design of a 200W DC to DC converter for vacuum tube RF power amplifier

[ashok.das81]

Hi all,

I am working on a Vacuum tube based RF power amplifier for my HAM Radio. I thought that I will find a suitable power supply for it but no such thing available from any reliable source. I am thinking to make a DC to DC converter that can convert 12v to 750v DC and can deliver 200W to the tube amplifier. I know this is not impossible and satellites already use this to power up the traveling wave tubes. The biggest discussion on this topic so far is this thread https://www.eevblog.com/forum/projects/switchmode-powersupply-for-powering-vacuum-tube-power-amplifiers/

I do not have any previous design experience of power converter, but I am sure I will be able to learn. So kindly share your ideas that i can use. Currently I need to understand what switching topology could the best option to make this. I know there are may topology on which a switch mode PSU can work.

Regards
Ashok Das

[moffy]

A push pull configuration is commonly used for the 12V to 400V DC to DC step up stage in 12V to 240V inverters e.g. https://schempal.com/12v-to-240v-inverter-circuit-diagram
Can't vouch for the specific circuit but it does illustrate the push pull configuration.

[Andy Chee]

For the 12V primary windings, push-pull topology is preferred because you only have a single switch device Vce/Vds drop.  This is despite requiring a centre tapped primary winding, in which half the winding does nothing half the time. 

You could use a H-bridge with a single winding primary, but even if you ignore the two switch Vce/Vds drop, the complexity of controlling the high-side switches may make you reconsider push-pull.

The high voltage secondary is almost always full wave bridge rectified.  But I've always wondered whether a full wave voltage doubler configuration would also work in this application? (which would permit lower secondary voltage winding).

[David Hess]

The usual topology would be a transformer coupled buck converter, which could be push-pull, half-bridge, or full-bridge, however a buck inductor swinging by 750 volts would be annoying.  The buck part however can be moved to the primary side and a simple rectifier attached to the output, which is essentially a buck converter driving an inverter.  Some old oscilloscopes use this method to generate multiple outputs, including high voltage outputs with low noise.  The input to the inverter could also be driven by a low dropout linear regulator.  200 watts at 12 volts will require almost 20 amps, but that is feasible on the input side.

The switching frequency is divided by 2 to drive the inverter, so every switch cycle reverses the transformer, and a full wave bridge can be used on the output.  The input buck converter can drive the transformer directly with its output current, so in a push-pull design, the transformer current will remain balanced and the output capacitors are only on the secondary side.

The input buck converter can use common parts, including the inductor, but the transformer probably needs to be custom wound.  It needs a turns ratio of perhaps 10:750.

Were you thinking of only powering the high voltage output while transmitting, or continuously?  Either is possible.

[T3sl4co1l]

A typical but improved approach would be like so:



I made this some years ago, which works fine. Build: https://www.seventransistorlabs.com/Images/InverterProto1.jpg

No driver was found necessary for TL598 into IRFZ34N.  I did note the shunt resistor's inductance (a 10mΩ metal-strip type) was high enough to seriously affect turn-off performance, which needed a bypass cap there.  Prefer chip resistors.

There's a UC3843 flyback for secondary side power, to run the oscillator and driver of a "modified sine wave" mains output.  H-bridge output, US 120VAC target, hence the 170V supply here.  Obviously that's not needed for a tube amp, and the direct DC output will suffice.

Scaling up output voltage is simply a matter of more secondary turns, finer wire, more inductance, and less capacitance.  This doesn't come for free, as the winding length inside the transformer matters.

Winding length can be partly compensated by simply using multiple DC outputs in series.  This may be desirable anyway, say if you need 300V for screen power.  This reduces the maximum secondary winding length by N, and has the downside of using N rectifiers in series, and more inductors and capacitors (but the same total amount for given capacity).

I would encourage a flyback (possibly QR type) for higher voltages.  Forward converters aren't great at high impedances: the transformer secondary needs quite high impedance, a contradiction for good transformer performance (low leakage is desirable); and the filter choke must be low capacitance as well (bank wound, preferably).  For example, here's a ~20mH <80pF inductor I made even longer ago:
https://www.seventransistorlabs.com/tmoranwms/Elec_Compound10.jpg
not so easy to do with commodity parts (e.g. divided bobbins on standard shape cores).

You might also simply avoid the high voltages and select a lower voltage type for the amplifier, if this is an acceptable route.  Sweep tubes for example have long been used in the SW band, perhaps with some neutralization as they aren't quite as RF-friendly as the proper types are; they're usable as low as 200V, albeit at some loss in efficiency.

Tim

[geggi1]

Ypu might get a way with a partly of the shelf solution.
Get a 12V to 230V (110V for US) inverter and match this with a standard tube amp power supply for the inverter voltage.
You might need to have the inverter oversized and a soft start on the tube supply.

[CaptDon]

Assuming you need negative grid bias voltage, screen voltage, filament voltage and maybe power for antenna relay I would consider building the amplifier like a normal 240VAC 50Hz mains operated unit and then run it from a 12VDC to 240VAC invertor. Otherwise you are going to need several buck or boost supplies besides just the 200 Watt plate supply. If you have a 100 watt transceiver like a Kenwood or Yaesu that run directly from 12VDC (Similar to my TS-690) or the TS-440 models then building an amplifier to go to 200 watts is pointless. In the end you may want to skip the amplifier project and buy a nice 12VDC transceiver if you don't already own one. Then you can 'cut the wire' or remove the 'magic resistor' or do the secret 'power on button press' and the unit will work on any band you are licensed for and the bonus is a good quality receiver and many added features. Remember, 1 S-unit is 6 dB so you must quadruple your power to gain a single S-Unit of signal strength. Also remember a 200 watt C.W. amplifier can only do 50 watts of carrier power on AM. A 100 watt transceiver is limited to 25 watts of carrier.

[boB]

I thought about doing this very thing.

One big thing to consider though is that a switchmode power supply will most likely make a LOT of RF noise.  I would think that if it can start up quickly, then you could leave the filaments running and switch on the HV PS just while transmitting.

But maybe you can made this supply run at say, 500 kHz to keep the interference spaced so it doesn't come up every, say, 25 kHz.

Now with GaN or maybe even Sic, you could do this or even higher frequency.

Sounds like a lot of work though to get it right.  I remember one supply that had a switching frequency fine tune adjustment.  Maybe that could be tied into the receiver tuning so that the EMI always misses the RX frequency.

Bigger transformers running at line frequency and bright and glowing tube rectifiers would be nice to look at but heavy of course.

boB

[David Hess]

I have worked with simple tube RF power amplifiers in the past which ran from automotive supplies, and they often just had a Royer converter.  Regulation of the high voltage is not that critical, so any inverter design can be suitable.  These also only turned on the high voltage during transmit because they responded so quickly.

[T3sl4co1l]

Nah, high 100s kHz or MHz would be harder to do, and send harmonics directly up the antenna just as well.  Low 100s is easy enough to filter; 10s would take extra work.

Shielding and filtering is required regardless; get it right, and just use enough for the job. Input and output.

Related, a long time ago I made a 20m receiver set, with a DC-DC for plate voltage.  With 2-stage filtering on the box, it's quieter than atmospheric noise:
https://seventransistorlabs.com/Radio_20m/Images/Radio_Wide.jpg
I later mounted it (the module on wires, top-left) on the "chassis".

Also, it's not a case of frequency band trickery or anything; that module has fairly spread-spectrum operation, and probing inside the box with an inductive loop probe plugged in as "antenna" clearly receives hiss and hum.  That is, using the radio to receive its own power supply's noise.  The harmonics don't actually extend much above 20MHz; it's a discrete design and the limited gain-bandwidth of such a circuit naturally limits switching speed and harmonics.  Makes it easier to filter too.

Tim

[NiHaoMike]

Ypu might get a way with a partly of the shelf solution.
Get a 12V to 230V (110V for US) inverter and match this with a standard tube amp power supply for the inverter voltage.
You might need to have the inverter oversized and a soft start on the tube supply.

Open up the inverter, remove the HVDC to AC stage, and replace the output side of the DC/DC stage with a voltage multiplier. A 240V modified sine inverter would have about a 320V DC bus, so a doubler would probably get close enough.

[ashok.das81]

Thank you all for your suggestions.

Its a very ambitious first project, but we will learn and built in a span of 1+ year time frame.  I also have a electrical engineer friend who is a transformer expert. He will do the transformer design and construction for this.

I brought the topic here to take the various opinions and suggestions that will help us to decide better. The inverter based solution is what I already thought and may be a secondary project as a backup.

Why I am making this? To do field operations with vacuum tube amplifier. This is a challenge and achievement.

My amplifier will be a single 6146B as designed by Greg Latta, AA8V https://www.frostburg.edu/personal/latta/ee/6146amp/6146amp.html.

I will use low power radio to drive this amplifier. The idea to turn off the high voltage during RX is good. I need to see how this can implemented so that the high plate can be turned on when transmitting only. This HV psu will output only 2 voltages, Plate voltage and screen voltage. For negative grid bias and filament supply I will use ready made converters. All of them will get 12 V DC from a big battery.

Thanks
Ashok Das

[ashok.das81]

Dear T3sl4co1l

As you have already done similar project, we can take this a reference. Is it possible that you share some more detailed photographs and a more clear circuit diagram? Also some more detailed description of the circuit if possible ?

Regards
Ashok Das

[T3sl4co1l]

This is I think the "original" post on that project. I don't have any closeups.
https://www.eevblog.com/forum/projects/considerations-for-a-high-power-forward-converter/msg3253022/#msg3253022