If I understand Steve's circuit correctly, the HE'ing is for the driver stage HD and not the output stage since he is "sampling" the difference at the grids of the output tubes not the cathodes/filaments. Its a very cool concept, I hope it doesn't get lost and more folks explore the idea.
It would help for those of us who aren't as proficient in theory if he simplified that portion of the schematic and explained what he is trying to acheive. *I THINK* I get it, but still would be helpful.
I wonder if Steve will read this thread?
It would help for those of us who aren't as proficient in theory if he simplified that portion of the schematic and explained what he is trying to acheive. *I THINK* I get it, but still would be helpful.
I wonder if Steve will read this thread?
Hybrid Cascodes
It would be still interesting to hear his position on that
To quote Nelson Pass, and I hope I am quoting this correctly since I can't remember where it was: "The upper device (in a cascode) imparts very little of it's own sonic signature".
By that account the hybrid cascode input stageis, from a sonic perspective, really very much a SS stage.
It would be still interesting to hear his position on that
To quote Nelson Pass, and I hope I am quoting this correctly since I can't remember where it was: "The upper device (in a cascode) imparts very little of it's own sonic signature".By that account the hybrid cascode input stageis, from a sonic perspective, really very much a SS stage.
Re: Hybrid Cascodes
I don't remember a Bible from Nelson Pass. Which topology did he mean writing about the upper device?
The bottom device sees no voltage swing, it sees a current swing. The upper device sees both. Common base transistor and common base tube stages exhibit different characteristics, even driven by high dynamic resistance.
MRupp said:
I don't remember a Bible from Nelson Pass. Which topology did he mean writing about the upper device?
The bottom device sees no voltage swing, it sees a current swing. The upper device sees both. Common base transistor and common base tube stages exhibit different characteristics, even driven by high dynamic resistance.
JoshK said:
Hmmm - I'm pretty sure Steve is compensating for output stage distortion, not the drivers. Correcting the drivers is hardly worth doing, since it's the output stage that is the dominant source of distortion (in a well-designed amplifier).
I agree that Steve is not a minimalist with his circuits - there are probably lot of ways to do this. The basic concept is that a balanced PP stage is equivalent to a balanced-diode modulator in terms of the nonlinearity generated; thus, you need a tunable source of odd-harmonic distortion to be pre-injected into the final stage of the amplifier.
The cleanest output tubes will need the least, and "dirtier" tubes with greater proportions of high-order harmonics will need more. The latter point has been experimentally confirmed by John Atwood on his Audio Precision test setup, so it is not just conjecture.
WEHE isn't feedback; WE invented feedback just a few years earlier, and the team that came up with WEHE probably knew Black, and it is possible that Black himself was involved. There weren't that many people involved in the Bell Labs high-performance audio (moviesound) team, and they must have been working together on the same projects.
I think it's significant that WE reserved global feedback for the 91A SET amplifier, which was their lowest-end amplifier intended for the smallest movie theaters. The PP 86 and 92 amplifiers intended for larger, more high-end venues used WEHE, not feedback. With Black on the staff, they certainly could have used feedback, but chose not to.
In one of the weird ironies of history, the SET fans re-create the 91A, but leave off the WE-original feedback network, while PP amps, which WE certainly made in great numbers, are typically used with feedback, exactly the opposite of WE's mid-Thirties practice.
In modern terms, we can think of WEHE as a specialized form of feedforward that applies to balanced circuits.
Lynn Olson said:
So, mine are not (I mean Pyramid-V and Pyramid-VII)... Below 1/8 of a max sinus power they generate 70 dB below the 2'nd order only, no others are visible, however above that power the 1'st pentode starts showing up the 3'rd and the 4'th... The driver and the output are both symmetrical so they can't show the 2'nd without the 3'rd. Are they not well designed?
Lynn Olson said:
Big transformers caused stability issues?
I have no idea how the Pyramid-V and Pyramid-VII are designed, so I'm not in a position to say anything about them, pro or con. If Wavebourn has come up with a way to cancel output stage nonlinearities, congratulations are in order - it's not an easy task.
I was speaking of power amplifiers in the broadest sense, where the output stage typically generates the most distortion, particularly the most audible upper harmonics (5th, 6th, 7th, et al). Since 2nd is nearly inaudible, we can discount that, but harmonics from 3rd on up are another matter. D.E.L. Shorter and Crowhurst suggested weighting harmonics by the square of their order, and that makes sense for me. Chasing these out of the output stage is a non-trivial task - feedback decreases overall distortion, but it does not change the ratio of the lower to upper harmonics, so the amplifier retains its general harmonic structure - its "tone", if you will.
The output stage has to contend with many sources of nonlinearity: reactive loads, the possibility of the Class AB transition at low levels, a potential A2 or AB2 transition at high levels, (if the driver stage is capable of it), and impact of the reactive load on these transitions. All messy problems that are not a major concern of the driver stage - except for driving the Miller capacitance of the output tube grids. Really easy for a EL84, really hard for an 845.
One the most significant comments by Steve Bench is the smoothness of the AB2 transition. This is very rare and hard to do. It's smooth in the Karna amplifier, but that's because it has a very overdesigned Class A PP 45 DHT driver stage that is transformer-coupled to the output - a de facto 3 watt amplifier in its own right. This is an absurd amount of overdesign, and not something I'm very proud of. Steve is getting the same results with a lot less.
I was speaking of power amplifiers in the broadest sense, where the output stage typically generates the most distortion, particularly the most audible upper harmonics (5th, 6th, 7th, et al). Since 2nd is nearly inaudible, we can discount that, but harmonics from 3rd on up are another matter. D.E.L. Shorter and Crowhurst suggested weighting harmonics by the square of their order, and that makes sense for me. Chasing these out of the output stage is a non-trivial task - feedback decreases overall distortion, but it does not change the ratio of the lower to upper harmonics, so the amplifier retains its general harmonic structure - its "tone", if you will.
The output stage has to contend with many sources of nonlinearity: reactive loads, the possibility of the Class AB transition at low levels, a potential A2 or AB2 transition at high levels, (if the driver stage is capable of it), and impact of the reactive load on these transitions. All messy problems that are not a major concern of the driver stage - except for driving the Miller capacitance of the output tube grids. Really easy for a EL84, really hard for an 845.
One the most significant comments by Steve Bench is the smoothness of the AB2 transition. This is very rare and hard to do. It's smooth in the Karna amplifier, but that's because it has a very overdesigned Class A PP 45 DHT driver stage that is transformer-coupled to the output - a de facto 3 watt amplifier in its own right. This is an absurd amount of overdesign, and not something I'm very proud of. Steve is getting the same results with a lot less.
I think it is fair to say that these designs follow from the AT&T Bell Labs research on auditory perspectives done by Harvey Fletcher just a decade earlier. The Fletcher team did a great amount of research on perception of sound, speech, music, hearing, all that. It is an eternal credit to Fletcher that they did not assume an engineering attitude about music reproduction. They actively sought out Leopold Stokowski and much fruitful collaboration followed, to the benefit and enjoyment of both parties. Stokowski would hang around Bell Labs and listen to their equipment and Fletchers team would enjoy live Stokowski concerts. For a brief period, just a year or two there, Blumlein also worked for Fletcher. What an amazing alignment of deities.Lynn Olson said:
It would be worthy of a masters thesis in History of Technology to go back and dig out the story of the AT&T/ Bell Labs / WE / ERPI / Fletcher / Stokowski music years of the 30's and early 40's . They certainly got it right. We still revere those theater designs, the WE91's, the WE86's, the theater horns, the WE274, the 300B. They had no polarizing partisan problem about audio design, such as we have today with everyone dumping on the engineers. Back then, WE bypassed the golden ears and went straight to the source, the accomplished maestro musician for auditory guidance. Is this another WE lesson we could relearn?
As for myself, I imprinted on the Stokowski sound as a result of those early Disney movies.
Trivia question, there was yet another AT&T company not named here.
We know of AT&T, Bell Laboratories, Western Electric, and ERPI.
Who was the fifth company in the family?
Graybar Electric was named after the WE founders Gray and Barton. They are still around. www.graybar.com
We stand on the shoulders of giants
Back in 2006, I did a short presentation at work on reflecting 100 years after the famous American surgeon George Crile devising the classical operation of radical neck dissection, which is still the gold standard to treat neck lymph node metastasis of cancer. I did an side track presentation afterwards, celebrating 100 years of invention of the triode, which is still the most linear amplification device in electronics. I asked if any advance had been made in a century, both in medicine and electronics. Of course we have, but we cannot discount something just because they are not "modern".
We can now look further, because we stand on the shoulders of giants.
gallon said:
Back in 2006, I did a short presentation at work on reflecting 100 years after the famous American surgeon George Crile devising the classical operation of radical neck dissection, which is still the gold standard to treat neck lymph node metastasis of cancer. I did an side track presentation afterwards, celebrating 100 years of invention of the triode, which is still the most linear amplification device in electronics. I asked if any advance had been made in a century, both in medicine and electronics. Of course we have, but we cannot discount something just because they are not "modern".
We can now look further, because we stand on the shoulders of giants.
Correct, I thought that was one main reason for having a cascode.
Not only does it see it a current swing, it "creates" that current swing due to it's gate/base being driven by the signal voltage. That current swing is following the lower device's transfer characteristics, and that is translated into the amplified voltage by the I/V (anode) resistor on top of the upper device, hence it would seem natural that the lower device dominates the performance of the cascode, right?
gallon said:
Hear, hear!!!
I think the experience I imprinted on was seeing and hearing Ben-Hur at the Japanese premiere in Osaka in 1959. It was a curved-screen 70mm presentation, with what must have been five Altec A4's (288s, multicells, and 515s) behind the screen. It was the first time I'd ever heard stereophonic sound, and at the age of ten, I was absolutely stunned by the dimensionality and power of the sound. What set a lifelong passion for audio in place was hearing an Ampex stereo recorder in the home (playing Invitation to the Dance on RCA Living Stereo 1/2 track tape) and then also attending a full-scale live performance of Handel's Messiah the same year.
They say some things are life-changing experiences - these were the ones for me.
Lynn Olson said:
Actually it is easy, if to design topology properly, select working points, apply good amount of feedback, and use modern capacitors and solid state devices to make voltages stable like a rock, especially screen grid voltages. Actually, all this factors have to be blended and weighted carefully to get the optimal result, so the 1'st stage only (dealing with an error signal) will be left dominant in terms of distortions.
Most probably GE engineers went without negative feedback in their monsters because of expensive PS filtering (stability on lows) and huge transformers needed (stability on highs). With modern capacitors and SS devices available I believe they would do it easily.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.