Golborne Vintage Radio

The simplest idea is an FET shunt regulator on a heatsink in a perforated insulated box, Maybe with a 12V fan.
The series ballast will use tungsten lamps. The only universally available kind are 26W for ovens, thus rated for up to 360 °C. There may be some old stock regular halogen lamps, or 60W red ones for fireglow effect where the heat makes a disk spin. I do have some 28W halogen cooker hood lamps.

The best way to test really high power valves is in the actual equipment, so I will go with 26W screw in oven lamps.

The most available HT transformer is the 110-120/220-240 isolated shaver outlets. These have a usually a shunt to limit power to under 20W and have good isolation. There is usually a themistor which can be removed. Obviously people can source other transformers or use two transformers back-to-back.

Two cheap FETs used in 500W SMPSUs can be used. They are 1nF to 5nF gate capacitance, but a 10K in series close to the gate is a good idea. A separate +9V (approx, I'll check) supply via a 100K pot will set gate bias to vary shunt current and thus HT after the series lamp(s).

Likely a number of 26W lamps are used in series, or possibly parallel, depending on selection of 150V, 300V or 600V ranges and load. Ideally to test valves you have two duplicate HT circuits.

The 1N4007s will give a nominal +160, +320 and +640V HT with no load. Feeds via 1M, 2M and 4M approx can be used for leakage tests. It can also give a nominal -160. A reversed 1N4148 can regulate to about -110V to -120V assuming a series resistor for under 200uA. A load of a 2M2 Ohm pot is less than 60uA. Perhaps a 1M or even 470K pot is feasible.

The supply for fan and FET bias can be a 9V or 12V transformer, giving between -12V or -24V for grid bias (9V) or about -16 or -32V (12V AC). So zeners for 9.1, and 22V or 12V and 28V could be added.

The 470K to 2M pot(s) can be thus switched to 3 different negative supplies.

Cheap DMMs or panel meters used to display the manually adjusted voltages, and optionally (ideally also) load currents.

There is no explicit HT regulation. Just measure it and adjust the FET bias knob. The negative supply could have the pot calibrated for situations where no or little grid current flows.

Next I will lash up a "breadboard" (a case where a literal one may suit) as I have a good few lamps, FETs, pots, caps, rectifiers, zeners, diodes etc. I'd not trust heatsink insulators at these voltages so the entire of the pair of heatsinks will be live. I've loads of suitable ones off PII style CPUs. Only up to 20W needing cooled, as worst case is HT max / 2 and no external load.

My approach has been to use MPSA42 and PHE13009 bipolar transistors in darlington format as a fairly simple series regulator. I have also used similar for active smoothing. In fact the PSU I have just built uses this for the 125V regulation although I have used them for well above that in the past.

I'm not suggesting that it is any better at all and probably beyond the scope of your supply plus also perhaps a bit more complicated but just another possible option.

Tracy

The idea with this is simple and short circuit proof, hence the low wattage lamps as series elements, which probably double resistance on a short.

Shunt regulated supplies are inherently short circuit proof, except that the series resistor might overheat. Using a lamp or lamps like this as the series resistor is a simple and clever way to give automatic current limiting.

Thanks, Jeffrey.

Also LEDs etc are not going to replace the oven lamps, so they seem to be a safe choice for availability. The local mini-market at the petrol station has them. While B&Q or similar are not the best place to get a shaver transformer, they have them. It's the most available high isolation HT transformer, though limited to 20W, and has 110-120V centre tap. I use them with an extra hole drilled for earth pin of a 5A plug for mains/battery sets that often have no isolation and dubious safety. Also even isolated ones may have poor isolation.

The design will also use 200V electrolytics (common in SMPSUs and easier to get than 300+ types). The parts can all be bought new, or you can recycle most old PC AT SMPSUs for parts.

More soon, inc photos.

(03-06-2024, 12:54 PM)Mike Watterson Wrote: [ -> ]Also LEDs etc are not going to replace the oven lamps, so they seem to be a safe choice for availability.

I'm not 100% sure abut this. I'm pretty sure I've seen an oven using LEDs and a light pipe of some kind. But oven lamps will certainly be around for a good while.

Many of us also have stashes of GLS lamps just in case we need them

I found even with a -48V PSU that "short circuit proof" was tricky. Off the shelf solutions seem only reliable up to about +/- 30V

A series regulator that's short proof with up to 650V or more instant HT across it is possible. There are 200W 1200V FETs. It's not simple.

Also I'll see how useful manual adjustment of FET bias to adjust HT is. Plan B would be a comparator and feedback. But unless using an off the shelf solution oscillation is a risk. Again an open loop "shunt regulator" is simpler than series for HT, because of shorts.

A more complex solution (and now more common) is a low voltage supply and a boost regulator (a similar concept to a Line Output stage). These are conceptually simple and there are controller ICs. You might need a custom inductor and RFI can be an issue. The Cûk version of a SMPSU is easier to filter. I've built one for +90V using a reed switch for on/off. A custom coil on the reed was in series with the filament supply (regulated 1.35V or 6.8V, or D cells). The battery valve sets are so seldom used that replica battery packs make more sense.

This HT PSU is for bench tests, not building into equipment, which would use fixed HT.

(03-06-2024, 01:08 PM)ppppenguin Wrote: [ -> ]

(03-06-2024, 12:54 PM)Mike Watterson Wrote: [ -> ]Also LEDs etc are not going to replace the oven lamps, so they seem to be a safe choice for availability.

I'm not 100% sure abut this. I'm pretty sure I've seen an oven using LEDs and a light pipe of some kind. But oven lamps will certainly be around for a good while.

Well, you can retrofit LED lamps to fridges, fluorescent tubes, incandescent room lighting, street lights, cars etc. I agree a light pipe on a new design of cooker is possible, but they are still selling new cooking things with the lamp too close to heat for LEDs. No way to put LED an oven, grill, air-frier etc designed for incandescent.

Beware of the safe operating area of the FETs, caught out many before. An example, I needed 10A at 400V maximum for a constant current source, in the end 96 TO220 MOSFETs did the job, yes 96!

It's a less than 20W transformer. I think worst case is no load and about 50% volts? At max volts there is no current. Worst case current at zero volts there is a 26W lamp fed by 240v (less here), about 110 mA max, or if two lamps in parallel, 217mA, but that ignores the power limiting of the shaver transformer's core design, so I doubt more than 150mA is possible. Or less than 75mA on the voltage doubler mode. I'll put an old Pentium II or similar heatsink and see.

An EL34 ought to be tested in ita amp. It's about 90mA anode and 20mA screen. The FET in that situation would be off. Pa 25W The vg for cut off might be -40V
An EL84 is about 50mA anode max and 6mA screen max.

With two lamps in parallel and the nominal 300V supply and output at 150V and no load you might see about be 25W to 30W on the FET, but that's impossible due to the transformer limiting at 20W. I can't imagine more than about 8W to 12W worst case on the FET with no external load. Hard to say without plotting V/I for transformer rectified and R vs I or V for lamps.

Some actual measurements. lamps are rated 230V-240V
Cold (about 19 °C)
15W (small bayonet) 333 Ω
25W (small ES, old) 195 Ω
26W (small ES, new) 169 Ω

Hot @ 121.5V AC (based on measuring AC current)
25W (small ES, old) 1841 Ω
26W (small ES, new) 1599 Ω

Hot @ 235V AC based on wattage
25W (small ES, old) 2209 Ω
26W (small ES, new) 2124 Ω

Approx Current mA
Vs 25W 26W
30 27 44
40 32 48
50 35 53
60 43 --*
121.5 66 76 (Actually an autotransfomer. I only have insulated one screw holder, so didn't test the bayonet model.

[* can't read my note!]
So due to much hotter at rated voltage, the resistance swing is far more than for valve heaters.

The only new filament lamp in the regular shops now is the Philips 26W smaller ES oven lamp.

So the series limiting action of these simple filament lamps is much more than 25W to 100W halogens I've tested before. A staggering 10:1 resistance change. The old iron wire and hydrogen filled barretters are even more impressive. I think I've one somewhere.
Most lamps are a coiled coil rather the the straight wires on these.