A little off topic but I just posted the circuit for the HP 15118A condenser microphone preamp here: http://www.diyaudio.com/forums/parts/40046-b-k-microphone-power-supply-help-needed-4.html#post2827046
PMA, I think that Thorsten HAS shown you an example of 'improved' response with a higher drive impedance. Perhaps, your speaker likes low drive impedance, so what?
I design my amplifiers with intrinsically low drive impedance. I do this by using power mosfets as DRIVERS for a number of output device pairs operating at 'highish' quiescent current, and then adding negative feedback, yet I do NOT add an output coil.
Now why does this give a number of advantages?
First, my OPEN LOOP output impedance over frequency is very low. Why? Because the mosfets separate the voltage drive stage from the output stage, and the mosfets automatically give low drive Z to the output bipolar pairs that then reduce it yet another 50 times. This means that even without global feedback, I have low output impedance. Then add 30-40 dB of feedback, and the output Z goes to virtually nothing. Now this is not all. By eliminating the output coil, I KEEP the output Z low with frequency, perhaps not perfectly, but better than most.
Now WHY do I design this way? For low output impedance, exclusively? NO!
I design it this way to reduce IIM distortion, and to turn the output devices into TRUE emitter followers, rather than BETA MULTIPLIERS, that most people have, unless they use a triple Darlington output stage. That frees my design from beta non-linearity, to a great extent, and gives constant Z for the voltage drive stage that does most of the voltage swing in the amp. The output coil, all else being equal would increase output Z with frequency by its very nature. I keep it down to a VERY low value at 20KHz, check it out in 'Stereophile'.
I design my amplifiers with intrinsically low drive impedance. I do this by using power mosfets as DRIVERS for a number of output device pairs operating at 'highish' quiescent current, and then adding negative feedback, yet I do NOT add an output coil.
Now why does this give a number of advantages?
First, my OPEN LOOP output impedance over frequency is very low. Why? Because the mosfets separate the voltage drive stage from the output stage, and the mosfets automatically give low drive Z to the output bipolar pairs that then reduce it yet another 50 times. This means that even without global feedback, I have low output impedance. Then add 30-40 dB of feedback, and the output Z goes to virtually nothing. Now this is not all. By eliminating the output coil, I KEEP the output Z low with frequency, perhaps not perfectly, but better than most.
Now WHY do I design this way? For low output impedance, exclusively? NO!
I design it this way to reduce IIM distortion, and to turn the output devices into TRUE emitter followers, rather than BETA MULTIPLIERS, that most people have, unless they use a triple Darlington output stage. That frees my design from beta non-linearity, to a great extent, and gives constant Z for the voltage drive stage that does most of the voltage swing in the amp. The output coil, all else being equal would increase output Z with frequency by its very nature. I keep it down to a VERY low value at 20KHz, check it out in 'Stereophile'.
Hi,
You may believe as you please.
I was talking about broader areas of uneven frequency response. It is easy to take the frequency response, print it out and to then estimate for (say) third octave sections the effect of a 4 Ohm source resistance on frequency response and to then draw the corrections with a coloured pen into the original diagram.
Actually, even that is not needed, it is quite obvious that areas of broad elevated SPL in the FR are mirrored by dips in impedance and the reverse.
Ciao T
You may believe as you please.
I was talking about broader areas of uneven frequency response. It is easy to take the frequency response, print it out and to then estimate for (say) third octave sections the effect of a 4 Ohm source resistance on frequency response and to then draw the corrections with a coloured pen into the original diagram.
Actually, even that is not needed, it is quite obvious that areas of broad elevated SPL in the FR are mirrored by dips in impedance and the reverse.
Ciao T
Of course without any additional resistor in series any real speaker has bumps and dips.
Here is the picture from the same Stereophile:
Manipulation? Misinterpretation of what I have said?
except for the language problem - "intrinsicically low drive impedance" is a practical impossibility
the only "no feedback" way to achieve low output impedance is to have a output stage shunt R much smaller than the intended load - which would soak up many times more power than the load
practical low Z outut stages rely on negative feedback, local and/or global
the only "no feedback" way to achieve low output impedance is to have a output stage shunt R much smaller than the intended load - which would soak up many times more power than the load
practical low Z outut stages rely on negative feedback, local and/or global
John, influence of R to speaker Q is a trivial thing. It may "improve" overdamped response of a single bass driver, but this I have said several times. Anyway, the method is not universal, it works only with some drivers in some kind of enclosures, and in general it is much less universal than a voltage drive. Demian has written a good post covering the problem.
It was me who should ask this question. If you totally agreed with 1audio, what you are arguing about?
My point was, zero output impedance of all universal amps and all speakers designed universally for such amps is not the only valid approach. 1audio said that this approach is obsoleted, you said you agree. What else, Pavel?
Last edited:
1audio said that this approach is obsoleted, you said you agree. What else, Pavel?
This is only a small part of his post. Misquote, again.
No, this is the part of his post that highlights my own position: I always view amps+speakers as the system and don't see such a bright future when all amps will be universal for all speakers, and all speakers will be universal for all amps. So we must deal with reality, instead of idealized models. And Stereophile tests of amps in terms of how close they are to such idealized model to drive idealized speakers are not valid, especially when testing amps especially designed to drive totally different speakers.
Anyway, voltage drive approach of a conventional speaker is much more valid than a current drive or added resistor in series. We may tune our own systems, but then our results are only valid for these unique systems.
Stereophile approach is perfectly valid as it MUST be as universal as possible.
Stereophile approach is perfectly valid as it MUST be as universal as possible.
Now everyone, I know that this sounds like a 'match' between PMA and me, but it is not.
Let me give you an example of the 'tradeoffs' in amp output stage design. Each tradeoff has its own merits and demerits.
Bipolar transistor output: intrinsically low Z (Re+1/B drive Z)
Mosfet output: medium Z, (1/Gm + Rsource)
Collector or Drain output: High Z, must be reduced by feedback.
Now the bipolar output is potentially the lowest Z, BUT drive Z MUST be low to keep it that way. IF you don't, then you have a beta multiplier, rather than an emitter follower.
Charles Hansen does this with a Darlington TRIPLE rather than just a pair. This lowers the output Z perhaps 100 times compared to a Darlington pair.
I use high current operated complementary mosfets that give low intrinsic drive impedance to the output devices.
Either way will work. Now if you add global feedback then the measured output impedance goes even much lower. Charles doesn't do this, so his measured output impedance will be higher than mine, but so what? Charles has it low enough for virtually any practical speaker example.
Now why do I put this up here? Charles Hansen is my greatest competition. He and I virtually design in an area directly competitive to each other. He often wins, more often than I am always comfortable with. But at least it is honest competition. He is of the 'no feedback school' and I of the 'modest feedback school' are in head-to-head competition, but leaving behind many in the 'high feedback school' made in traditional ways with traditional thinking that Charles and I gave up on, perhaps 30 years ago.
Let me give you an example of the 'tradeoffs' in amp output stage design. Each tradeoff has its own merits and demerits.
Bipolar transistor output: intrinsically low Z (Re+1/B drive Z)
Mosfet output: medium Z, (1/Gm + Rsource)
Collector or Drain output: High Z, must be reduced by feedback.
Now the bipolar output is potentially the lowest Z, BUT drive Z MUST be low to keep it that way. IF you don't, then you have a beta multiplier, rather than an emitter follower.
Charles Hansen does this with a Darlington TRIPLE rather than just a pair. This lowers the output Z perhaps 100 times compared to a Darlington pair.
I use high current operated complementary mosfets that give low intrinsic drive impedance to the output devices.
Either way will work. Now if you add global feedback then the measured output impedance goes even much lower. Charles doesn't do this, so his measured output impedance will be higher than mine, but so what? Charles has it low enough for virtually any practical speaker example.
Now why do I put this up here? Charles Hansen is my greatest competition. He and I virtually design in an area directly competitive to each other. He often wins, more often than I am always comfortable with. But at least it is honest competition. He is of the 'no feedback school' and I of the 'modest feedback school' are in head-to-head competition, but leaving behind many in the 'high feedback school' made in traditional ways with traditional thinking that Charles and I gave up on, perhaps 30 years ago.
As I said already many times, I don't accept any extremism no matter how well it looks in someone's imagination. In reality (and in theory as well) no system made of optimal subsystems can be optimal. No speaker exists, and no speaker will exist in near future, that when driven by ideal zero ohm voltage source produces flat frequency response on all audio band.
- Status
- Not open for further replies.