM broadcasters, the BE FM-1.5A, FM-3.5A. When dealing with big plate supplies doing this, one must use polypropylene dielectric capacitors, not oil, if you want them to survive. The peak AC voltage at 120 Hz across that capacitor (its like a parallel trap) is high, and so dielectric heating and corona are both problems. Paper/Oil caps can burst and leak, worse yet, cause a fire. In an FM transmitter it was easy, due to constant load. It was mainly done to reduce the stored energy in the power supply, and to keep the HV meter from pegging the pointer, when the exciter load was unplugged or drive removed. It made the plate voltage supply act like it had the critical value of input
inductance to keep constant current (instead of spikey cap charging current) through the rectifiers. I chose the cap first, and having a custom choke designed to resonant with that at 120 Hz, only when the load was removed (the L value with no DC in it). Capacitors are readily available for this, from microwave oven power supplies, used in the HT supply. I used a pair of 0.97 uF 2500 VAC caps in series to get ~0.485 uF and the caps were cheap ($10) so we could select them from large order. 

So the choke needs to be well characterized, L known over a spread of DC current that will be seen in the transmitter, so that you can determine where it will run into resonance. That is measured using a bridge, and DC bias current through it. The Q of the circuit needs to be small (which it is) so that it is a smooth transition, and to keep the circulating current down, during resonance.

Finally, the caveat of this is that it only filters 120 Hz with a notch when it is in resonance. When it is loaded heavily, the L falls, the resonant frequency rises > 120, and you have virtually no filter on your power supply unless you go back to a big capacitor again. So a resonant L system must either have a large shunt C afterwards, or a second LC filter section to low pass filter the higher harmonics of powerline, 240, 360 etc. Otherwise you can hear a whine from the power supply noise. 

Terman, in Radio Engineering, mentioned the resonant choke scheme, and so does the 1964 Collins Radio SSB book by Pappenfuss. However, for AM with such varied loading, it might be more trouble than its worth. Anyone want to try it out and be a test bed? 

On the AM Window pages, where the technical tutorials are located, I posted a copy of a GE Ham News article from many moon ago, on dynamic regulation of power supplies for AM, CW and SSB. Look for it there. 



> Message: 3
> Date: Thu, 15 Feb 2007 19:33:07 -0600
> From: "Jim candela" <jcandela at prodigy.net>
> Subject: RE: [AMRadio] Re: Gates BC1-T operational again
> To: "Discussion of AM Radio in the Amateur Service"
> 	<amradio at mailman.qth.net>
> Message-ID: <NHBBLOLJILJMABPBLLFLKEIKEIAA.jcandela at prodigy.net>
> Content-Type: text/plain;	charset="iso-8859-1"
> -----Original Message-----
> From:D. Chester
> To: amradio at mailman.qth.net
> Subject: [AMRadio] Re: Gates BC1-T operational again
> What I did to make the BC1-T operational on CW would apply to many other
> broadcast transmitters.
> Don k4kyv
> Hey Don,
>    This is a great story, and a labor of love. Thanks for sharing this with
> the group.
> One passing remark I wondered about while reading this essay is whether it
> was feasible to stick with the stock 10 Henry choke, and resonate it to 120
> hz with a suitable oil type capacitor with a value somewhere around 0.15 uf
> @ 5 KV. (10H || with 0.15 uf = ~ 130 Hz resonance). I have limited
> experience with this approach, but the one time I tried it, it seemed that
> the critical inductance need dropped leaving everything else constant, and
> the AC ripple coming out of the filter was far less than a normal choke
> input filter. I did not do dynamic load variation tests however, so who
> knows..
> Great Story Don!
> Jim
> --

More information about the AMRadio mailing list

This page last updated 19 Oct 2017.