What driver tube for 200Vrms?

[SpreadSpectrum]

I am looking at getting linear swing in the 200Vrms level for amps with heavy local feedback in the output stage.

Basically my approach will be to load this driver only with a source follower biased with 1M resistor and having 5.5pF reverse transfer capacitance, so AC load will be relatively light on this driver. I am just concerned with linear voltage swing under these conditions.

I have achieved the following results with a 6SN7 loaded with 10M90S cascode, biased at 6mA and 380V on plate, and followed by FQPF2N90 source follower:

1.2% THD @ 200Vrms, output taken from source follower source.

Spectrum is attached.

My question is, has anyone done better and if so with which tube?

[Eli Duttman]

Look at the GE data sheet, which suggests for 'GTAs and 'GTBs you can go as high as 450 V. on the plate. It's tough to get a resistively loaded triode to yield a linear swing in exceeding 50% of the anode voltage. Choke loading might help.

JJ's ECC99 can take 400 V. on the plate and its low R_P makes it a better candidate for choke loading than the 'SN7 is.

[Wavebourn]

You may use gyrator made on a decent MOSFET, instead of CCS, to stabilize anode voltage.

[SpreadSpectrum]

I guess I neglected to mention that these tests are with about 800V B+. Remember everybody 200Vrms is almost 570Vpk-pk.

Eli,

Unless I am mistaken, no choke will give me a high enough impedance to get 200Vrms @ 20Hz starting at 6mA. Inductance would have to be unrealistically high. AC load line needs to stay very flat across all frequencies to avoid clipping on the anode upswing.

I have wondered if I would get better performance with this circuit and an ECC99 at higher currents. It is hard to tell how a tube will perform when you are swinging almost twice as high as the published curves go. Is the 5W limit for ECC99 per half or for the sum of both? I imagine it is the sum. What would be really cool is maybe an octal ECC99 with a limit of 8W per half.

Wavebourn,

Gyrator might be a better design due to the stable anode voltage, but I still have to make sure the impedance stays very high all the way down to 20Hz or I will clip at low frequencies. I have noticed that there is about a 20V variation in where the anodes end up when I plug in different tubes.

[SemperFi]

How 'bout 6BL7 or 6BX7? Rated 500V continous and 2000V during peaks.

Transmitter tubes? Like 809? Perhaps a bit fancy and expencive as driver.

You could also use a current mirror and swing the voltage into a negative rail. That way you can use just about any tube you like. Actually, come to think of it, I am not sure you'd find PNPs or P-MOSFETs with high enough voltage ratings.

[Eli Duttman]

I think that's 5 W. per triode, for the ECC99. Remember the '99 is similar to the 5687 and the 6Н30П (6n30p).

60 H. should be enough as the load for ECC99 sections, as that yields a X_L at 20 Hz., which is > 3X R_P.

[Wavebourn]

You may use triode strapped GU-50 with 450V on anode (800V B+) Or even in pentode mode, if to keep screen grid on stable 250V.

Do you mean electrostatic speakers? Load would be about 4K on 20 KHz.

[SpreadSpectrum]

Originally Posted by SemperFi

How 'bout 6BL7 or 6BX7? Rated 500V continous and 2000V during peaks.

Transmitter tubes? Like 809? Perhaps a bit fancy and expencive as driver.

You could also use a current mirror and swing the voltage into a negative rail. That way you can use just about any tube you like. Actually, come to think of it, I am not sure you'd find PNPs or P-MOSFETs with high enough voltage ratings.

Those look like some good options. I don't think I want to use large tubes for a driver. The cool thing about this approach is that you don't need to burn a bunch of power in the driver as it is driving a very light load. The source follower does the heavy lifting on the power tube grid. I AC couple driver to source follower and bias the gate of the fet, which is DC coupled to the power tube.

I have had difficulty with the lack of high-voltage p-fets in the past as well. 500V is about as good as it gets.

[SpreadSpectrum]

Originally Posted by Eli Duttman

I think that's 5 W. per triode, for the ECC99. Remember the '99 is similar to the 5687 and the 6Н30П (6n30p).

60 H. should be enough as the load for ECC99 sections, as that yields a X_L at 20 Hz., which is > 3X R_P.

5W per triode is reasonable. I will definitely have to get some and experiment.

I guess I'm still not on board with the choke idea. Maybe I am missing something. Being able to swing these large voltages is only possible with these low power tubes with light AC loads, hundreds of k or megaohms. I just don't see how low frequency performance can be adequate with the relatively very steep load line that will exist only at low frequencies. Am I missing something? The low frequency response gets trashed enough in the OPT, I don't want to make it worse.

[Eli Duttman]

3X R_P is quite sufficient for good linearity and gain retention. It gets better as the freq. rises. Ask a real expert, like Doc. B. or Paul Joppa about choke loading. They've done quite a few parafeed designs.

A big plus to choke loading is the ability to swing voltages above the "idle" anode potential.