So if I asked you to design a Blameless tube amp, it might end up as a hybrid with a MOSFET follower output stage?
Possible. Unlikely, but possible.
So if I asked you to design a Blameless tube amp, it might end up as a hybrid with a MOSFET follower output stage?
Interesting question. Proably not. If you read my posts in other recent threads, you may notice that I am not in favour of surrounding tubes with solid state bits - CCS's, voltage regulators, etc etc. In my view they a) are not necessary in getting impecable performance, b) they just add more to go wrong (and get wrong), and c) it conflicts with tube aesthetics.
But MOSFETS in a direct signal role is another matter. While I haven't given any thought to combining MOSFETs with tubes, I'm inclined to discount MOSFETS driven by tubes as MOSFET capacities are much much higher.
I would say that anything related to amplification has to be tube based, anything that supports the tubes whilst doing that can be SS. I.e. I would certainly use SS in the power supply, CCS, etc. etc.
Putting any SS in the actual signal path quickly turns this into a hybrid design, and then the question becomes where do you stop? You could easily make a composite amplifier with a nice opamp frontend, and tie the NFB loop around the OPT. Now we've got a real shot at building a tube amplifier with 10ppm THD20K, but only because the opamp allows for this.
In my opinion the real challenge is to make sure the tubes are comfortable enough to give it their best, obviously helped by a topology and gain structure that makes sure they are configured optimally.
That brings me full circle again, as despite using SS to create stable, low noise and low impedance supply rails etc. this exercise starts with identifying distortion mechanisms in tube amplifiers and address them to ultimately end up with a topology that's the best compromise possible to allow for the most accurate amplification of the input signal.
Ha ha! very Small how funny
OT
I recall that in his original paper (1964?) from wireless australasia, or something, he briefly shows a tube output stage arranged to synthesize the desired system Q by means of negative impedance output.
Last edited:
That doesn't make a whole lot of sense, like a lot of Morgan Jones's stuff.
The things that limit how much feedback is the overal gain, or low or high frequency phase shifts, whichever limit comes first.
Some commercial tubes amps eg GEC 88-50, have more GNFB than many SS amps, and they have local feedback as well, and as they were designed in the early 1950's, there's no silcon anywhere. Nothing but tubes.
My argument was as follows:
The amount of GNFB that can be used in a tube amp is limited by the phase shifts in the output transformer.
By using a MOSFET output stage, we could drive the speaker directly.
Having got rid of the output transformer, we could use a level of GNFB that Doug Self would approve of, except that tube small-signal circuitry wouldn't provide enough open-loop gain...
My point is that if we use rational engineering design and aim for minimum distortion, if we are allowed to use solid-state then the amp will become completely solid-state.
Oh c'mon, you're taking the fun out of it, the challenge is to design a tube amplifier that performs as well as a good SS amplifier whilst retaining the amicable and magical presentation a tube amplifier offers, that's one thing a tube amplifier offers over a SS amplifier, those glowing tube are intriguing and wonderful to look at (at least in my opinion).
Top hit on Google:
Murray Tube Amplifier
But that wouldn't be blameless (if for no other reason that tubes are a massively inefficient alternative to transistors). Tubes automatically carry more 'blame' than transistors. Anything you could do with tubes you could also do with transistors, but with even less blame. You can't have a 'blameless' valve amp, by definition, because they will always be 'blameful' devices compared to transistors.
Last edited:
Those are my thoughts as well. Mosfets make good followers. Due to their extremely high Gm they make a better follower than most tubes.
In a mosfet source follower the only capacitance spec that matters is the reverse transfer capacitance, Crss.
The huge gate to source capacitance is a non issue since the source follows the gate in a follower, thus the two terminals remain at the same offset voltage (bootstrapped). The source to drain capacitance may be fairly large but the source is at AC ground, and the drain can easily drive this capacitance.
The gate to drain capacitance (Crss) IS directly seen by the driving tube, so we MUST select the right mosfet. Mosfets with a Crss in the 1 to 10 pF range are common, but we MUST look for a fet where the Crss remains CONSTANT over the voltage range where it will operate. This curve is usually given in the data sheet, and fets with flat curves above 25 volts or so are common.
I generally use a fet with a Crss of about 4 pF flat from 25 volts to hundreds of volts for medium sized triode tubes like the 6SN7. I use the LND150 for guitar amps or other 12AX7 type applications. Even a wimpy 12AX7 can drive the LND150's 3/4 of a pF (less than some tubes).
I do not believe in using a mosfet in a tube amp for anything other than a follower type buffer, a power supply regulator, or a current source. Using a mosfet in a common source configuration DOES invite all sorts of VVC related effects because the capacitance changes over voltage at the voltage levels seen in tube amps are too great.
If you are prejudiced against mosfet followers in the signal path you CAN use a vacuum tube cathode follower, however if we are assigning blame (or absence of blame) to circuits found in tube amps, and we were to make some comparisons of a triode cathode follower, a pentode cathode follower, and a mosfet source follower, each directly driving the grid of an output tube in A2 or AB2, the triode follower most often seen in tube amps might just end up in last place.
Some people would discredit A2 or AB2 operation. If your tube amp has the 10 to 20 db of headroom needed to avoid EVER being overdriven, this is OK. For most of us, this isn't possible, so a good engineer MUST consider this as a design requirement. In fact every amp I design will be tested by me plugging in my guitar preamp and blasting away with everything I have got. I have found and fixed more that a few issues early in the design this way. The best way to avoid overload recovery issues is to anticipate them and design for them. This is what led me to discover the mosfet buffer about 15 years ago. A 6SN7 cathode follower will run out of drive current capability on transit peaks even driving a triode wired KT88. Somewhere I have the scope trace pictures to prove it.
Again, for me it is all about choosing the right component for the job considering the task it will see, using good engineering principles, without preconceived prejudice.
Would I use mosfets in the output stage, even a pair of complmentary followers, in a TUBE amp. NO.....it wouldn't be a tube amp.
Have I built "Darlington pairs" with a tube and a transistor or mosfet in the output stage incorporating an OPT, yes. Did I call it a tube amp, no, I did not.
Is my TSE design which uses a CCS chip load on a 5842 followed by a mosfet buffer driving the grid of a 45 or 300B a tube amp? Most people seem to think so, but some purists may call it a hybrid.
Can you drive the grid of a big transmitting tube like the 845 or 833A 100 volts positive with a cathode follower, yes you can. Does a mosfet follower do a better job when said grid draws over 100 mA on peaks, yes, it does. This an extreme case, but this is exactly where I went to test my mosfet drivers in the real world. These tubes and smaller tubes like the 811A REQUIRE positive grid transitions. The usual driver for direct coupled applications is a 6V6 or a 6L6GC, and even these distort at the current levels seen during signal peaks.
SY, myself, and a few others have been experimenting with screen driven sweep tube amps, and possibly driving both G1 and G2 simultaneously. It is possible to directly couple a cathode follower to the screen grid of a sweep tube, but the combination is susceptible to bias drift and runaway. those of us who have successfully tames these designs have all been using mosfet drivers.
I have a rather novel design for a high efficiency push pull amplifier design breadboarded and tested. It was the first time that I designed an amplifier completely in LTspice, and had the real world tested results match up rather well for an unconventional circuit. It will not work without a mosfet follower in the signal path.
I have recently moved, and will not have my lab up and running until well into next year. I can't publish this design yet since there are people here who will copy my work and sell it.
control theory is a little broader than the audiophool's caricature
its only simplistic if you restrict yourself to strawman constructions of stupid "conventional engineers" - even that further constrained by adding "audio designer" blinkers
control theory is a little broader than the audiophool's caricature
try: Amazon.com: nonlinear control: Books
Reducing nonlinear effects is the first, and major motivating advantage claimed for the invention of negative feedback in Black's patent - the objection that linear theory doesn’t apply because we're interested in nonlinear effects is naive; from the outset of negative feedback engineers have been exploring theoretical extensions such as "local linearization" and the small gain theorem to explain and expand on the distortion reducing effect of negative feedback observed in real world circuits
Negative Feedback, extending the applicability of the linear approximations is a major theme in those nonlinear control nonlinear books, courses
as is understanding the nonlinearities effects on controllability/stability and signal fidelity
particularly ironic in a tube amp discussion when triode relative linearity is a consequence of internal negative feedback from the plate V - specifically nonlinear feedback following the same Lagmuir-Child law as the gm gain nonlinearity
Tubes 201 - How Vacuum Tubes Really Work looks like a good start for tube refs
[Edit: engineers need good sound bites too - added it to title]
The preponderance of design mediocrity suggests not all of them did
...2) Negative feedback
The simplistic failure here is that the error correction is assumed to pass through a linear amplifier on it's travels back to the output, whereas in fact it passes through exactly the same non-linear one you are trying to correct.
True believers in GNFB don't understand that their pet theory only really works on a linear amplifier - i.e. one that doesn't need correcting (barring side effects like lowering impedance).
It took the Japanese to realise that GNFB was just moving problems - not solving them.
Unless we look at a music signal as it travels through the amplifier (and stop fixating on a low-signal 1kHz tone) we'll miss the subtle mechanisms that mangle the signal on the way to the air.
To a large extent this means a short path between input and speaker terminals - GNFB multiplies that short path (by definition) into a variable set of competing amplifiers at various levels and phases and we listen to them all - again it took the Japanese SET movement to realise that listening to just one one amplifier, once, was better.
And goodness knows why you'd want to listen to your OPT more than once lol
its only simplistic if you restrict yourself to strawman constructions of stupid "conventional engineers" - even that further constrained by adding "audio designer" blinkers
control theory is a little broader than the audiophool's caricature
try: Amazon.com: nonlinear control: Books
Reducing nonlinear effects is the first, and major motivating advantage claimed for the invention of negative feedback in Black's patent - the objection that linear theory doesn’t apply because we're interested in nonlinear effects is naive; from the outset of negative feedback engineers have been exploring theoretical extensions such as "local linearization" and the small gain theorem to explain and expand on the distortion reducing effect of negative feedback observed in real world circuits
Negative Feedback, extending the applicability of the linear approximations is a major theme in those nonlinear control nonlinear books, courses
as is understanding the nonlinearities effects on controllability/stability and signal fidelity
particularly ironic in a tube amp discussion when triode relative linearity is a consequence of internal negative feedback from the plate V - specifically nonlinear feedback following the same Lagmuir-Child law as the gm gain nonlinearity
Tubes 201 - How Vacuum Tubes Really Work looks like a good start for tube refs
[Edit: engineers need good sound bites too - added it to title]
Last edited:
Everything is a mathematical function of the components in the circuit.
I detect no information in your post?!
I was wondering where you read that the PSRR of a common cathode stage is always 6dB?Everything is a mathematical function of the components in the circuit.
I detect no information in your post?!
"look at a textbook design of (e.g) a common cathode stage, it's PSRR will be stated as 6dB. "
Lol, Williamson is great if you like hearing echos of your OPT and amplifier's non linearities all over the harmonic spectrum.
Personally I only want to hear my amplifier ideally once, and if I am forced to listen to bits of it multiple times it certainly won't be the OPT I choose!
Williamson is great for mid-fi but has no place in hi-fi, sorry. Enjoy your echoes
Once you decide everything has already been done, innovation stops. Everything can always be improved.
Why?
Any signal balanced mid way between power and ground will be the same. It's not rocket science, it's a simple fraction
The correct way of avoiding PSRR is to feed a cancellation signal into the circuit where required.
Agreed. The power-drive is an excellent stage and a FET is really the only way to properly avoid blocking distortion.
The non CSS version might be a mid tapped SRPP driving the MOSFET, with the bias circuit's resistance designed to keep/load the SRPP at it's sweet spot. E.g the SRPP exits into a coupling capacitor, which connects to the gate and the bias - the bias being 'earthed' via Rideal for the SRPP.
Do you have data showing these "echoes" which no one else has ever seen?
Any signal balanced mid way between power and ground will be the same. It's not rocket science, it's a simple fraction
Only for the case of RL = rp(eff), which is never seen in even a half-competent audio design, much less a textbook. Typically, RL is 5 or more times higher than rp(eff). Sometimes a LOT more. In which textbook did you read this?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.