Adjustable Bench Power Supply - Redux

[Alan0354]

I don't know how useful is this schematic. I designed two guitar amp with power scaling.....that is adjust the +B voltage from high voltage to low. I never have problem of burning anything. Most likely the MOSFET oscillates and burn itself out. It is very common for MOSFET if you don't have gate stop resistor and careful layout.

In my amp, I use 8 position switch instead of continuous variable. But you can easily use a zener to set the voltage and use a volume pot that can take high voltage to adjust the voltage driving the gate.

In my circuit, I also adjust the grid bias proportion to the +B to keep the class AB amp with constant bias current.

In my circuit, I put a 180ohm resistor in series with the source of the MOSFET, this is to mimic tube rectifier and make the rail not as stiff as SS rectifier to give it some "sag" that is good for guitar amps. You can put lower value if you want stiff power supply.

 

[thorpej]

That seems like the right thinking. One other thing to consider is how much current you dump down the Zener. My Zener runs a bit high. About 124V. So my screens run a little high too.

I could have "fixed" that by running the Zener at less current, I.e making the resistor in series with the Zener larger, but then I don't think it would regulate as well. So I picked a middle ground so to speak.

Spent some time simulating / tweaking this tonight... as drawn, it does 116.3V at 20mA, and 115.4V at 80mA. Some tweaking may be necessary when I actually build it... the simulation uses 100V + 20V Zeners because I didn't have a spice model for the 1N5380BG handy. I ended up reducing the resistor value in the Zener divider (to increase the current through the Zener) in order to bring the gate voltage up where I wanted it.

Kevin, you spec'd the STP16N50M2, but I guess the STF16N50M2 would work fine since the MOSFET only ends up dissipating a couple of watts tops (according to simulation)?

 

[kward]

That's a good question (about the 47K bleeder). Best guess at this point is it was to control heat dissipation in the zener. (if that Zener fails open, the screen voltage jumps, and that could cause the tubes to be put into a very stressful condition under certain output loading conditions). It surfaced in one of the later versions of my schematic as I was thinking about failure/boundary conditions. (I had 11 schematic versions in all, tweaking ideas, rethinking if specific ideas would work, etc). You might be able to omit that resistor (or one like it) in your build since you have a different frontend on your screen power supply.

I spec'd the STP part because it has a metal back plane. I thought it would conduct heat to the heat sink better than the STF part with the plastic back plane. I wanted to keep the mosfet as cool as possible so it would have less chance of ever failing, When mosfets fail, I'm not sure how they fail...do they fail "open" or "closed"? In my previous testing when I was considering using mosfets in place of the power tubes, they always failed "closed," meaning the gates lost their ability to control the voltage across the device, so the source pin saw the same voltage as the drain pin. If that happened in this power supply, it could be catastrophic to the power tubes under the right output loading conditions. This might be another reason I added that 47K bleeder...

 

[thorpej]

Spent some time playing with the B+ supply in simulation. Kevin, not sure which choke you used, but I opted for a 600mA Edcor model... but alas, different inductance. I selected some long-life TDK caps for the first filter due to the high ripple current there, and put some slightly larger JJs in the second cap stack to tamp down some of the ring in the CLC filter.

 

[dcgillespie]

If I may, all American produced 6550 tubes carry a plate dissipation rating of either 35 watts or 42 watts, depending on the rating system used. When Tung Sol introduced the tube, the Design Center rating system was in place, and so the tube was rated accordingly using that system with a Pd of 35 watts. When the rating system changed to the Design Maximum system (1957), Tung Sol of course updated their published data to rate the tube under the new system at 42 watts (20% increase), which is why you can find some TS data rating the tube at the lower rating, and some at the higher. If you check the rating system used for any data being referenced however, the two numbers hold true to the two rating systems used, while the actual dissipation capability of the tube itself has never changed -- only the rating system has.

Now RCA -- reluctant as ever to even include the tube in its manuals due to its superiority -- continued right up to the end to show the tube with a 35 watt rating (never updating the data), not wanting it to best their own 7027(A) tube. That tube is also rated at 35 watts -- but under the Design Maximum system, while the 6550 carried its 35 watt under the Design Center system. When GE began manufacturing the tube, the Design Maximum rating system was firmly in place by that time (being the final rating system used), so their tube hit the street as a 42 watt tube. But in terms of actual plate dissipation capability, they're all the same -- from the first Tung Sol offering, up to the last MPD bottle to roll off the line.

Tung Sol (to my knowledge) never produced a 6550A. Their product always carried the 6550 label regardless of which rating system was in place throughout their production of the tube. I believe that the later slender straight side offering from Sylvania was also only offered as a 6550. Only GE/MPD attached the A suffix to their tube, which actually carries a lower Maximum Screen Grid voltage rating in UL service of 450 vdc, versus Tung Sol's rating of 500 vdc when operated in UL.

Dave

 

[kward]

Jason,
Looking good. I used Edcor's 2H 350 mA choke. (and it looks like I have a typo on the published schematic, because that choke has a 27 ohm DCR rating, whereas I listed it on my schematic as 37 ohm DCR).

Other things to think through.

1. With 400 mA max current supported, you may need to tweak the size of that 2 ohm resistor directly in front of the 317. It's there for current limiting under short circuit conditions. I did a lot of trial and error testing looking for the smallest value resistor that would support the max current I wanted. Since your supply supports 100 mA more output current than mine does (which is a significant amount I think), you might need to make that 2 ohm resistor a bit larger in value--maybe 4.7 ohms. Check the wattage dissipation also. You could make it much larger if you wanted, it just starts to have a detrimental effect on max current supported.

2. When you get to the stage of short circuit testing, I'd suggest starting with one single 6550 installed, such that current will be limited to ~100 mA. Do your short circuit testing in that configuration first. If that all works, then move up to two, three, and finally four 6550's installed, testing short circuit conditions after each addition.

 

[thorpej]

...and now the same "different transformer" treatment to the negative bias supply ... instead of using the Hammond transformer, it struck me that the Triad F4-120 would be sufficient; not much current required, and it seemed like 120Vac secondary would be enough. Simulated it with a generic PNP instead of a KSA1013 for lack of a spice model, and substituted MJE5730 for the MJE5731 because the application is well within the 5730s ratings (and besides, Mouser's out of stock on 5731s at the moment ).

Simulation showed that with a 5mA load, the bias voltage actually goes positive if the pot is dialed back to 0, so I back-stopped it with a 10 ohm resistor.

 

[kward]

That looks good Jason. Check the size of R3. With your lower voltage transformer secondary, there will be maybe 0.5 mA quiescent current through the Zener. You may want to lower the size of R3 to something like 15K so that you have about 1.5 mA through the Zener. At least in my build I thought 1.5 mA would be about right to regulate somewhat effectively at max load where the secondary voltage might be pulled down a bit. May not matter a whole lot. Just an observation.

 

[thorpej]

That looks good Jason. Check the size of R3. With your lower voltage transformer secondary, there will be maybe 0.5 mA quiescent current through the Zener. You may want to lower the size of R3 to something like 15K so that you have about 1.5 mA through the Zener. At least in my build I thought 1.5 mA would be about right to regulate somewhat effectively at max load where the secondary voltage might be pulled down a bit. May not matter a whole lot. Just an observation.

Ah, good catch. I was wondering about that... Adjusting that resistor to 10K yields 1.6mA through the Zener, and keeps Q2 below 3W at 30mA load at any voltage.

 

[Alan0354]

I though OP want adjustable supply, the few circuits in this page ( except my circuit) are all fixed voltage with zener diode.

If you change your mind to a fixed voltage. Tcdrive designed a regulated supply that is very good, you can read it from the top post in page 4 of this thread:http://audiokarma.org/forums/index....-not-commonly-used-in-tube-amps.790283/page-4
I think this is much superior than zener diode type. If you give a big enough heatsink for the MOSFET, you can either adjust R7 or R3+R4 to adjust the voltage. You adjust R8 for different current limits.

These are all very easy circuit and they work. Time to just build it.

 

[kward]

In a fully adjustable linear supply that supports say 10V to 500V adjust range at up to 500 mA, there exists a design challenge in the pass stage. Even today's best power transistors weren't designed to drop 500V at 500 mA continuously. A single pass stage topology will blow the transistors even if you're using 1200V devices and have them configured in an array (in parallel with matched devices, and current sharing resistors), and using massive heat sinks with forced air cooling. The mechanism that fails is is not the Vds or Vce rating, but rather the ability for the gate or base to control the voltage. The hybrid fet/bjt types fail the same way at those continuous voltage/current conditions.

Two ways around it (not the only ways) are to cascade pass devices (connect them in series), or use a different kind of pass device all together, which is what I did. For a cascaded transistor approach, if you could rig a control circuit that could evenly split the voltage drop difference, I think you would have a solid solution. I don't think this would be terribly difficult--you would need two error amplifiers that delivered adjust range from 0 to Vout/2, and Vout/2 to Vout, respectively, as well as make them both auto adjust to the Vout selected. I think that could be done with a dual ganged adjust pot and two parallel voltage divider circuits in the error amplifiers. Anyway this type of solution was more work than I was willing do, since I was fairly certain a vacuum tube could handle it. And I turned out to be right.

Now on the other hand if you can limit adjust range (i.e, limit voltage drop across the device) to say 100V, the traditional pass device approach works very well indeed. My experience anyway.

 

[thorpej]

Heh, fair enough. The F6-120 fits the bill at 250mA. (I plan to do a 4-tube / 400mA build.)

Hah, well wouldn't you know it... the AnTek toroid I'm using happens to have a 70V tap as well (this wasn't mentioned on its product page), so I could use that with a doubler setup and save myself the space of the extra transformer in the case. I don't know what the current rating for that winding is. I've used this transformer before in an amplifier, so you'd think I would have remembered that it's there, but I guess not!

 

[kward]

On the Antek toroid, I bet that 70V winding is beefy enough...

You could test it by loading the HV secondary up to max the supply will deliver (appropriately scaling the max DC current load your supply will deliver to the AC current delivered from the transformer), then load the 70V winding up to the current you need, and see how much it sags. Might need some beefy wire wound power resistors to do that test though.

 

[thorpej]

On the Antek toroid, I bet that 70V winding is beefy enough...

You could test it by loading the HV secondary up to max the supply will deliver (appropriately scaling the max DC current load your supply will deliver to the AC current delivered from the transformer), then load the 70V winding up to the current you need, and see how much it sags. Might need some beefy wire wound power resistors to do that test though.

Yah. I emailed them asking what the current rating on that tap was and they said "same as primary, but don't draw too much current from it". OK, THANKS, THAT'S HELPFUL GUYS!

With the required voltage-doubler, I'd need to draw a max of ~200mA DC I guess (assuming a max of 20mA screen current from each 6550 -- but maybe that's too low). In your schematic, the 80V tap on the Edcor is labeled 250mA (same as the rest of the HV secondary). On this particular AnTek, there's actually TWO 400Vac @ 500mA windings that I'm going to attach in parallel, and thus there are also 2 70Vac secondary taps, so I'm pretty sure there's enough juice. Using simulation, I've already adjusted the circuit for the 70V option, and would just need to order a few more resistors and another of the Nichicon 220uF caps.

 

[thorpej]

Yah. I emailed them asking what the current rating on that tap was and they said "same as primary, but don't draw too much current from it". OK, THANKS, THAT'S HELPFUL GUYS!

With the required voltage-doubler, I'd need to draw a max of ~200mA DC I guess (assuming a max of 20mA screen current from each 6550 -- but maybe that's too low). In your schematic, the 80V tap on the Edcor is labeled 250mA (same as the rest of the HV secondary). On this particular AnTek, there's actually TWO 400Vac @ 500mA windings that I'm going to attach in parallel, and thus there are also 2 70Vac secondary taps, so I'm pretty sure there's enough juice. Using simulation, I've already adjusted the circuit for the 70V option, and would just need to order a few more resistors and another of the Nichicon 220uF caps.

D'oh! I just realized -- I can't use the 70Vac secondary on the AnTek because it's a TAP of the 400Vac winding, not a separate winding. Which means I can't float it on the configured output voltage.

So I'm going to have to go with my original plan of using the separate Triad F6-120 for the 6550 screen supply.

Good thing I realized this before I built the board!

 

[kward]

Good catch! Measure twice, cut once as they say.

 

[thorpej]

Finished up the front panel supply board today. I had actually built it several weeks ago, but didn't get a chance to test it until today. Worked like a charm, but I am going to make one tweak -- jumper around the resistor in series with the fan... my fan draws enough current that the drop across that resistor is about 6V... I'd rather have the fan spinning faster. The black goop you see is some of that liquid electrical tape stuff -- an insurance policy against shorting some of the wires below deck to the heat sink via those plated through-holes. My panel meters take 9V, 9V, and 5V. The 12V regulator is just warm to the touch even with that small-ish heat sink (edit: after running for an hour or so, the hottest spot I can find on the regulator package + heat sink is 53C).

 

[s-petersen]

i just bought a different kind of power supply, 500v at 500ma, it needs better filtering, but that could be done.
I think it uses PWM on the primary of the HV transformer, it runs cool at 200MA 500v.
http://www.audiokarma.org/forums/in...r-supply-ec-135-osp-135.799795/#post-11103816

 

[Brice]

Cool but that one is not a linear supply isn't it ?

 

[s-petersen]

It's regulated,constant voltage, or current, but IDK if it's linear.
There must be some way to combine the 2 styles of PS like is done on some power amps, changing the charging voltage to the main caps, before the pass regulator on the fly.