The day before yesterday I stumbled across an article by L. Stellema (University of Sydney, Department of Aeronautical Engineering & Electrical Engineering).
Published in September 1980, it's a pretty well-designed hi-fi amplifier with impressive specifications.
in advance
the final stage of the small, clever amplifier - All transistors are mounted on a common heat sink!
Published in September 1980, it's a pretty well-designed hi-fi amplifier with impressive specifications.
in advance
the final stage of the small, clever amplifier - All transistors are mounted on a common heat sink!
Attachments
Thank you very much,
but this compilation is very extensive, only pages 144 to 146 could be extracted.
It's just a basic diamond buffer,
Q5 Q6 are current sources.
https://en.wikipedia.org/wiki/Diamond_buffer
But if you are interested in output diamonds, there is something better than Wikipedia:
this article is good - https://www.angelfire.com/az3/dimitri/images/ewww1192.pdf
Audio Research commercialized a topology "input voltage amplifier -> output voltage unity gain diamond", and there were very good amps like D100, D400 etc.
Andrea Ciuffoli (seems he is banned here IDK for the reasons) is a guy impassioned in this topology, and he is realist (lot of practice + LTSpice in reasonably sufficient quantities) - https://www.audiodesignguide.com/Ibridone/index5.html
this article is good - https://www.angelfire.com/az3/dimitri/images/ewww1192.pdf
Audio Research commercialized a topology "input voltage amplifier -> output voltage unity gain diamond", and there were very good amps like D100, D400 etc.
Andrea Ciuffoli (seems he is banned here IDK for the reasons) is a guy impassioned in this topology, and he is realist (lot of practice + LTSpice in reasonably sufficient quantities) - https://www.audiodesignguide.com/Ibridone/index5.html
There's no need to go on and on about the fact that it's all about basic circuits (and their components), the so-called Lego bricks.
The author simply calls the output stage a “voltage follower”, and we all know that it is a collector circuit - even in push-pull mode.
There was no fancy name for this configuration at the end of the 1970s. It is also generally better not to memorize names (alone), but to be able to read and understand the function.
I like the whole circuit and repeat my assertion to be on the safe side: a well thought-out, clever circuit.
Regards,
HBt.
The author simply calls the output stage a “voltage follower”, and we all know that it is a collector circuit - even in push-pull mode.
There was no fancy name for this configuration at the end of the 1970s. It is also generally better not to memorize names (alone), but to be able to read and understand the function.
I like the whole circuit and repeat my assertion to be on the safe side: a well thought-out, clever circuit.
Regards,
HBt.
Let's simply scale everything to (more than sufficient) 20W and ask ourselves whether a commercial concept that is frequently used today is technically better than the present design study.
Better?
I have the /this amplifier on the shelf - it sounds fabulous and yet I don't like it, the entire construction is a disaster - not only that the power transformer hums and sets everything into mechanical vibrations ..., but that would be a topic for the “grief box”.
Better?
I have the /this amplifier on the shelf - it sounds fabulous and yet I don't like it, the entire construction is a disaster - not only that the power transformer hums and sets everything into mechanical vibrations ..., but that would be a topic for the “grief box”.
Attachments
The advantage of the circuit in #1 is that both half-waves see the same two complementary transistors, so their audible different characters can be heard in both half-waves: The half-waves sound audibly less different.
Q5 and Q6 are not "identical". The largest audible difference between the half-waves is created here.
Q5 and Q6 are not "identical". The largest audible difference between the half-waves is created here.
The disadvantage of the diamond buffer as a power amplifier output stage is that the driver pair needs lots of quiescent current in order to provide enough base current to the output pair on signal peaks. You end up running drivers at the current and dissipation similar to that of the output pair.
In contrast, with the more common double or triple emitter follower, the drivers can provide more current on demand than the quiescent current, even if you keep them in Class A.
The diamond output stage works well where the peak output current is relatively small, e.g. in headamps.
In contrast, with the more common double or triple emitter follower, the drivers can provide more current on demand than the quiescent current, even if you keep them in Class A.
The diamond output stage works well where the peak output current is relatively small, e.g. in headamps.
There should be a resistance that goes from Q23 emitter to the negative rail, otherwise the recovery is not good.
Also a serial resistance from its collector to ground would be a good thing.
Otherwise the value of R58 is too low and the parraleled lead lag cap is not strictly necessary, you ll have to check what part of the circuit has a frequency response such that it s required to render the whole thing stable, guess that it s a byproduct of the shunt compensation, see if a TMC compensation isnt better, it should be more efficient to yield a good THD number without resorting to such daredevils tricks.
By the way: bipolar transistors in particular have a fine veil in their sound. They generally sound "technical" - compared to many Fets, for example. A summation, regardless postition and function, of these is always audible and ends in a grey, colorless sound without joy and elegance. The supposed advantage of more current, for example, is bought at the price of a powerless, colorless, slowed-down sound.
That s a current limitation for the predriver since there s a separate voltage rail for the driver stage, if there s no limitation
then the predriver Q26 will sink all it can in the darlington base if there s a clipping, with the consequences one can imagine
given that the predriver transistor is a small signal one with limited current capability.
I just noticed your clear scrupulousness towards historical details, so I’ll just finish about diamond buffers.
And here's another thing - the Lego analogy with regard to circuit design only works at the level of using integrated circuits.
At other levels, circuitry is a language, and specific circuits are sentences of this language, and everyone is free to build new sentences based on already known ones, but some very vivid sentences become idioms.
Diamond buffer is a typical idiom of a circuit design language; you can write a whole new story from this one idiom.
Unfortunately, in the late 70s this very characteristic idiom was widely known, and in 1971 it became the output stage of an commercial integrated operational amplifier (by Harris, in those days, circuit diagrams of integrated circuits were still published).
You can call it differently - a voltage follower (but this is not entirely accurate, because a voltage follower has a voltage gain of 1, but a diamond as a kind of emitter follower. always has a voltage gain of less than 1), a [power] buffer (a wide-range meaningless name), but because this circuit was invented by Mr. Joseph M. Diamond and published in 1969, it is commonly called the [D|d]iamond buffer:
https://ieeexplore.ieee.org/abstract/document/1162052
And since very few people now know about Mr. Diamond, it can be assumed that the name is so old that its origin has been lost.
Output diamonds very often (always) have stability problems in practice.
You can easily check this, the diamonds are very simple and in a week you can assembly 4-5 different ones to play with.
Abstract reasoning and LTSpace are not very helpful in this case.
A good PCB tracing helps much more, I checked.
John Broskie at his beautiful "TubeCAD Journal" blog has a lot of very interesting stuff related to diamonds.
This is a very bright idiom of the language of electronic circuits, a fact.
And here's another thing - the Lego analogy with regard to circuit design only works at the level of using integrated circuits.
At other levels, circuitry is a language, and specific circuits are sentences of this language, and everyone is free to build new sentences based on already known ones, but some very vivid sentences become idioms.
Diamond buffer is a typical idiom of a circuit design language; you can write a whole new story from this one idiom.
Unfortunately, in the late 70s this very characteristic idiom was widely known, and in 1971 it became the output stage of an commercial integrated operational amplifier (by Harris, in those days, circuit diagrams of integrated circuits were still published).
You can call it differently - a voltage follower (but this is not entirely accurate, because a voltage follower has a voltage gain of 1, but a diamond as a kind of emitter follower. always has a voltage gain of less than 1), a [power] buffer (a wide-range meaningless name), but because this circuit was invented by Mr. Joseph M. Diamond and published in 1969, it is commonly called the [D|d]iamond buffer:
https://ieeexplore.ieee.org/abstract/document/1162052
And since very few people now know about Mr. Diamond, it can be assumed that the name is so old that its origin has been lost.
Output diamonds very often (always) have stability problems in practice.
You can easily check this, the diamonds are very simple and in a week you can assembly 4-5 different ones to play with.
Abstract reasoning and LTSpace are not very helpful in this case.
A good PCB tracing helps much more, I checked.
John Broskie at his beautiful "TubeCAD Journal" blog has a lot of very interesting stuff related to diamonds.
This is a very bright idiom of the language of electronic circuits, a fact.
This kind of buffer was used in the 70s by Bang and Olufsen in their Beocenter 2800 and in the hefty 7000,
eventually it s from these designs that Hbtaudio got the idea.
Like here they also used a shunt compensation centered in the IPS to wich was added an inclusive compensation
enclosing the OS and Vas, and last but not least a lead lag compensation from output to inverting input, but this didnt
prevent te amp to have a little frequency peak in the 500kHz region that they surely tamed with an output RL filter.
eventually it s from these designs that Hbtaudio got the idea.
Like here they also used a shunt compensation centered in the IPS to wich was added an inclusive compensation
enclosing the OS and Vas, and last but not least a lead lag compensation from output to inverting input, but this didnt
prevent te amp to have a little frequency peak in the 500kHz region that they surely tamed with an output RL filter.
I didn't know the name as a person, and I first came across the nickname for a /this container, a kind of collective term for a topology (I'm deliberately not writing basic circuit or Lego brick !) 15 years ago through the DIY scene.
So no offense and many thanks for this information, which does not make L. Stellema's 25W amplifier from 1980 look any worse.
HBt.