Does anyone has an idea of output stage that has gain G=1? Not 0.99, not 0.9, not 1.1, but exactly 1?
Something like Hawksford EC or NP-PMA, but with minimal component.
I'm looking for simple EC (2 transistors maybe), that this 2 transistors can be used for setting bias, also for local Error Correction that gives G=1 exactly.
Something like Hawksford EC or NP-PMA, but with minimal component.
I'm looking for simple EC (2 transistors maybe), that this 2 transistors can be used for setting bias, also for local Error Correction that gives G=1 exactly.
lumanauw
Does anyone has an idea of output stage that has gain G=1? Not 0.99, not 0.9, not 1.1, but exactly 1?
Something like Hawksford EC or NP-PMA, but with minimal component.
I'm looking for simple EC (2 transistors maybe), that this 2 transistors can be used for setting bias, also for local Error Correction that gives G=1 exactly.
----------------------------------------------------------
ZEN series
Does anyone has an idea of output stage that has gain G=1? Not 0.99, not 0.9, not 1.1, but exactly 1?
Something like Hawksford EC or NP-PMA, but with minimal component.
I'm looking for simple EC (2 transistors maybe), that this 2 transistors can be used for setting bias, also for local Error Correction that gives G=1 exactly.
----------------------------------------------------------
ZEN series
Hi, EVA,
Sometime ago, I remember that you opinioned that what is called "Error Correction" is not clear, because the whole audio amp system is an EC itself
You are right, but this is my thinking.
I wanted to make a "local" EC around output stage, so it can level input and output of that EC output stage as close as it can get to G=1, in order to make the work of the differential is much light-er. If the differential pair works light-er, I feel that the sonics is better.
Something in the spirit of "making the elements as linear as possible before applying feedback", so the differential pair will work much light-er (not burdened by the total mess of the whole amp)
Sometime ago, I remember that you opinioned that what is called "Error Correction" is not clear, because the whole audio amp system is an EC itself
You are right, but this is my thinking.I wanted to make a "local" EC around output stage, so it can level input and output of that EC output stage as close as it can get to G=1, in order to make the work of the differential is much light-er. If the differential pair works light-er, I feel that the sonics is better.
Something in the spirit of "making the elements as linear as possible before applying feedback", so the differential pair will work much light-er (not burdened by the total mess of the whole amp)
That's what I'm thinking, at least I need 2 transistors with emitors tied together or base tied together. Or input to base, and feedback to emitor (current feedback).
But trying to make with those always result in NP-PMA or Hawksford EC.
Do you have any ide to make output stage with "differential" to control input=output, that can follow full swing from +rail to -rail with G=1?
Hi, Jarek,
Thanks. I will study it.
Even an opamp follower is not exactly unity gain.
Making rail to rail and unity gain is very hard as the input is also rail to rail and that usually involves having two input amplifiers, one pnp, the other npn in parallel. The result is typically a nasty gm and input offset glitch somewhere as these front ends hand over.
My thread http://www.diyaudio.com/forums/showthread.php?s=&threadid=71728
could be modified for a gain of -1 (inverting) and does not have this problem but I think its better to have gain and avoid problems earlier.
Making rail to rail and unity gain is very hard as the input is also rail to rail and that usually involves having two input amplifiers, one pnp, the other npn in parallel. The result is typically a nasty gm and input offset glitch somewhere as these front ends hand over.
My thread http://www.diyaudio.com/forums/showthread.php?s=&threadid=71728
could be modified for a gain of -1 (inverting) and does not have this problem but I think its better to have gain and avoid problems earlier.
lumanauw said:
In other words.
Adding a very good local feedback ( Gain near 1.0 )
will reduce the need of a lot of global feedback
and result will be hopefully a better total amplification.
I will follow with interest what this thread can come up with, lumanauw.
There are plenty of output voltage followers, without feedback, error correction.
Because of this they will not be good Gain=0.99 amplifiers.
What I miss is:
A good unity gain power ampifier with feedback.
This would be a blessing!
Imagine letting a good discrete, a good OP-amp or a good tube preamp be voltage amplifier.
And add to this another untity gain power amplifier.
I am talking about two amplifiers.
With only feedback within each amplifier = locally.
I have not managed to make a good (theoretically) unity gain power amp
using feedback, with less than 12 transistors.
It is Class A with feedback from output to input.
And DC voltage is same at output as at input.
I still have to figure out several things, though.
That is why I do not want to post any circuit.
Maybe I can come up with a better solution.
I am surprised not to have found a dedicated unity gain power amp
using error correction, feedback.
The practical possibilities to use such an amp, would be endless!
Apa kabar, Kuala Lumpur? Petronas Tower?
Ahh....That's it . You got it
I think the heaviest burden for a differential pair is to "FIX" the output stage. The VAS, the Drivers, all usually works in classA with bias current much bigger than needed. Not much mess here.
If we can make G=1 output stage, then alot of differential pair's burden will be removed, much less phase mess, better HF properties, better(less) intermodulation problem, better sonics hopefully
Ahh....That's it . You got it
I think the heaviest burden for a differential pair is to "FIX" the output stage. The VAS, the Drivers, all usually works in classA with bias current much bigger than needed. Not much mess here.
If we can make G=1 output stage, then alot of differential pair's burden will be removed, much less phase mess, better HF properties, better(less) intermodulation problem, better sonics hopefully
Hi lumanauw,
I think you are on the right track. I have noticed that the less circuit correction the diff pair has to do, the better the sound is. I've been playing with this idea myself lately. The trick is, you don't need exactly one as a gain. What you need is linearity so the diff pair only needs to touch up DC offset and output voltage. Avoiding the discontinuity correction in the output stage.
Essentially, you want to apply a ton of negative feedback in the output stage alone. Very few phase shifts to worry about this way. Then you can run a linearised voltage amp stage at lower feedback ratios and have a great sounding amplifier. It will measure well this way too.
-Chris
I think you are on the right track. I have noticed that the less circuit correction the diff pair has to do, the better the sound is. I've been playing with this idea myself lately. The trick is, you don't need exactly one as a gain. What you need is linearity so the diff pair only needs to touch up DC offset and output voltage. Avoiding the discontinuity correction in the output stage.
Essentially, you want to apply a ton of negative feedback in the output stage alone. Very few phase shifts to worry about this way. Then you can run a linearised voltage amp stage at lower feedback ratios and have a great sounding amplifier. It will measure well this way too.
-Chris
Khabar baik
I agree that making the output stage linear is the way to go, Trying to solve everthing only with staggering amounts of global feedback just does not seem to work.
The normal emitter follower output is of course an example of feedback in the output and in well designed circuits like the Leach amps works well. The problem is the inefficiency.
I was in Bandung at Indonesian Aerospace last year
(For outsiders Malay and Indonesian are very closely related languages that cause plenty of problems when we think that we understand each other but don't)
I agree that making the output stage linear is the way to go, Trying to solve everthing only with staggering amounts of global feedback just does not seem to work.
The normal emitter follower output is of course an example of feedback in the output and in well designed circuits like the Leach amps works well. The problem is the inefficiency.
I was in Bandung at Indonesian Aerospace last year
(For outsiders Malay and Indonesian are very closely related languages that cause plenty of problems when we think that we understand each other but don't)
Hi, myhrrhleine
I really don't know which Hawksford schematic you mean. I know the Hawksford EC biasing cct (that is used in Etalon power amp).
In Hawksford's paper (the ones that explaining about this EC biasing technique), it is very good solution, with solid mathematical backup. It can produce output stage with G=1 even can be adjusted with G> or <1.
If this is the one you mean, I see an aspect that is not soo good (in my opinion) in this Hawksford EC.
The base of the biasing transistors (2 of them in this Hawksford EC) is connected to output node, while the emitors are connected to front end. The "output" condition is conected to the base of EC transistors, and the "input" is connected to emitors.
Base is the far more sensitive for receiving signals compared to emitors. In hawksford EC, the "output" is more controlling the EC itself than the "input".
If the speaker is fully resistive (not impedance), this Hawksford EC will work fine.
In case of back EMF from speakers (where current and voltage are not in phase, rather in (i+jw) format), this back EMF will control the EC more than the input from the front stage, because this back EMF (with bigger magnitude) is injected direcly to base, while the emitors (which are less sensitive than base in receiving signal) only receives small signal front the front end.
I'm looking for something like this Hawksford EC with altered condition. It can be adjusted for G= or < or >1, but with "input" signal going to base and "output" signal comes to emitors (tilted from Hawksford EC).
I try with paper and pencil here, and haven't succeeded in making 1 drawing yet.
I really don't know which Hawksford schematic you mean. I know the Hawksford EC biasing cct (that is used in Etalon power amp).
In Hawksford's paper (the ones that explaining about this EC biasing technique), it is very good solution, with solid mathematical backup. It can produce output stage with G=1 even can be adjusted with G> or <1.
If this is the one you mean, I see an aspect that is not soo good (in my opinion) in this Hawksford EC.
The base of the biasing transistors (2 of them in this Hawksford EC) is connected to output node, while the emitors are connected to front end. The "output" condition is conected to the base of EC transistors, and the "input" is connected to emitors.
Base is the far more sensitive for receiving signals compared to emitors. In hawksford EC, the "output" is more controlling the EC itself than the "input".
If the speaker is fully resistive (not impedance), this Hawksford EC will work fine.
In case of back EMF from speakers (where current and voltage are not in phase, rather in (i+jw) format), this back EMF will control the EC more than the input from the front stage, because this back EMF (with bigger magnitude) is injected direcly to base, while the emitors (which are less sensitive than base in receiving signal) only receives small signal front the front end.
I'm looking for something like this Hawksford EC with altered condition. It can be adjusted for G= or < or >1, but with "input" signal going to base and "output" signal comes to emitors (tilted from Hawksford EC).
I try with paper and pencil here, and haven't succeeded in making 1 drawing yet.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.