|
|
transconductance varies throughout the tube's operating range. The mu, AKA amplification factor, comes the closest to being a constant. Thus the rp maybe much higher at the chosen bias point and thus result in the higher output impedance. Increasing the idle current tends to lower the output impedance by virtue of moving to a lower rp point on the plate curves.
The second reason is the more likely culprit. My guess is that the amplifier is being run in Class-AB, as true Class-A operation would exceed the dissipation limit for the 6C33. The formula you provided is only accurate for Class-A amplifiers, as only when both tubes are conducting hard does the output impedance fall to one half that of a single cathode follower.
What is the answer to lowering the output impedance of your amplifier? You can add more gain to the amplifier and then wrap a feedback loop across it. Or you can use a low-winding-ratio output transformer. The impedance ratio of a output transformer is equal to the square of the winding ratio. Thus a winding ratio of 5 equals an impedance ratio of 25, which would lower the output impedance of your amplifier to less than one ohm. I understand that a output transformer ruins the OTL feature, but it is a practical choice if your goal is a feedback free amplifier with a very low output impedance. The only other alternative is to use more output tubes in parallel.
2nd E-Mail
Thanks for the reply. Yes the amplifier is running in Class-AB, plate voltage is 150v and the idle current is about 200 mA. The impedance I gave was for zero feedback. So the impedance is as high as Circlotron bridge amplifier's.
In actual operation, I use only the feedback port around the phase splitter. With 12AT7, it lowers the impedance down to about 1.5 ohm and yet it still has some gain (about 3.5 @ full power) and with 6H0PI (Russian tube) to about 2.5 ohms. It sounds more open this way than it does with any global feedback method (although the out impedance would be much lower, but it
|
|
|
|
sounds more like a transistor amplifier with lot of feedback).
I like the sound of this arrangement best. If I use both feedback loops at once, it doesn't sound that good either.
So my choice is to use only feedback around phase split with high gain tube like 12AT7, then again 6H30PI is much more linear and thus sounds more accurate (especially in low bass) than 12AT7. Even with 2.5 ohms Zo, it sounds good on my cheap Roe Anderson with 8" woofers. I think 2.5 ohms Zo is not too bad after all, is it? The other option to lower the Zo is doubling the number of output tube, so Zo now is 1.5/2=0.75 ohm. What else can I do to lower Zo using only feedback to phase splitter?
I also prefer very short feedback loops (a triode is effectively a tube with a built-in short feedback loop). Theory dictates otherwise: one global feedback loop allows the greatest reduction of distortion. Yet you and I are not pleased with the results. My guess is that we need a better theory, a more complete one, one that falls I line with perception.
Until then, I recommend that you try replacing the 12AT7 with a 5965. I have had some great results with this tube. But as it has a lower mu than the 12AT7, do not expect any further reduction in gain, and consequently, no further reduction in output impedance. One possibility is the 12BZ7. This tube is effectively two 12AX7s in parallel. Its mu is 100; its rp, 32k, its Gm, 3200µS. This tube will result in a higher output impedance, but this might be for the good, as the 6C33 is often used with very large valued grid stopping resistors, for example 10k to 100k. Topologically, using a cascoded differential amplifier will increase the gain of the phase splitter as would using pentodes instead of triodes in the existing differential amplifier. But I am not convinced that the lower output impedance is all that important. With a critically tuned bass reflex sub-woofer maybe, but then we are unlikely to want use an expense, low wattage tube OTL amplifier for driving that load.
|
|
