Sunday, March 27, 2011


It's surprising how quickly you get into trouble, fiddling with power supply designs. Duncan's PSU2 software does help keep the worst mistakes inside the computer instead of on the test bench, but there's no substitute for the real thing.

Fred Sutter's original "QST-40" power supply is a nice study in empirical design. He used a choke-input supply for stability under varying loads and a fairly low-resistance bleeder resistor to keep the B+ from soaring between Morse elements. The power transformer is one a contemporary ham could have readily salvaged and the rest of the parts were pretty stock -- in 1938.

Today, we have fewer choices. Specs are more stringent, too; there's usually more safety margin for a given current rating and modern measurements of choke values are less optimistic. Hammond transformer has probably got the widest range and availability of any outfit still making tube-suitable iron, so I've worked from their catalog. It's good news and bad news; the plate transformer will run $80 to $100 and the filter choke is another $46 to $50, but you can get them;* the rest of the parts are quite reasonable.

Here's the untested, not-yet-built power supply so far:(Click on it for a large version) There are a couple of options for the transformer and different filter circuits for each one -- either a 279X (850VCT at 170 mA) using a choke-input filter (leave out the 8 uF filter condenser!) or a 273BX (700 VCT, 175 mA), with a pi-section filter. As I have noted on the drawing, you really don't want to try the pi-section filter with the higher-voltage transformer.

5U4 or 5Z3 is your choice; the former is an octal tube, the latter 4-pin but they're otherwise the same. 83V is another possibility.

I also detail two options for the bleeder resistor, which additionally supplies lower B+ voltage to the screen of the 6L6.

The biggest change is in how the B+ is turned on and off; the five and dime no longer sells nice ceramic-based knife switches (more's the pity) and even if they did, it's not a very user-friendly means to that end. So instead, I added separate filament transformers for the 5u4/5Z3 rectifier and 6L6. This results in two good things and one "gotcha:"

1. Now you can turn the B+ on and off by switching the primary voltage; you can even do fancy pilot lights for filament and plate voltage, just like the big rigs!

2. With no heater voltage being drawn, the plate transformer's got a little oomph to spare. (If it has too much, Hammond provides primary taps for 115 and 125V mains).

3. Mind that heater transformer for the rectifier! It's only providing 5V to the tube but it's got the full B+ voltage on the secondary. Hammond hi-pots them at 2kV and it uses "two-bobbin" construction, with primary and secondary on their own plastic spools, but fabricat emptor caveat fabor, or however Cicero would've said it.

The output connector is notional at best; you can still find the old Amphenol plugs and sockets at places like Radio Daze and hamfests. A terminal strip would work as well -- just keep your pets and extremities clear!

I'll close by repeating that this circuit has not been built and tested. While I have done my best to predict how it might behave, I cannot verify its safety or suitability; proceed at your own risk. (On first power-up, I'd want a big old 100W or 60W incandescent in series with the hot side of the incoming power, and I'd use a GFI-switched outlet strip to turn it off and on. I'm not much for excitement.)
* Angela Instruments, listed under "Sources" in the sidebar, seems to have the best prices at this writing. Antique Electronic Supply, Radio Daze and Parts Express all stock Hammond transformers.


  1. Ah yes, good old leaky heater winding insulation: Gave me my surprise off of a Simplex (nee Western Electric) theater amplifier by putting the chassis (and, more particularly, the handle of the power toggle switch) at c.a. 1kv. Smarted some.

    (Still have a couple of those somewhere in the basement...)

  2. Youch! Grouch, There's reason those are pronounced Kill-a-Volts!

    I'm playing with some room heaters right now.

  3. I just about made my head explode understanding how the B+ voltage gets to the filter when the rectifier cathode is hooked up to 6V, but I might have it now.

    The two sides of the transformer secondary going to the rectifier heater at 6V relative to each other, but relative to ground they are 425V. But since the filter is tapped into the middle of the rectifier's heater secondary, it sees 425V +/-3V. Since the rectifier is full wave, the filter sees a 428V half-wave, then a 422V half-wave, then a 428V half-wave, and so on. But any filter that can turn a 425V half-wave into DC shrugs off 3V either way, so the variation doesn't matter.

    How am I doing?

    What makes a knife switch unfriendly? Exposed high voltage?

  4. Exposed high voltage -- and proximity of your fingers to it.

    He's using it A) to keep the 83 happier (it's a mercury-vapor rectifier, though more forgiving that most, and they don't like seeing HV until heated up) and B) as a safety feature, so a (brave!) op can change bands without shutting all the way down. I'd advise killing the primary power.

    ...I'm rethinking my design, as there are filter-cap charging-current issues that would be mitigated by allowing the non-mercury 5YU4/5Z3 rectifier to heat up with B+ applied.

    Your analysis is okay except -- that 5V heater AC on the rectifier doesn't show up on the rectifier output: there's only one point of connection between the two circuits: it's a trick! You can hook the smoothing filter to either side of the heater instead of the center-tap; heater emission will be a little better distributed using the tap, is all.

    This makes more sense if you redraw it as a couple of solid-state diodes and a pilot light for the heater.

  5. There is a comment from Chuck K that's not showing up and it darned well should:

    "I know why the HV is switched - the 83 is a mercury vapor rectifier, and you don't want to apply HV to the plates until the tube is warmed up.

    I've got a homebrew rack of push-pull amplifiers I got from a silent key in Elgin, IL years ago - he worked for WGN as an engineer. The rack is nice and tight, with RF gasketing and copper screen over ventilation openings - but the power supply was transformers, two 866 tubes, caps, and filter chokes sitting on a metal rack in the room behind his shack, all wired together with 12 gauge solid copper wire. Talk about exposed HV! I've got grandkids running around here - so can't do it that way. One of these days, I'll try and get that back up.

    I've got some Amperite delay relays in glass packages somewhere - put one of those in series with the HV transformer primary and power the heater from the filament transformer - now you don't have to worry about sequencing the power."

    (In re the 83, I'll note the 83V is a non-mercury substitute. Current/voltage ratings slightly less than 5U4/5Z3).

  6. Somebody (probably Arrow-Hart or Hubbell) once made a round-and-round rotary switch especially for this application. Wiring was:

    1 - Off
    2 - 1-2 (Filaments)
    3 - 1-2 + 3-4 (Plate)

    The one I saw would only rotate one way (couldn't back up) and had some kind of mechanical limiter so you had to pause and release tension in each position.

  7. Sure, a progressive switch -- I like the "full stop" feature, though.

  8. Having spent a few times around with that stuff.

    A .005uF/600V from line to ground and neutral to ground.

    A MOV of suitable size on the primary side to protect the HV transformer from transients that may break down the inculation.

    The 5y4/5u4/5ar4 would have a the fils warm before ac applied as a happier thing. The filliment supply should be on the same switch as the HV for that reason.

    Having the 6.3/12.6Vac line on the primary power on is good as many radios (VFOs) will settle down better then longer they warm up. That and better life from tubes is had when not thermal cycleing heaters.

    The old tube hands (I was) do know the hardware and how to make it sing. I know from doing old Land mobile and repeaters plus the usual commercail radio.


  9. One last item, Those caps should be Epeak+10-20% where Epeak is 1.414 * Vrms (425*1.414-596.7) and despite voltage drop across the tube when lightly loaded your above 550V. Minimally I'd use a pair of caps at 2x the value in series with a bleader string across them of 56K@3W per cap. For caps I'd use 450V units of 16uf and 47/50uf. When is series with resistive balancing across them the voltages is controlled and you have multiple bleeders should any one fail. When building a cap bank like that I prefer to mount them between two peices of leaxan/lucite/G10 fiberglass as a finger shield
    and blast shield should a cap fail. Also spacing should be wide as even a small spider web can cause a nasty arc.

    Having be hit by flying bits and also 1800V in my life safety is not taken lightly.

    Enjoy! Those glowgugs are fun and produce excellent radios and transmitters. While I design with the latest and greatest I do run a selection of older Heath, Hallicrafters, Tempo-One and HB firebottles on HF.