compactron output stage and power supply

[engels]

I've got a pair of 33GY7 and a pair of 38HE7 compactrons which contain output pentode and damper diode. In an attempt to design an output stage I stumble upon too many questions.

First, here's the datasheets:

33GY7

38HE7

With plate voltage of around 130V I'm thinking about using a 230V:115V stepdown transformer. The diodes in these tubes look seductive and I was thinking about some weird hybrid power supply that I haven't seen before, I'm probably wrong so please tell me why. I'm also not sure the diodes may withstand the currents (around 120mA total) and voltages (155 DC input).

Pentodes are also questionable, the cathode resistor and load should be calculated somehow... any ideas?

Here's a primary schem:

[ilimzn]

You don't have to limit the plate supply, in fact, you could probably benefit from it being considerably higher. What you do have to limit, and regulate VERY well, is the G2 supply. The maximum for this is 150V but youa r emore likely to use 100V or even lower! because of this, in order to get the required G2 supply, do NOT just use a voltage drop, use a proper regulator. If you look at the published curves for these tubes, you will see that G2 current has very 'bumpy' shape as plate voltage nears G2 voltage, so you must keep G2 voltage very stable. If this 'bumpy' current is alowed to modify the G2 voltage, this will be a form of feedback which will make the plate curves look much worse.

Regarding the damper diodes, no need for hybrid power. In fact, I would not recomend it because of the declared maximum voltages between heater and cathode of these diodes. The best way to use them as rectifiers, would be with teh classical arrangement with a center tapped transformer.
In such an arrangement, the ANODES of the tubes are tied to the transformer winding ends, and the cathodes become the DC output voltage. This way they will sit at +B or close to that, which is important because your heater supply, which is common to the diodes and the pentodes, will be close to 0V. The heater-cathode potentials given specify that the heater can be up to 500V NEGATIVE with respect to cathodes, which will be Ok since your B+ is surely less that 500V. On the other hand, if i remember right, heater-cathode for the diodes, with heater positive WRT cathode is only some 50V.

Regarding current through the dampers, this should be no problem. These are very sturdy, and besides, ina center tapped trafo supply, each diode passes current 50% of the time. Because voltage drop is so low with damper diodes, it is also possible to use larger filter caps after them. One thing which you may want to do is create separate B+ for each channel of your amp, from the same center tapped trafo, using two of the 4 available damper diodes per each B+. I would recomend this over paralleling damper diodes.

Regarding initial calculations for operating point and load, cathode resistors etc, the first thing you need to do is extrapolate what the curves might look like for a low Vg2. Normally, everything shifts downwards. In your case this is quite important because most of teh useful parts of the curves is over the limit for the max dissipation, on the curves in the datasheet (Vg2=130V IIRC).
Alternatively, you need to work in class AB. This automatically means fixed bias and obviously no cathode resistors (except perhaps very small ones to measure bias current with). In this case you will not get far without plotting the composite curves for the PP arrangement, see here:

http://members.aol.com/sbench102/composit.html
http://www.ax84.com/media/ax84_m225.pdf

[Tweeker]

Why would class AB preclude cathode bias? Resistor based versions might not be ideal, but its certainly been done. Did you mean AB2?

[ilimzn]

Originally posted by Tweeker
Why would class AB preclude cathode bias? Resistor based versions might not be ideal, but its certainly been done. Did you mean AB2?

It does not preclude cathode bias but it is not ideal. In fact, in this case it may be far from it, looking at the curves the idle current will be quite small. So, whichever way you do it, either separate or common cathode resistor, you would get rectification effects and 'sliding bias'. Of course, it depends on what you want - it might be just the ideal thng for a guitar amp.

[Tweeker]

One method to largely avoid the effect is diode bias. LEDs seem to be fairly popular.

[ilimzn]

Yes, or a shunt regulator of some sort. But, if you think about it, it's not that different from fixed bias. LED's have veery low internal resistance (especially compared to the usual cathode resistor values), so most of the self-stabilising benefits of cathode bias are not there anyway. What is there, unfortunately, is the power dissipation - bias current times bias voltage - no way around that. This is not present with fixed bias. OTOH fixed bias does have some pitfalls with relatively high gm tubes such as these sweep tubes, though at the expected bias curent I think it will be just fine.

[engels]

Thanks a lot for tons of usefull ideas!

I may put a 0B2 or 0C3 regulator on the screens but I'd like to know WHY the datasheets always refer to the 130V screen voltage in typical operation section (no other screen voltage mentioned)? I understand I may lower the screens and raise the plates but then I will need to meet dissipation conditions and they're pretty strict.

Is it possible to squeeze more than ~12 PP audio watts out of two above mentioned tubes anyway? (just to remind everyone, the dissipation is 9W plates and 3W screens while our EL84 or 6V6 have 12W pates and 2W screens)

I'm not sure I'm going to spend on a 230V CT transformer (115-0-115) for this experiment. What I can find is non-CT trannies ("american voltage converters").

At this point I'd probably leave the diodes hanging unconnected... I just hate the standby switch!

And btw, I don't care if this is going to be a (cathode-biased)guitar amp, I have only one pair of 33GY7's and one pair of 38HE7 so there's no chance for a stereo hi-fi, only a monoblock.
Maybe a stereo SE.

I was thinking about making an experimental mono PP with 33GY7's, learning something in the process and then making a better amp with a pair of bigger 21GY5's (18W plates/3.5W screens dissipation).

[SY]

Is it possible to squeeze more than ~12 PP audio watts out of two above mentioned tubes anyway?

Yes, if you have the design skills to pull it off and the test bench to work out the design. You want a high B+, relatively low loading, screen drive with control grid connected to the cathode, idle near class B, something like a source follower or a really good cathode follower doing the drive duties, high voltage drive input stage. Not trivial, but absolutely do-able.

[revintage]

Think you where on the right way in your first schematic. Just take the B+(155V) directly from the bridge with a 50-100u cap to ground.
Maybe you could use the diode voltage drop to get the 130V screen voltage.
For 38HE7, Ua 125V Iq 60mA/tube, 2000 ohms a-a you get a common cathode resistor of 180ohm, 3W class-A output and just above 10W class-AB. These are theoretical values.

You can use my www.revintage.se/PPABAMP.xls for your own calculations.

[engels]

thanx, revitage! this is probably what I was expecting - running the tubes close to the specified ratings and getting 10-12 watts into a relatively low load. I don't think I'm going to order custom OT, probably a 12V tranny with dual primaries (230-115-0) for the center tap (which should result in something like 2.5K plate to plate load into 8Ohms).

Still what SY tells is true - looks like it's possible to do something a bit less conservative and more efficient. Unfortunately, it requires design skills and a bench and I've got a soldering iron only.