I would like to talk about circuit topologies once again. Even after much has been made about the standard op amp design, and how it has been refined up to the present, some people seem to be stuck on designs similar to what Charles Hansen, Erno, or I might make, and have made for nearly the last 4 decades. Why does something that looks like it was designed in 1973, get any notice today? Perhaps, because it was designed originally on 'first principles' rather than economy or compatibility with IC design processes.
This sort of design was not the first type of design that was popular back in the early days of solid state audio design, but it is still useful and 'popular' today. It came to be when the first complementary jfets became available, replacing the complementary symmetry transistor input stage, developed in the late 1960's. The jfets made input biasing easier, direct coupling practical and possible, and the lower Gm of the jfets allowed for a much higher slew rate, approaching 100V/us, compared to maybe 10-20V/us with earlier transistor inputs, as they were normally designed. (more later)
This sort of design was not the first type of design that was popular back in the early days of solid state audio design, but it is still useful and 'popular' today. It came to be when the first complementary jfets became available, replacing the complementary symmetry transistor input stage, developed in the late 1960's. The jfets made input biasing easier, direct coupling practical and possible, and the lower Gm of the jfets allowed for a much higher slew rate, approaching 100V/us, compared to maybe 10-20V/us with earlier transistor inputs, as they were normally designed. (more later)
Entertainment? Ever notice how popular some very simple circuits are with the DIY community, some threads go on for years?
Like my new sig? Sounds good set to a little three cord mile a minute rock.
Of course, many circuit designs are built for 'entertainment' on this website. Quasi comp, IC's + transistors, etc. However, there are certain circuit design topologies that are used today that just 'work' a little bit better than some others. One of these is the complementary differential jfet input stage. You might compare it to the 'boxer' engine layout, that I believe was first developed by Fd. Porsche, many, many decades ago, (even before my time), yet it is still used with Porsche's 'best' designs.
This is not to say that one topology is the best, everywhere, but if it lasts 40 years, even with parts becoming much more difficult to get, there is probably a good reason for it.
This reason is the inherently low distortion, and the minimum thru-path required to make a complete circuit from it, coupled with the reasonably low noise, (about 1nV/rt Hz is possible over a broad input impedance range).
Yes, the circuit is simple looking, but elegant in fact. AND unfortunately, no IC has been able to make it as of yet.
A practical engineer would just select an IC from the many available to make a phono preamp. For a power amp, why not start with an IC, then add a high voltage power stage with discretes, or even class D? Is this not the future?
This is not to say that one topology is the best, everywhere, but if it lasts 40 years, even with parts becoming much more difficult to get, there is probably a good reason for it.
This reason is the inherently low distortion, and the minimum thru-path required to make a complete circuit from it, coupled with the reasonably low noise, (about 1nV/rt Hz is possible over a broad input impedance range).
Yes, the circuit is simple looking, but elegant in fact. AND unfortunately, no IC has been able to make it as of yet.
A practical engineer would just select an IC from the many available to make a phono preamp. For a power amp, why not start with an IC, then add a high voltage power stage with discretes, or even class D? Is this not the future?
Even if all this is true complementary JFETs are no longer made by Toshiba.
So it's rather theoretical than practical.
Even if all this is true complementary JFETs are no longer made by Toshiba.
So it's rather theoretical than practical.
It is still practical, since the parts are still available, though not cheap and not in abundance.
Yes, you sure know the concept of never surrender. We rarely fill in the blanks $15,000 preamps beget $20,000 turntables and $3000 cartridges and so on and so forth. This is not DIY. Trying to carry on a discussion to help folks get 99.9% of this on an ordinary budget is greeted with as much abuse as you attribute to your attackers.
Last edited:
It is true that 'the very best' costs real money, that also goes for telescopes, autos, cameras, etc. DIY, the very concept, is just fine. However, not knowing WHAT to DIY, because some people don't discern between different designs, makes the SAME effort next to useless in one case, and a real performer in another.
The circuits that I was describing in a previous input, are SIMPLE, ELEGANT, and BUILD-ABLE. The only real requirements are matching and availability of parts, that are still available, especially to amateurs with low quantity needs. Why build a Ford if you can build a Porsche with the same effort?
The circuits that I was describing in a previous input, are SIMPLE, ELEGANT, and BUILD-ABLE. The only real requirements are matching and availability of parts, that are still available, especially to amateurs with low quantity needs. Why build a Ford if you can build a Porsche with the same effort?
John,
you are either treading a fine line here between genuinely believing this stuff, or just plain ignorant of the current state of the art. Ditto CH, but I'll try to steer clear from raising that 'dirty little bit of sand' story again . . . especially since some CFA's running open loop were snuck into one of his designs.
There is no 'very best' circuit design or topology in high end audio. Its all in the listeners head, and that's what determines whether its high end or not. I can promise you, if you spend 10 grand on a system it WILL sound good to your ears. If you're student and broke and you hack something togther on a piece of stripboard for $20, it WILL sound awsome. Period.
you are either treading a fine line here between genuinely believing this stuff, or just plain ignorant of the current state of the art. Ditto CH, but I'll try to steer clear from raising that 'dirty little bit of sand' story again . . . especially since some CFA's running open loop were snuck into one of his designs.
There is no 'very best' circuit design or topology in high end audio. Its all in the listeners head, and that's what determines whether its high end or not. I can promise you, if you spend 10 grand on a system it WILL sound good to your ears. If you're student and broke and you hack something togther on a piece of stripboard for $20, it WILL sound awsome. Period.
Bonsai, this is just an excuse IMO (if someone spent 20K it would sound great to him...). This is not only a question of money, but real high sound quality products are very expensive. I have a chance to participate in comparison tests quite often, and I have never found low-cost consumer product sounding perfect. The very well sounding products have always been very expensive, though high price does not automatically mean good sound.
Last edited:
I would like to say it straight. IF you or anyone thinks that whatever you hear is virtually perfect, perhaps because you have been influenced by constructing it, then all you have to do, is to give it to a neutral listener to compare against other audio components. THEY will set you straight. Some of us care about this stuff, I have spent almost 50 years investing time and effort into hi fi. I have had others positively approve of my efforts from the beginning, but that did not mean that I knew 'everything' there was about audio. Just slightly more than the average college student. As time progressed, I went from 'acoustic suspension' to K-horns, to Magnepans, Rogers LS3-5A, and finally to WATT-Puppy's.
Each major switch involved 'trade-offs', some of which I evolved into, yet I might miss the 'dynamics' of a 3 way horn loaded system, but the 3 way K-horn would annoy me in other respects. Each and every one of these 'tradeoffs' are known, yet usually rejected by the individual speaker designers, such as path delay by Klipsch, or Doppler distortion by Ed V, designer of the AR loudspeaker. You must be your own personal judge of audio components.
My challenge in life, for the last 45 years or so, was to make the best solid state amp in the world, and I have come close. Your life challenge may be something different, but IF you don't have a life challenge, then you may have little idea how much I have personally invested into my goal, GLADLY, because it gives me pleasure to pursue it. Here, I am only sharing some of what I have learned, on the way, and I hope that others will share their serious audio design experiences with me, much like Jack Caldwell just shared on another thread. It is interesting to learn for its own sake, at least it is for me.
Each major switch involved 'trade-offs', some of which I evolved into, yet I might miss the 'dynamics' of a 3 way horn loaded system, but the 3 way K-horn would annoy me in other respects. Each and every one of these 'tradeoffs' are known, yet usually rejected by the individual speaker designers, such as path delay by Klipsch, or Doppler distortion by Ed V, designer of the AR loudspeaker. You must be your own personal judge of audio components.
My challenge in life, for the last 45 years or so, was to make the best solid state amp in the world, and I have come close. Your life challenge may be something different, but IF you don't have a life challenge, then you may have little idea how much I have personally invested into my goal, GLADLY, because it gives me pleasure to pursue it. Here, I am only sharing some of what I have learned, on the way, and I hope that others will share their serious audio design experiences with me, much like Jack Caldwell just shared on another thread. It is interesting to learn for its own sake, at least it is for me.
[snip]There is no 'very best' circuit design or topology in high end audio. Its all in the listeners head, and that's what determines whether its high end or not. I can promise you, if you spend 10 grand on a system it WILL sound good to your ears. If you're student and broke and you hack something togther on a piece of stripboard for $20, it WILL sound awsome. Period.
Refreshing....
jan
imo, every pro designer has the right to stand behind his choices/preferences, and it would be rather strange if he didn't.
No one should be ridiculed or criticized for having a preference.
Personally, i don't hold a boxer as the ideal engine configuration, mainly because i've had to swallow a lot of combustion engine theory.
But if someone said a boxer is the preferred engine choice for the back of a car, right inbetween the rear axles, i'd agree or respect the choice.
Amusing thing, Mr Bonsai has a power amp schematic on his homepage, which he refers to as equivalent to a 10k$ commercial amp.
The custom wound 2KVA toroidal transformer he used for the power amp alone would add several thousand dollars to the retail cost of a Western manufacture commercial product.
Reading words of a preference as the ultimate truth and then criticising the statement is amusing, and seems dyslectic.
Trivia :
for fun, i've been calculating stuff like $/W and retail-$ per output device figures over the last 30 years, of just about any commercial power amplifier in the upper tiers of the audio market.
Amusingly, despite the current wide spread of power amps in the 10k$ and higher range, once corrected to present value, factors as MSRP $/W and $/device have not gone up, but on average gone down.
(same story with automobiles, btw, MSRP$/HP has never been lower)
My audio buddy mentioned long time ago that he rather spend the amount of money for a car on audio gear, much lower depreciation, the car would be as good as worthless six years later.
Last edited:
N-channel Complementary IPS
The complementary JFET input stage popularized by John is indeed elegant, but the lack of suitable complementary dual JFETs makes it difficult and expensive to put into products these days. It also requires good JFET matching to achieve its best performance. Fortunately, there is more than one way to skin a cat.
The circuit below is after Figure 7.13 in my book Designing Audio Power Amplifiers. It achieves the same functionality while using only N-channel JFETs. Q3 and Q4 are common-base stages that harvest the source currents of input JFETs Q1 and Q2. R1 and R2 provide source degeneration. The key to circuit operation is to drive the bases of Q3 and Q4 with an accurate replica of the common mode voltage of the input signal. This is accomplished by emitter followers Q5 and Q6 which sum the source voltages of Q1 and Q2 to drive the bases of Q3 and Q4 with the replica common mode signal. Current source I1 conveniently sets the stage operating currents by creating a DC voltage drop across R3 and R4 that is also forced across source degeneration resistors R1 and R2.
Cascodes Q7 and Q8 complete the IPS cell, resulting in cascode outputs to both the top and bottom loads of the cell. Load resistors R5-R8 are just illustrative, and the stage can be used as the IPS in many different complementary architectures, including folded cascodes. The circuit is equally suitable for use in line-level stages and power amplifier input stages.
It is notable that, with the exception of the small base currents of Q5 and Q6, the same current of each input transistor is conveyed to the top and bottom loads, making the transconductance to the top and bottom essentially the same. This is sometimes not the case with the classical complementary JFET IPS because the gm of the P and N-channel JFETs may not be the same for the same operating current.
In fairness to the classic JFET complementary IPS, this stage will often have slightly higher noise, since the classic design effectively places two pairs of input JFETs in parallel and can operate with no source degeneration if so desired. The source degeneration shown in this circuit can of course be made smaller.
Cheers,
Bob
The complementary JFET input stage popularized by John is indeed elegant, but the lack of suitable complementary dual JFETs makes it difficult and expensive to put into products these days. It also requires good JFET matching to achieve its best performance. Fortunately, there is more than one way to skin a cat.
The circuit below is after Figure 7.13 in my book Designing Audio Power Amplifiers. It achieves the same functionality while using only N-channel JFETs. Q3 and Q4 are common-base stages that harvest the source currents of input JFETs Q1 and Q2. R1 and R2 provide source degeneration. The key to circuit operation is to drive the bases of Q3 and Q4 with an accurate replica of the common mode voltage of the input signal. This is accomplished by emitter followers Q5 and Q6 which sum the source voltages of Q1 and Q2 to drive the bases of Q3 and Q4 with the replica common mode signal. Current source I1 conveniently sets the stage operating currents by creating a DC voltage drop across R3 and R4 that is also forced across source degeneration resistors R1 and R2.
Cascodes Q7 and Q8 complete the IPS cell, resulting in cascode outputs to both the top and bottom loads of the cell. Load resistors R5-R8 are just illustrative, and the stage can be used as the IPS in many different complementary architectures, including folded cascodes. The circuit is equally suitable for use in line-level stages and power amplifier input stages.
It is notable that, with the exception of the small base currents of Q5 and Q6, the same current of each input transistor is conveyed to the top and bottom loads, making the transconductance to the top and bottom essentially the same. This is sometimes not the case with the classical complementary JFET IPS because the gm of the P and N-channel JFETs may not be the same for the same operating current.
In fairness to the classic JFET complementary IPS, this stage will often have slightly higher noise, since the classic design effectively places two pairs of input JFETs in parallel and can operate with no source degeneration if so desired. The source degeneration shown in this circuit can of course be made smaller.
Cheers,
Bob
Attachments
- Status
- Not open for further replies.