Member
Joined 2009
Paid Member
The one thing I've learned from my limited foray into SS (no, this is not a SS vs tube thing) is the belief by many that the sound of the amplifier is more than 'linearity'. These discussions often go off into the topic of harmonics and saturation behaviour to try and correlate the technical performance vs sonics.
I haven't been in the tube world long enough to get a feel for what you look for in choosing a tube to produce good sonics - if like SS it isn't sufficient to have only great linearity then what other properties should a 'perfectly linear triode' have to it's name for good sonics ? (e.g. mono vs twin plate - does this affect linearity, black anode vs non-black etc.)
Hi, before we discuss tubes, please advise which SS transistors have great linearity (?), so that I / we can understand your perspective.
Thank you,
G.
In terms of linearity I believe that triodes better or edge out transistors or SS. Where SS wins the linearity battle is with greater overall gain and copious amounts of time smearing negative feedback. Feedback can even fill in the gaps in crossover distortion. Which is better I do not care and will leave alone. They come from different worlds. I play with tubes for fun. The odd thing is that the test equipment is mostly digital and SS. Bigun if you find a definitive sonic difference between black plates and not let me know.
DT
All just for fun!
Not proprietary or even secret. It was developed live right here on this forum. See the 6L6GC in AB2 thread. The driver board uses two cascaded stages of 6SN7's in LTP feeding my usual pair of mosfet followers. I have a pair of these boards mounted on a piece of wood with screw terminals for quick change output tubes. I have used this setup with everything from 45's to sweep tubes including KT88's and of course 6L6GC's. The "mystery tube" was also revealed in that thread and tested in this amp.
For fully differential class A only operation I simply lifted the cathodes of the output tubes from ground and wired a CCS between the two cathodes (tied together) and ground. To give the CCS enough headroom I set the bias voltage on the grids of the output tubes to about +10 volts. The cathode voltage settles in at + 50 volts or so depending on the output tube and B+ voltage. Tweak the bias of ONE output tube for minimum distortion at low frequency. Since this only works in class A1, the mosfet drivers aren't really needed, and the positive grid voltage could be eliminated if the CCS was tied to a negative voltage source. This source must handle the full output tube current (mine can not) and in either case the usual 10M45 generally can't handle the current needed, and an LM317 can't handle the voltage. I used a Fuji mosfet with a string of LED's to bias the gate.
I am guessing that the fully differential trick might be a problem in pentode mode because of the screen current, but I haven't tried it yet. I will eventually.
These experiments took place last year. The amp has been idle ever since. I haven't had any tube time lately. I took the breadboard off the shelf and grabbed this quick (and slightly fuzzy) picture.
For fully differential class A only operation I simply lifted the cathodes of the output tubes from ground and wired a CCS between the two cathodes (tied together) and ground. To give the CCS enough headroom I set the bias voltage on the grids of the output tubes to about +10 volts. The cathode voltage settles in at + 50 volts or so depending on the output tube and B+ voltage. Tweak the bias of ONE output tube for minimum distortion at low frequency. Since this only works in class A1, the mosfet drivers aren't really needed, and the positive grid voltage could be eliminated if the CCS was tied to a negative voltage source. This source must handle the full output tube current (mine can not) and in either case the usual 10M45 generally can't handle the current needed, and an LM317 can't handle the voltage. I used a Fuji mosfet with a string of LED's to bias the gate.
I am guessing that the fully differential trick might be a problem in pentode mode because of the screen current, but I haven't tried it yet. I will eventually.
These experiments took place last year. The amp has been idle ever since. I haven't had any tube time lately. I took the breadboard off the shelf and grabbed this quick (and slightly fuzzy) picture.
Attachments
Member
Joined 2009
Paid Member
I don't think I asked my question very well judging by the response.
My question started out with an interesting comment from DualTriode
This made me curious. I was asking myself what kinds of things tube amplifier designers actually consider relevant for their designs. My only point of reference is SS amplifiers and my experience was that most designers felt that linearity was insufficient to guarantee good sonics and they spend time speaking of even vs odd order distortion amongst other things. In fact, some SS designers think chasing the best linearity can even be counterproductive for good sonics.
It didn't cross my mind that tube amplifier designers might approach things differently and I wanted to know more about it.
Hello Bigun,
If you want to get inside a tube designers’ head read anything and everything that you can written by Norman Chrowhurst and George Cooper. These guys work focused on tubes before SS was the big dog. There is much best of worlds to check out when you are done with Norman and George
Credit goes to Pete Millett, thank you Pete.
Check out this book.
High-Fidelity Circuit Design, Norman Crowhurst and George Cooper, 1957, 296 pages - Courtesy of John Atwood
The title pretty much says it all here.. this is a practical guide to the design of audio circuits.
Technical books online
DT
All just for fun!
If you want to get inside a tube designers’ head read anything and everything that you can written by Norman Chrowhurst and George Cooper. These guys work focused on tubes before SS was the big dog. There is much best of worlds to check out when you are done with Norman and George
Credit goes to Pete Millett, thank you Pete.
Check out this book.
High-Fidelity Circuit Design, Norman Crowhurst and George Cooper, 1957, 296 pages - Courtesy of John Atwood
The title pretty much says it all here.. this is a practical guide to the design of audio circuits.
Technical books online
DT
All just for fun!
Member
Joined 2009
Paid Member
Everything depends on how to measure distortions. My favorite example is an average temperature of patients in the clinic, it is very similar to THD. If an average temperature is higher than normal, that means patients are ill. But if some of them die their temperature goes down to the room temperature, so the resulting average temperature looks closer to the norm, or even normal! Does it mean we cured all patients, as the measurement suggest?
Similarly, if to make the 2'nd order error lower introducing some higher order errors, the result seems to be "more linear". But does it mean more linear according to human perception of linearity? No. The sharper is bent the curve, the more audible is an error. Smoothness should be measured, not linearity.
Similarly, if to make the 2'nd order error lower introducing some higher order errors, the result seems to be "more linear". But does it mean more linear according to human perception of linearity? No. The sharper is bent the curve, the more audible is an error. Smoothness should be measured, not linearity.
Sounds to me that there are procedural and sampling errors in the clinic patient temperature reporting protocol, dead people are not patients they are mistakes.
Speaking of amplifiers and amplifier testing (leaving human hearing for another day) linearity is well defined, smoothness is not. Linearity or lack of it can be seen in a transfer function. You can also see linearity or lack of it in a plot of grid voltage vs. gm. The straighter these curves the more linear the tube output.
Put two matched tubes in a push pull amplifier and ideally the 2nd harmonic distortion will cancel. The 3rd harmonic will remain. In the big picture push pull reduces THD but is it better? We will need to take the amplifier to the psychophysics lab in the Psychology building and do some human testing. You do know what killed Pavlov’s dog? We do not want to cause experimental neurosis in audiophiles!
Can you hear the difference? Which is better?
DT
All just for fun!
DT,
It wil work with pentodes too. I have an OTL amplifier, that uses a LTP driver using 6CL6 with the screens tied together to a common feed to B+ (265V), the same as the plate resistor packs. The cathode is at -365V (with B- at 615V). The set is absolutely stable into the Mhz range. from DC. [It is a Krohn-Hite DCA 50 laboratory amplifier.] Tying the screens together this way forces the pentode into a linearizing area, I guess.
albert
It seems to me that both linearity and smoothness are well defined. Linearity is about the degree of departure from a straight line. Smoothness is about the speed of departure from a straight line.
To get mathematical, if an amplifier transfer function is written as a Taylor series then what we want (for high fidelity) is that no terms beyond the linear term appear in the expansion. In reality there will be higher terms. Linearity says that the sum of these higher terms should be small. Smoothness says that all these higher terms should be individually small, and the higher order terms should be much smaller than the lower order ones. Our ears seem to prefer smoothness to linearity, with the caveat that there is a some preference for a small amount of second-order distortion rather than perfection.
To get mathematical, if an amplifier transfer function is written as a Taylor series then what we want (for high fidelity) is that no terms beyond the linear term appear in the expansion. In reality there will be higher terms. Linearity says that the sum of these higher terms should be small. Smoothness says that all these higher terms should be individually small, and the higher order terms should be much smaller than the lower order ones. Our ears seem to prefer smoothness to linearity, with the caveat that there is a some preference for a small amount of second-order distortion rather than perfection.
Member
Joined 2009
Paid Member
For Single End Triode fans you are right on target. Catch me on a Push Pull day and I would disagree.
DT
All just for fun!
Hi guys,
thanks Merlin,love the discussion (and I have your book).
Does anyone have a set of curves for a triode conected 6AK5?
I have over 50 of these and am using them as triodes, pentodes - and Merlin's morph control is on the drawing board - for guitar and bass single tube stompbox preamps.
JimG
thanks Merlin,love the discussion (and I have your book).
Does anyone have a set of curves for a triode conected 6AK5?
I have over 50 of these and am using them as triodes, pentodes - and Merlin's morph control is on the drawing board - for guitar and bass single tube stompbox preamps.
JimG
Funny, I can low-pass filter a signal to achieve any abritrary degree of "smoothness".
Tim
Funny, I can low-pass filter a signal to achieve any abritrary degree of "smoothness".
No. It is frequency independent.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.