Because of the high impedance feed to the top grid you could get strange results if the triode grid current model is not right. As the circuit/simulation is trying to balance one thing against another I can easily imagine that the simulation might have trouble converging.
Well, the potential divider is set up to drop about 1/4 of the available HT. If the capacitor was connected to ground, instead of the bottom triode cathode, the top triode would be a capacitance multiplier, for said capacitor, so a form of voltage stabilization for the rest of the circuit. But... it's connected to the cathode resistor of the bottom triode, which is essence modulates the HT for the bottom 2 triodes by an attenuated version of the input signal with some added distortion. What that does is another story... given it ends at the output through the plate of the middle triode.
Based on a couple of quick simulations with the 5687 this circuit seems to do little except to slightly degrade the overall linearity of the SRPP. I noted that the second and third harmonic were a couple of dB worse relative to the fundamental, other than that nothing significant outside of some ripple rejection which IMHO is irrelevant since a simpler, better thought out PSU design would take care of that issue. (And would be cheaper to make not requiring that third tube.) I will try again with the 6SN7 model once I get that sorted out, and of course this is a simulation, real world behavior may be better or worse. I don't however see any particular practical benefit except as a way to sell more tubes.. I guess it is a nothing more than a marketing gimmick unless the poster who stated that we don't understand this circuit can explain what it is meant to do.😀
It converges fine, but there is an error in the tube model which IIRC I have had some reason to suspect in the past. The bode plot results make no sense at all. Substituting another triode model and the results seem to make sense and are I think something of an indictment of the design premise baring more evidence.. (I need to recheck the inter-electrode capacitances in that model, and perhaps a few other things.)
I wouldn't call it junk however, just not the way I would choose to implement this old warhorse circuit. (I've built a lot of SRPP based line stages in the past to my current chagrin 😀, but I do like the way they sound, and technically if not the ideal topology performance has been reasonably good, and the simplicity helped keep the cost down as well.)
Last edited:
While he's explaining what it is meant to do, he might also explain why it's not particulairly good at it 🙂
My call would be, marketing gimmick. The top tube basically acts as a filter offering about 27dB reduction in ripple. It would actually do a better job if the 0.22uF capacitor was larger, and it's other end simply connected to ground. As it stands, it offers less ripple reduction where it counts most, at 50Hz, due to the haf-wave power supply, while at the same time slightly increasing distortion and reducing the total voltage headroom for the remaining 6SN7s to operate in. Why all this would be needed after several RC filters and a stack of zeners in the power supply, is also for that knowing poster to explain.
Just to clarify, I have simulated with the 0.22u cap to ground, also with comparatively extreme loads, clipping due to input overdrive, 10Vpp of ripple. I can't find anything to justify including the top triode, assuming a relatively noise-free 300V supply. And that would have been easier to do if only the remaining 3 diodes for full wave rectification were added - certainly less expensive than an extra tube + stuff around it. Connecting the 0.22u cap to the cathode of the bottom triode only seems to make the top triode ineffective in ripple reduction at very high frequencies, though I am sure someone 'guru n the know' will ascribe it practically magical properties, as well as give it a fancy name.
Mixing in a bit of HF noise from the power supply perhaps adds 'air' and 'sparkle' and stops the thing from sounding 'neutral' or 'clinical'. After all, we don't want to hear exactly what the mixing engineer in the studio heard, do we?
There have been a few theories as to what was in the circuit "designers" mind when he made this circuit. I will offer yet another.
It is obvious from the power supply that the designer either didn't understand the design and cost trade-offs, or he had a big box full of surplus zener diodes and needed to use them. A FWB, two small electrolytic caps, one zener, one or two resistors and a cheap mosfet would cost less than this mess and work a whole lot better.
Considering the highly effective power supply design, lets consider the amp. My guess is that the designer just looked up some popular circuits from the internet and wired two of them together. I will admit to doing the same thing years ago. Some of them actually worked quite well, some sucked big time.
The bottom two triodes is textbook SRPP, a circuit that was once in vogue, and can be made to work reasonably well with some tweaking. Note that the top tube in an SRPP circuit resembles a cathode follower.
The top triode in the circuit is the bootstrap tube from the augmented cathode follower, or SLCF. Its intention is to keep the voltage drop across the cathode follower constant. Perhaps the guru behind this circuit attempted to combine the two topologies somewhat unsuccessfully and then experimented with the connection of the bootstrap cap until he arrived at this design with its magical properties.
It is obvious from the power supply that the designer either didn't understand the design and cost trade-offs, or he had a big box full of surplus zener diodes and needed to use them. A FWB, two small electrolytic caps, one zener, one or two resistors and a cheap mosfet would cost less than this mess and work a whole lot better.
Considering the highly effective power supply design, lets consider the amp. My guess is that the designer just looked up some popular circuits from the internet and wired two of them together. I will admit to doing the same thing years ago. Some of them actually worked quite well, some sucked big time.
The bottom two triodes is textbook SRPP, a circuit that was once in vogue, and can be made to work reasonably well with some tweaking. Note that the top tube in an SRPP circuit resembles a cathode follower.
The top triode in the circuit is the bootstrap tube from the augmented cathode follower, or SLCF. Its intention is to keep the voltage drop across the cathode follower constant. Perhaps the guru behind this circuit attempted to combine the two topologies somewhat unsuccessfully and then experimented with the connection of the bootstrap cap until he arrived at this design with its magical properties.
So if the cap were returned to the upper tube cathode, the circuit would presumably work better.
I think the upper tube cathode is the more appropriate return point for the cap. Not that any of this will make any difference...😀
edit: I meant "middle" tube, i.e., upper tube of the SRPP.
edit: I meant "middle" tube, i.e., upper tube of the SRPP.
The proper place to connect the bootstrap cap in the SLCF or augmented cathode follower is to the output of the cathode follower. In this circuit that would be the cathode of the middle tube. I can't say where or if there is a proper place to connect the cap in this circuit.
The augmented cathode follower was a rather ingenius circuit developed and patented by Ross MacDonald back in the 1950's. It was used for extremely high input impedances and excellent driving capability. I believe the SLCF was derived from this work, although it could have been an independent design. I took the augmented cathode follower and performed the usual Tubelab method of cranking up the power, using it for an output stage. The circuit can be configured to have gain and I got a very clean 30 watts out of it, only limited by my wimpy power supply. Now that I have a big supply I must revisit the design. I made a complete amp on a PC board. I might even know where it is.
Last year I received an email from Ross MacDonald thanking me for keeping the design alive. He stated that he still tinkers with electronics but he is strictly a solid state kind of guy now.
As Allen Wright acknowledged, the SLCF was taken right out of Tektronix designs of the 50s and 60s. I didn't know that Ross McDonald invented it, but I'm not surprised.
Connecting the cap to the middle triode cathode reduces distortion by an order of magnitude. Maybe this product could be bought and modified to full wave rectification, and the cap connected to the appropriate point to form a SLCF version of a SRPP, to get something that actually has merit.
So, now we just might have done some proper design work for it's manufacturer, for free...
So, now we just might have done some proper design work for it's manufacturer, for free...
He was working at TI at the time.
The patents and published papers on the augmented cathode follower and lots of other good stuff related to tubes, solid state physics, and stuff that I can't comprehend are all on his web site. Look at "scientific publications by date", the tube stuff in dated in the 50's.
James Ross Macdonald
For the SLCF, see #12.
More CF stuff #25, and #40
Want to see a cool multi feedback P-P amp squeezing 65 watts from a pair of 807's? Look at item #26. Did the Citation II start here?
Distortion cancellation which is known as predistortion today and commonly used in high power RF power amps in cell phone towers. #52, #54, and #59.
Interesting, in simulation I got twice the swing and about the same distortion as for a conventional SRPP. (This using the 6587)
- Status
- Not open for further replies.