Very true. After E.B. left Hafler Co, i was called to interview to take his place and/or help to improve the last amps (MOSFET HD200/500). I flew out fom calif and talked with the enginnering group and had a group meeting with david hafler. I didnt like D.Hafler enough to work for him. But I did help get the distortion down with simple things to the design and got a couple free amps for my trouble. Design is usually a group effort with contributions from many.
-Richard Marsh
I thought Rockford Corporation nearly killed the Hafler name when they didn't initially believe in device matching in production
Last edited:
Kevin I appologize for the rhetoric but the behavior of wires and resistors is determined by some basic physics. First principles are not subject to opinion, if you have extraordinary claims present the evidence after that minds might open.
It really makes no difference where the product is compromised - all it takes is one place, one time, to make it compromised. What use excellent design if its execution lets it down?
I cannot speak for others, but it's the sound quality which defines a manufacturer as High End in my book. Not many competent designers will say that they do High End, rather that they do things the way they think things should be done.
In my view, what let Bob down was his weakness in succumbing to doubtful build quality, or more precisely, in sizing. He will stick in many output power trannies, but will save on electrolytic capacitors.
As for his products going up in flames, obviously he had a build glitch somewhere. If memory serves, he did resolve the issue as his later day products did not burst in flames.
OT posts deleted. The fact that Dougs book is on the shelves means that this thread has really become redundant.
The main thing is double VAS gets discussed as it offers befits at very low cost . Rod Eliot was criticized earlier . Rod is the other person who tries to make a difficult subject easy to understand . I feel his down to Earth and dare I say musically orientated focus is looked upon by some as stupidity . I have worked and lived in Oxford ( now Shire ) all my 57 years . Feeling stupid is an everyday possibility in Oxford . That goes for all so we learn just to like one another . Seldom do we care if someone is right . We care mostly that they have passion . Right is a house keeping exercise , a tidy up which must be done .
Douglas, on page 223 you note that including an emitter follower in the minor loop does not increase forward path gain significantly.
The reason for this observation is the transimpedance stage (TIS) requires large source and load impedances for maximal transimpedance gain; this is accomplished by using a current mirror for the transadmittance stage (TAS) and a high impedance output stage, such as a triple with cross-coupled drivers.
Without the current mirror, there is little to be gained in respect of increasing forward path gain because the transimpedance stage significantly loads the transadmittance stage.
Indeed, Edward Cherry, in his "Ironing out Distortion" articles in Electronics World, indicated that there was little point in using a buffered TIS unless a current mirror was used in the TAS. He further declared that this is one of the great advantages of using a current mirror in the TAS.
The reason for this observation is the transimpedance stage (TIS) requires large source and load impedances for maximal transimpedance gain; this is accomplished by using a current mirror for the transadmittance stage (TAS) and a high impedance output stage, such as a triple with cross-coupled drivers.
Without the current mirror, there is little to be gained in respect of increasing forward path gain because the transimpedance stage significantly loads the transadmittance stage.
Indeed, Edward Cherry, in his "Ironing out Distortion" articles in Electronics World, indicated that there was little point in using a buffered TIS unless a current mirror was used in the TAS. He further declared that this is one of the great advantages of using a current mirror in the TAS.
Last edited:
Precisely.
Please remember that in those days, we had no High End as a clearly defined market segment, rather it was in the process of formation. Arguably, Marantz started the show with their last all tube models in the mid-60ies, but SAE picked up on this rather quickly, and Mark Levinson got John Curl to design what is commonly taken as the very first true High End units.
And from the beginning of the 70ies, there was Phase Linear. Despite some of my reservations, I would have no trouble calling them High End, even if their prices were not as high as those of some others.
From my correspendence with the late James Bongiorno, I know he never once even hinted at his being in the High End segment, with either SAE, or Sumo, or Ampzilla, or even his currwnt models. He always said he made them as he believed they should be made, period.
I don't know John Curl, I've only just met him on this forum, but he also never even hinted at being or wanting to be in the High End, he just talked about the sound only (and, on occasion, his vintage Porsche, which I can understand and sympathise with perfectly). I am sure he recognizes that his circuits can come out rather expensive, but hey - no pain, no gain. 😀
I don't have to agree with someone to be able to respect him for what I feel is creditable work. Rod Elliot is just such a case, even if I believe his circuitry is relatively outdated.
I'm with you on Rod, Nige.
The Douglas Self book is an exercise in sanity as is Rod Elliot's site . It can not operate in a vacuum as a reference . That's why we need a diverse discussion . I think most who are contributing here have a grasp of the facts and welcome input like #1854 . I love when someone tells me something I have known for perhaps 30 years as something I should know . Perhaps it is something I choose not to do . All the F1 guys know stuff , doesn't stop them loosing . That's when the armchair expert gets to know the truth .
If you disapprove of the book's content and still choose to post O/T here, then you must have some reason beyond just shooting the breeze, rambling from one anecdote to another about unrelated past industry figures, history and personal experiences - actually, it's the same ramble across several different threads, so I don't buy that disagreement argument for one second.
My opinion probably doesn't matter much but we are all free to say what irks us regarding on-topic material here as any other. If our arguments or objections won't stick and we still want to air them, we could try something innovative like starting our own thread - in the lounge, perhaps? 🙂
Douglas, on your section on the slew rate limitations of the Thompson arrangement, it would have been worthwhile to consider the amelioration of these limitations by simply using wholly independent amplified negative feedback (ANF) current sources for the input stage and the second stage.
Using a single ANF current source as a reference to bias a second transistor is really shabby and wholly inelegant, and cannot be justified on the grounds of economy as the extra components needed add less than 0.1% to the total cost of the amplifier; as you said in your book, the cost of all the small signal components is less than 1% of the total cost of the amplifier.
Using a single ANF current source as a reference to bias a second transistor is really shabby and wholly inelegant, and cannot be justified on the grounds of economy as the extra components needed add less than 0.1% to the total cost of the amplifier; as you said in your book, the cost of all the small signal components is less than 1% of the total cost of the amplifier.
Last edited:
Douglas, in the chapter on the transimpedance stage you state that the non-linear base-collector capacitance is the significant distortion mechanism with the single transistor.
This was pointed out by Edward Cherry in his "Ironing out distortion" article of January 1995 in Electronics world.
This fact should have been pointed out in your book and appropriate reference given.
Further, Cherry makes no mention of Early effect as a distortion mechanism in the second stage. Rather, he suggests that transistor beta and non-linear Cbc are the main distortion mechanisms of the second stage.
This was pointed out by Edward Cherry in his "Ironing out distortion" article of January 1995 in Electronics world.
This fact should have been pointed out in your book and appropriate reference given.
Further, Cherry makes no mention of Early effect as a distortion mechanism in the second stage. Rather, he suggests that transistor beta and non-linear Cbc are the main distortion mechanisms of the second stage.
Last edited:
I will drink to that ( # 1882 ) . I think it was said in previous editions of Self it is nicer to have low Ccb as it allows us to choose how much . Also an external capacitor tends to show a nicer distortion curve as it's anchor point is more advantageous . It is hard to say for sure if that is true as a capacitance which appears inside a transistor surely is a very fast path . All the same I will live with excepted wisdom on this . Could it be that low Ccb is just a better transistor and somehow the internal verses external idea grew from nowhere ? In my experiments the supposed stability advantage of a single input stage seems never to be a reality . More likely it was the high Ccb of the older designs . Mr Dvv has always advocated cascode VAS . There is a need to set voltages correctly and consider a stage 2 emitter resistor if so ( 33 R is where I start if 8 mA , consider is conditional on testing ) . My tests on all types of cascodes including valve/ transistor hybrids is very often positive . Absolutely no evidence the extra transistor is there except massive gain . Gain of 600 is not out of the question . Beta is a distortion mechanism ? Surely it is lack of beta ? Lack of beta is like driving with the handbrake slightly on .
Last edited:
Cherry states that sensitivity to beta and Cbc can be reduced by using Douglas's modified Darlington (with the buffer's collector connected to ground), or, in the case of sensitivity to Cbc, using a cascode.
From a PSRR point of view, there seems to be some benefit to decouple this buffer's collector (Tr12, page 636, figure 26.7) to the negative power supply (for an amp having a PNP input) : page 641, figure 26.12.
This what Samuel Groner has done in an elegant scheme in his SGA-HSA-1_r1 (PNP input) or SGA-JFA-1_r1 (N-Jeft input) :
SG-Acoustics · Samuel Groner · Discrete OpAmps
The decoupled collector shares the same node as the collector of an emitter follower making the current mirror of the input stage being of EFA type.
Last edited:
Forr, I can't find SGA-HSA-1_r1 (PNP input) or SGA-JFA-1_r1 (N-Jeft input) in the link you gave...
