It looks like a simple series-pass voltage regulator, except with the grid modulated by the SRPP gain stage so as to help maintain dynamic regulation under signal drive conditions without resorting to feedback. It instead appears to utilize feedforward A.C. error correction. Would be very helpful in such an otherwise basic 6SN7 based follower regulator, what with it's approximately 330 ohm output impedance. The trick would be to derive the optimum amount of feedforward signal by setting the ratio of the two divider resistors, which apparently are 300k and 910k for the schematic shown. Interesting.
Last edited:
The anode of the middle triode looks into a cathode follower formed by the top triode.
Without the 0.22u cap, the top triode simply buffers the voltage divider to provide roughly 3/4 of the power supply voltage to the middle triode's anode.
With the 0.22u cap, the top triode slightly modulates the middle triode's anode voltage in anti-phase its cathode voltage, i.e., it slightly increases the AC voltage across the middle triode.
It doesn't seem likely that the effect is significant though. I'm not sure why it's there.
Without the 0.22u cap, the top triode simply buffers the voltage divider to provide roughly 3/4 of the power supply voltage to the middle triode's anode.
With the 0.22u cap, the top triode slightly modulates the middle triode's anode voltage in anti-phase its cathode voltage, i.e., it slightly increases the AC voltage across the middle triode.
It doesn't seem likely that the effect is significant though. I'm not sure why it's there.
Yes, you can reconfigure as a cathode follower and have unity gain- which is all most people need for 21st century sources.
400V caps are cheap, why take chances?
I dunno, this seems a rather clever little circuit to me. I see it as a feedforward type of cathode-follower voltage-regulator. Coupling signal from off of the SRPP lower triode's cathode to the regulator triode's grid has the potential to dramatically lower the A.C. output impedance of the regulator, without resorting to feedback and second triode which would otherwise be necessary.
Spice simulations indicated it performed quite poorly. Someone might want to simulate it again. My simulation indicated that performance was considerably better without the top triode..
Yes, I can believe that the execution may be far from optimal, yet I see nothing terribly wrong with the concept. I've certainly seen some more silly circuits published here. One of the potential oversights in simulating such a regulator could be the assumption that there is no noise and a zero impedance from the supply rail - I'm not saying that you made such an oversight. Both simulation assumptions, which if applied would, of course, be false. At the very least this regulator should provide some increased level of PSRR, which would be far more effective if they had used a pentode instead of a triode for the regulator. The A.C. output impedance also is not helped by the choice of a relatively low gm triode.
I can imagine that the selection of a high gm pentode, or better yet, a high voltage MOSFET, plus the optimum application of feedforward signal could result in excellent A.C. performance from so simple a regulator and not employing feedback.
Last edited:
what this circuit does, when properly set up, it injects the noise on the cathode into the series regulator that feeds the plate and cancel the total noise. Theoretically, The sampling effect of the cathode signal to the series regulator dynamically adjust its bias for current demands of the input signal. the bias shift is small in comparison to the bias shift in the main grid circuit. (as the signal effects bias (also known as AC bias)). But the drawback of this design is its potential to over exaggerate negative peaks if signal input is high and skewing the power gain to bandwidth response due to negative feedback. this is much more apparent in higher gain tubes (especially if a 12ax7 or an 6sl7 is used) . The more more sloppy the power supply is designed, the better it runs. The bad thing about this circuit is the 100K attenuator in the front needs to be 1M and the grid bias 1M. Better yet, no 100k attenuator. the output impedance is 10k or higher. I hate it when people throw crappy attenuator/capacitorology on the input that totally kills the bias set up. Think people: a 470k and 100k will give you an effect grid bias resistor or 82K ish, so that textbook 470k is a different animal and since the grid bias impedence is too low, 10-30% of the potential signal is heater noise. The best output loading would be closer to 200k. A good tube output amp or a fet input based audio amp would have best results with this amp.
Last edited:
the power supply has too high in output impedance for the series regulator tube.
the output resistor is wrong size.
I'm sure people are tired of seeing this poorly designed line amp.
Theoretically, the design concept is neat, but this is a very bad implementation of it.
If I was a professor at a school, I would suggest to the student that came up with this design to look at the amplifier as a whole and not just a summation of sub-sections.
the output resistor is wrong size.
I'm sure people are tired of seeing this poorly designed line amp.
Theoretically, the design concept is neat, but this is a very bad implementation of it.
If I was a professor at a school, I would suggest to the student that came up with this design to look at the amplifier as a whole and not just a summation of sub-sections.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.