Advice for budget valve amplifier

[smoking-amp]

re: goldenbeer

Very interesting driver with the neg. impedance loading (bootstrapped beyond unity). Does it oscillate without the global Neg Fdbk?

I wonder if "local" N Fdbks could still be added to the driver cathodes, without disturbing the 3rd H cancelling bootstrap loads.

 

[artosalo]

In part, yes, but only to a smaller fraction. An even larger part of the H3 is cancelled by the x-coupled positive feedback around the final and driver stages.

No, quite contrary.
The output stage (with common cathode resistor) alone is more linear than the combination of the bootstrapped driver stage and the output stage.
So the "the x-coupled positive feedback around the final and driver stages" more than doubles the THD compared to output stage alone.

Attached the simulation results.

Why would anyone use 8 ohm outputs with 4 ohm loads??

Any real designer takes varying load impedances into account when designing an amplifier.
Haven't you ever looked the impedance curves of loudspeakers ? Those are very far from fixed values.

Please also note that my circuit also uses less overall GNFB

Your circuit has 11.8 dB of GNFB (R6 = 3k3), my circuit has 10.6 dB.

...my circuit keeps better power balance at varying loads than yours, even with less effective GNFB...

Please define power balance.

Concerning the operation with varying loads:

Your circuit: Pout = 6,5 W to 4 ohms; THD = 2.2 %

My circuit: Pout = 6,5 W to 4 ohms; THD = 0.44 %

But you will see the ultimate result, which will be worse than the simulations let you expect, as soon as you get actual amplifier done.

 

[goldenbeer]

No, quite contrary.
The output stage (with common cathode resistor) alone is more linear than the combination of the bootstrapped driver stage and the output stage.

It makes no sense to evaluate the combination of bootstrapped driver stage and the output stage in isolation. Obviously, the positive feedback boosts the THD inherent to the tubes. (It also re-duces the actual THD due to better load lines on the drivers thru another effect.) But then it also provides more gain, which can be used for more global NFB around all stages, which in turn reduces the total THD of the whole amp. This is a totally all stages integrated design. It doesn't make sense to just look on the single stages independently.

Any real designer takes varying load impedances into account when designing an amplifier.
Haven't you ever looked the impedance curves of loudspeakers ? Those are very far from fixed values.

Yes, and a speaker with 4 ohm dips is not a 8 ohm speaker.

Your circuit has 11.8 dB of GNFB (R6 = 3k3), my circuit has 10.6 dB.

This view doesn't make sense. You need to take the inner positive FB into account as well.

cheers,
GB

 

[artosalo]

...my circuit keeps better power balance at varying loads than yours, even with less effective GNFB...

I am still waiting for the definition of the above.

Concerning the operation with varying loads:

Your circuit: Pout = 6,5 W to 4 ohms; THD = 2.2 %

My circuit: Pout = 6,5 W to 4 ohms; THD = 0.44 %

And waiting for your comment concerning the operation with different load impedance.
Or do really think that your design is fully OK when it generates 0.14 % THD at nominal output power to 8 ohms, but 2.2 % THD with half power to 4 ohms ?

It makes no sense to evaluate the combination of bootstrapped driver stage and the output stage in isolation...

What do you mean ”in isolation” ?

It is very important to evaluate the performance of the combined driver/output stage and also optimize it.
The valid design process is such that, at first the output stage is designed to work as good as possible. This means that it will be evaluated/simulated alone.
Then the preceding stage, already designed alone, will be combined to the output stage,
and this combination will be optimized to optimum performance.
Etc. Etc.

This view doesn't make sense. You need to take the inner positive FB into account as well.

At first I must make a correction. The GNFB in my circuit is 11,7 dB (when loaded with 8 ohms).

The Global Negative Feed Back is the difference of open and closed loop gain.
In your circuit the GNFB is simply caused by the signal thru R6 (3k3).
The inner positive FB is out of this study.

 

[smoking-amp]

One thing I would be wary of in using positive feedback in an inner loop, and solely N Fdbk in the global loop. The positive feedback has got good bandwidth. But the global loop does not, due to the OT. At HF the positive feedback loop may prevail. (which might not show up in a simulation, depending on the OT parameters used)

So I would be more comfortable with sufficient N Fdbk moved to a "local" loop to keep the positive Fdbk loop under control with similar bandwidth. Then whatever excess gain is left over would be used for the global N Fdbk loop.

It seems to me that N Fdbks (from output plates) back to the driver cathodes would leave the 3rd H canceling effect at the bootstrapped loads intact, it just reduces the effective drive voltage to the stage.

 

[smoking-amp]

Advice for Budget Amplifier:

Mickeystan has just posted a nice TV Sweep Tube amplifier with the latest drive technology. 6DQ5 tubes are like $10. Could also use 26DQ5 ($3), or 6HJ5 ($4), or 6CB5A ($5) or 21LG6 ($4), 12GE5 ($3) ..... Could use a lower B+ voltage with some design adjustments, like a 3.3K Ohm primary OT. These output tubes can handle the current easily.

http://www.diyaudio.com/forums/tube...nificent-television-tubes-67.html#post4675483