Hi Guys
Quote from Nigel:
"So, all I'd like to know is what Mr Self would do if all plugs were thrown out of the window and he was given a truly free hand."
You'll never see such a response. At least not until well after he has sold the design to someone and can make money publishing it a few times....
He has a "legacy of Lin" to protect and uphold; his bread and butter all these years. To suggest that another way is even as good or - shudder - better would appear tantamount to saying the rest was foible - which in fact it would not be. The heavily economic path Doug has tread is only the part of his life so far and even for that just a portion of what is the total of his life experience.
I believe it was Ian Finch who commented that he found APAD6 to be a bit defensive regarding past developments and writing. I agree with him. On the other hand, some of the new info is both enlightening and disappointing. A good bit was the demonstration of output-inclusive compensation, which has been around for a long time but out of sight as far as APAD goes - until now. A disappointment for me was with respect to the comp-diff and push-pull VAS chapter.
We have seen production and diy examples of comp-diff front ends since the 1970s. The Leach amp is the one most people are familiar with. Its THD is not spectacular but many like how it performs. There are aspects of its design that are very appealing, even if Doug's test results show its warts. Those tests are not exactly equitable, despite the appearance of using the same devices as for the Blameless. No one building comp circuits uses those test devices.
It may simply be a "negative" attribute about complimentary circuitry that it is sensitive to device matching, or maybe just matching the gains of the circuit halves? But I think the true benefit of such circuits is in their THD profile and dynamic qualities others have theorised are more important to sonic perception than static THD numbers. The overall circuit may be more important when complimentary symmetry is all around than for an asymmetric circuit like the modified Lin. To rephrase, breaking the circuit into portions to isolate distortion mechanisms may not be as enlightening or useful as for the modified Lin.
My icon for symmetric design is Bryston. Their circuits are fully symmetrical, use nested feedback, have output stages with gain, do not use current mirrors, do not use current sources, yet push the AP-2 to its limits in testing. All of these attributes are negatives in Doug's book - save maybe for nested feedback, which he denounces for cost - yet the total package is marvelous. Erno Borberly is also a promoter of symmetry, although his projects leave me cold most of the time.
For me, Doug's "proofs" are convincing but so is the proof of performance I have seen and heard of Bryston's "all wrong" designs. I've seen Bryston's own test numbers and they are fantastic. Their claimed performance is very conservative. So, a contradiction that can only be interpreted as "many ways to peel the onion of THD".
I like and respect Doug. He has strong opinions and can generally back them up. His sharing of research, even if for profit, has benefited more than he knows - even his competitors, no doubt. The longer this thread goes on the more books he will sell, even if it is just so someone can come back here and denounce what he said on, say page 57 (arbitrary number). Good for Doug.
Have fun
Kevin O'Connor
Quote from Nigel:
"So, all I'd like to know is what Mr Self would do if all plugs were thrown out of the window and he was given a truly free hand."
You'll never see such a response. At least not until well after he has sold the design to someone and can make money publishing it a few times....
He has a "legacy of Lin" to protect and uphold; his bread and butter all these years. To suggest that another way is even as good or - shudder - better would appear tantamount to saying the rest was foible - which in fact it would not be. The heavily economic path Doug has tread is only the part of his life so far and even for that just a portion of what is the total of his life experience.
I believe it was Ian Finch who commented that he found APAD6 to be a bit defensive regarding past developments and writing. I agree with him. On the other hand, some of the new info is both enlightening and disappointing. A good bit was the demonstration of output-inclusive compensation, which has been around for a long time but out of sight as far as APAD goes - until now. A disappointment for me was with respect to the comp-diff and push-pull VAS chapter.
We have seen production and diy examples of comp-diff front ends since the 1970s. The Leach amp is the one most people are familiar with. Its THD is not spectacular but many like how it performs. There are aspects of its design that are very appealing, even if Doug's test results show its warts. Those tests are not exactly equitable, despite the appearance of using the same devices as for the Blameless. No one building comp circuits uses those test devices.
It may simply be a "negative" attribute about complimentary circuitry that it is sensitive to device matching, or maybe just matching the gains of the circuit halves? But I think the true benefit of such circuits is in their THD profile and dynamic qualities others have theorised are more important to sonic perception than static THD numbers. The overall circuit may be more important when complimentary symmetry is all around than for an asymmetric circuit like the modified Lin. To rephrase, breaking the circuit into portions to isolate distortion mechanisms may not be as enlightening or useful as for the modified Lin.
My icon for symmetric design is Bryston. Their circuits are fully symmetrical, use nested feedback, have output stages with gain, do not use current mirrors, do not use current sources, yet push the AP-2 to its limits in testing. All of these attributes are negatives in Doug's book - save maybe for nested feedback, which he denounces for cost - yet the total package is marvelous. Erno Borberly is also a promoter of symmetry, although his projects leave me cold most of the time.
For me, Doug's "proofs" are convincing but so is the proof of performance I have seen and heard of Bryston's "all wrong" designs. I've seen Bryston's own test numbers and they are fantastic. Their claimed performance is very conservative. So, a contradiction that can only be interpreted as "many ways to peel the onion of THD".
I like and respect Doug. He has strong opinions and can generally back them up. His sharing of research, even if for profit, has benefited more than he knows - even his competitors, no doubt. The longer this thread goes on the more books he will sell, even if it is just so someone can come back here and denounce what he said on, say page 57 (arbitrary number). Good for Doug.
Have fun
Kevin O'Connor
Last edited:
Do try CfbPair VAS , my feeling is it could be best of all . The curve distortion must be very low . I never had a problem with stability . Cascode I feel is increasingly the better choice . BC 550/560 C or what you can get and BF720/721 .
Douglas, fig. 7.24 on pg. 187 has Q7 labeled "cascode". This is incorrect: you're conflating "cascode" with "common-base stage".
There are such designs that work very well and D. Self
not paying attention to this topology is not an aknowledgment
that they are not valuable.
There is only one design with complementary input diff. pairs that works well, and is on a par, as far as linearity is concerned, with D. Self's designs. I know because I designed it myself. You'll read about it soon.😎
EDN just published Part 2 of Bruno Putzey's article The F-word - or why there is no such thing as too much feedback.
jan
jan
Would be interesting , i ve one as well that do better than D Self
generic 741 derived amps...🙂
That is not to say, of course, that I recommend complementary diff. stages or complementary second stages over Self's design.
Self s lin monoculture has an advantage , though , just look
at the japanese production from late 70s to early 80s and you
can be sure to see all his past, present and future iterations
of this ubiquitous design.
at the japanese production from late 70s to early 80s and you
can be sure to see all his past, present and future iterations
of this ubiquitous design.
I hope its better than ALL the other circuits you've posted which have such badly chosen operating conditions, currents & topology that they have abysmal performance .. even in SPICE world. 😱
I doubt quoting obscure names will improve incompetent design.
Well said!
It is very frustrating to try to read these threads only to find that every 10th post or so includes something of value to read. Mikey has been complaining about VAS/TIS probably for soon to be a decade.
Mikeks wrote "And they're all rubbish I am afraid."
As with a few others around here, no lack of modesty there then! 😀
I trust that you built you new comp diff design, measured it and can back your claims up. I look forward to reading about it.
As with a few others around here, no lack of modesty there then! 😀
I trust that you built you new comp diff design, measured it and can back your claims up. I look forward to reading about it.
The book thread is re-opened on the understanding that this thread is to discuss Dougs book and the content therin. Discussion of material and ideas for future inclusion is also acceptable.
Off topic posts will be deleted, as will posts using this thread as a "sounding board" for personal opinions that go beyond common civility. Repeat offenders of such postings will see penalties handed out.
So enjoy the thread... which is what its all about... and keep it productive and informative.
Hopefully this is on topic ? For years I have used 2SA970 PNP ( 872A 1085 and 716 ) . 120 V , very low noise and high gain . Soon I will be forced to use something different . I notice in the book that the input transistors are in the nothing special category . 2N5401 comes to mind as a possible substitute if nothing special is OK . I would like to have a >100 V capability if possible as sometimes I build high voltage amps that resemble hi fi use . LM4702 will do that for that , prefer it is my own design . I have seen MPSA 92 used ( HH ) . Any thoughts ?
Noise figure . 8 dB ? OK ?
http://www.onsemi.com/pub_link/Collateral/2N5401-D.PDF
Noise figure . 8 dB ? OK ?
http://www.onsemi.com/pub_link/Collateral/2N5401-D.PDF
It is true that the input transistors are not at all critical. High beta is good as it reduces voltage offsets and input current distortion. (but you should always drive from a low impedance to prevent the latter being a problem anyway)
Noise is also not critical, in the sense that the amplifier with ordinary transistors (eg MPSA56) is quieter than almost any stage you can put in front of it.