Lo (voice coil inductance in braked state) is of no importance for energy exchange. It is only responsible for impedance increase at higher frequencies. Speaker is an acoustic/mechanic/electric converter and this results in familiar electrical schematic. The energy exchange is caused by acoustic-mechanic values transformed to electric side (as LRC resonant circuit), it has nothing to do with voice coil inductance. These trivialised explanations that can be seen here are missleading.
Charles said:
Your feedback comes from a node that for all intends and purposes mimics the speaker node.
You have found a clever way to obtain speaker node feedback without a galvanic connection.
Calling it zero speaker feedback is misleading.
Jan Didden
good grief
What is charles supposed to call this ?
Zero speaker feedback seems spot on, crystal clear to me.
.....on the basis that, in many different ways, a resistor is not a speaker...😉
mike
What is charles supposed to call this ?
Zero speaker feedback seems spot on, crystal clear to me.
.....on the basis that, in many different ways, a resistor is not a speaker...😉
mike
Speaker EMF and other trivia
I must assume everyone seriously involved in audio design is well aware of the Thiele - Small model for the electromechanical speaker. At least Pavel, Jan and certainly several others here do and this particular needs no further comment.
I am not so sure others do, or do understand the physical meaning for the various electrical - both direct or equivalent - parts of which it is composed.
I will try to condense hopefully as simply as possible this physical meaning with reference to the back EMF, and the importance or lack thereoff of taking care / controlling it. This description of necesity is simplified and not accurate but does correctly highlight the concepts involved.
Imagine a voice coil and piston freely suspended, that is with no restoring force from the cone surround or spider. If we apply a DC voltage, the winding current in the presence of the uniform magnetic field will feel a proportional so called Lorentz force and will begin to move against the mounting air pressure differential (front to back) caused by the associated cone movement.
This winding movement within a magnetic field in time induces an EMF in it, whose value and polarity are - were the coil resistance null - equal and opposite the driving voltage.
The product of the applied (equal to the back EMF) voltage and the coil current has the dimensions of power, and assuming there are no other energy losses, it is completely converted to work/time, numericaly equal at each instant to the product of the coil/cone speed times the opposing force of the compressed air. This work/time is no less than acoustical power.
In short, the speaker EMF **is an integral part of the electrical to acustical energy conversion process **.
Back to the real world, there is a mechanical restoring force we try to make as unobstrusive as practical, there is voice coil leakage inductance bothering in the high end, and voice coil resistance we try to make as small as practical (NOT THE NOMINAL SPEAKER IMPEDANCE).
In the real world, the active part of the electrical energy delivered to the speaker (inner product of voltage and current vectors) goes to sound pressure but unfortunately after diverting a good deal to ohmic heating in the voice coil resistance, hysteresis losses in the magnetic circuit, and to a lesser extent in mechanical non sound related losses in the motor structure.
To have a good control of the speaker voltage (eg. to make the amplifier output impedance as small a possible) is to have a good control (to a fair extent) of the voice coil speed, which (hopefully) translates to well controled sound pressure.
An this is what we are looking for after all.
Sorry for the long post, Rodolfo
I must assume everyone seriously involved in audio design is well aware of the Thiele - Small model for the electromechanical speaker. At least Pavel, Jan and certainly several others here do and this particular needs no further comment.
I am not so sure others do, or do understand the physical meaning for the various electrical - both direct or equivalent - parts of which it is composed.
I will try to condense hopefully as simply as possible this physical meaning with reference to the back EMF, and the importance or lack thereoff of taking care / controlling it. This description of necesity is simplified and not accurate but does correctly highlight the concepts involved.
Imagine a voice coil and piston freely suspended, that is with no restoring force from the cone surround or spider. If we apply a DC voltage, the winding current in the presence of the uniform magnetic field will feel a proportional so called Lorentz force and will begin to move against the mounting air pressure differential (front to back) caused by the associated cone movement.
This winding movement within a magnetic field in time induces an EMF in it, whose value and polarity are - were the coil resistance null - equal and opposite the driving voltage.
The product of the applied (equal to the back EMF) voltage and the coil current has the dimensions of power, and assuming there are no other energy losses, it is completely converted to work/time, numericaly equal at each instant to the product of the coil/cone speed times the opposing force of the compressed air. This work/time is no less than acoustical power.
In short, the speaker EMF **is an integral part of the electrical to acustical energy conversion process **.
Back to the real world, there is a mechanical restoring force we try to make as unobstrusive as practical, there is voice coil leakage inductance bothering in the high end, and voice coil resistance we try to make as small as practical (NOT THE NOMINAL SPEAKER IMPEDANCE).
In the real world, the active part of the electrical energy delivered to the speaker (inner product of voltage and current vectors) goes to sound pressure but unfortunately after diverting a good deal to ohmic heating in the voice coil resistance, hysteresis losses in the magnetic circuit, and to a lesser extent in mechanical non sound related losses in the motor structure.
To have a good control of the speaker voltage (eg. to make the amplifier output impedance as small a possible) is to have a good control (to a fair extent) of the voice coil speed, which (hopefully) translates to well controled sound pressure.
An this is what we are looking for after all.
Sorry for the long post, Rodolfo
mikelm said:
Oh, it is crystal clear allright. It is also wrong.
Cut the feedback and see how the speaker now reacts way different.
Or take out the dummy output stage (which changes the feedback) and see how the speaker reacts way different.
Or change the dummy load (which changes the feedback) and see how the speaker reacts way different.
Jan Didden
Harking back to definitions, Charles' amplifier has:
1. Voltage feedback from the LTP input and VAS since these sections are shared between the working output stage and the dummy output stage.
2. Voltage feedback from the dummy output stage, which, being a unity gain current amplifier, and with suitable resistive loading, should take care of the largely current driven non-linearities in the typical output stage of any amplifier. Presumably this (global) feedback would be no more successful at scotching crossover errors than any other, however.
3. No feedback from the working output stage, which leaves the non-linearities driven by phase shift and the back emf uncorrected.
All this may be good, may be bad. From my limited experiments in this area, it will make the amp sound more organic, more twee; in short, a little like a tube amp since damping factor cannot be the same and thus we rely more on the mechanical damping of the speaker system. Incidentally, this amp might sound a tad better with the sorts of speakers which sound good with tube amplifiers.
An extremely interesting result of the SPLIF, however, is that since the global feedback loop has only to deal with a resistive, non-reactive load at its dummy output stage, the lag compensation regime to ensure stability can be much less limiting since there will be far less phase shift across the amplifier. Nonetheless, it goes without saying that the working output stage has to be stable in and of itself - not too difficult with base/gate stoppers and good layout - but potentially it should be possible to realize greater speed in this amplifier.
Should we call this 'dummy global feedback', Jan, or 'global dummy feedback'?
Cheers,
Hugh
1. Voltage feedback from the LTP input and VAS since these sections are shared between the working output stage and the dummy output stage.
2. Voltage feedback from the dummy output stage, which, being a unity gain current amplifier, and with suitable resistive loading, should take care of the largely current driven non-linearities in the typical output stage of any amplifier. Presumably this (global) feedback would be no more successful at scotching crossover errors than any other, however.
3. No feedback from the working output stage, which leaves the non-linearities driven by phase shift and the back emf uncorrected.
All this may be good, may be bad. From my limited experiments in this area, it will make the amp sound more organic, more twee; in short, a little like a tube amp since damping factor cannot be the same and thus we rely more on the mechanical damping of the speaker system. Incidentally, this amp might sound a tad better with the sorts of speakers which sound good with tube amplifiers.
An extremely interesting result of the SPLIF, however, is that since the global feedback loop has only to deal with a resistive, non-reactive load at its dummy output stage, the lag compensation regime to ensure stability can be much less limiting since there will be far less phase shift across the amplifier. Nonetheless, it goes without saying that the working output stage has to be stable in and of itself - not too difficult with base/gate stoppers and good layout - but potentially it should be possible to realize greater speed in this amplifier.
Should we call this 'dummy global feedback', Jan, or 'global dummy feedback'?
Cheers,
Hugh
Charles,
You said the output impedance depends on the dummy load, darkferinz's simulation seem to tell that there's a quite tiny difference.
This picture is from his simulation where he forgot to put in any dummyload:
Infinit impedance dummy
and
8 Ohm dummy
Regarding the output impedance, you could find it out, why not saying with an 8 Ohm nominal load if you want to specify with a certain load.
Saying you don't have any idea could mean anything between zero and infinite, so that's why we find out some "standards" how and under what circumstances something should be performed to find out comaparable/referencing data.
There's standards for almost everything we humans design etc...
The output stages works ONLY under the same conditions when you have a load and a dummy load that corresponds to each other.
In a reactive area of the the speaker load you could have 1A runing through every each transistor while for the dummy it's 0A when the voltage is crossing the zero, under these circumstances the output transistor and the dummy output transistor are totally diffrent, the hfe, transconductance, ft.. everything is diffrent which gives a certain distortion pattern for every each working point.
If the output impedance is, for instance, 1 Ohm with a SPLIF modified amplifier but with a conventional FB 0,1 Ohm, well we could use the conventional FB amplifier and add a 1 Ohm resistor in series with the output, this resistor is far more linear than any transistor but gives the same output impedance... and who knows, maybe quite the same sound, but NOT the same of course as long as there is no transistor acting like an linear resistor with current amplification.
BTW why not giving it a shot, add a 1-2 Ohm resistor or more in series with the output on a conventiional FB amplifier, tha's so simple that even I might try it out! 😉
New listening experience is waiting...
What I try to say that adding a SPLIF topology to an amplifier not designed with very low output impedance even without FB will give lot of distortion.
Still the SPLIF have hope, no doubt, but the amplifier total design should be redesigned to a much greater extent if we want to reach any reasonable output impedance which means lot of output transistors, high biasing etc., then I think it can start to be interesting and have some potential.
But modifying conventional FB amplifiers.. 🙄
My first thought in the beginning when the SPLIF came up in this thread for some week ago was Susan Parkers verry simple amplifier with quite high output impedance which also seem to sound verry good by some, you should check in that thread because your amplifier designs have in a way something big in common! 😎
How about the "speaker cable problem running around the whole room" in combinationa with a conventional FB amplifier, I asked it before, could you tell what you think is the problem more specifically, what signals kan be picked up by the cable and disturb the FB circuitry function etc.?
Cheers! 😉
You said the output impedance depends on the dummy load, darkferinz's simulation seem to tell that there's a quite tiny difference.
This picture is from his simulation where he forgot to put in any dummyload:
Infinit impedance dummy
and
8 Ohm dummy
Regarding the output impedance, you could find it out, why not saying with an 8 Ohm nominal load if you want to specify with a certain load.
Saying you don't have any idea could mean anything between zero and infinite, so that's why we find out some "standards" how and under what circumstances something should be performed to find out comaparable/referencing data.
There's standards for almost everything we humans design etc...
The output stages works ONLY under the same conditions when you have a load and a dummy load that corresponds to each other.
In a reactive area of the the speaker load you could have 1A runing through every each transistor while for the dummy it's 0A when the voltage is crossing the zero, under these circumstances the output transistor and the dummy output transistor are totally diffrent, the hfe, transconductance, ft.. everything is diffrent which gives a certain distortion pattern for every each working point.
If the output impedance is, for instance, 1 Ohm with a SPLIF modified amplifier but with a conventional FB 0,1 Ohm, well we could use the conventional FB amplifier and add a 1 Ohm resistor in series with the output, this resistor is far more linear than any transistor but gives the same output impedance... and who knows, maybe quite the same sound, but NOT the same of course as long as there is no transistor acting like an linear resistor with current amplification.
BTW why not giving it a shot, add a 1-2 Ohm resistor or more in series with the output on a conventiional FB amplifier, tha's so simple that even I might try it out! 😉
New listening experience is waiting...
What I try to say that adding a SPLIF topology to an amplifier not designed with very low output impedance even without FB will give lot of distortion.
Still the SPLIF have hope, no doubt, but the amplifier total design should be redesigned to a much greater extent if we want to reach any reasonable output impedance which means lot of output transistors, high biasing etc., then I think it can start to be interesting and have some potential.
But modifying conventional FB amplifiers.. 🙄
My first thought in the beginning when the SPLIF came up in this thread for some week ago was Susan Parkers verry simple amplifier with quite high output impedance which also seem to sound verry good by some, you should check in that thread because your amplifier designs have in a way something big in common! 😎
How about the "speaker cable problem running around the whole room" in combinationa with a conventional FB amplifier, I asked it before, could you tell what you think is the problem more specifically, what signals kan be picked up by the cable and disturb the FB circuitry function etc.?
Cheers! 😉
hi all
one more question:
what is the standard damping factor of conventional amps??
as the signal from the speaker goes back to differential stage, it should be equal to amplifier's open loop gain, right?
so is it kind of thousends...
as far as i know many 'best' sounding amps (including tubes) have damping factor below, say 40
maybe damping factor this huge is not good???
maybe sensible damping factor is the whole 'magic' of splif???
maybe in audio systems with separate subwoofer amp there is no point in huge damping factor as 'softer' bass does not harm sound???
seems a bit alike THD issue
THD->0 then better sound ..... not very much true
damping factor->infinity then better sound..... ??????????
let's stop following parameters forever!!
cheers
one more question:
what is the standard damping factor of conventional amps??
as the signal from the speaker goes back to differential stage, it should be equal to amplifier's open loop gain, right?
so is it kind of thousends...
as far as i know many 'best' sounding amps (including tubes) have damping factor below, say 40
maybe damping factor this huge is not good???
maybe sensible damping factor is the whole 'magic' of splif???
maybe in audio systems with separate subwoofer amp there is no point in huge damping factor as 'softer' bass does not harm sound???
seems a bit alike THD issue
THD->0 then better sound ..... not very much true
damping factor->infinity then better sound..... ??????????
let's stop following parameters forever!!
cheers
Hi, Charles,
My amp has a feedback voltage divider of 10k+1k. This makes 11k in series.
What if I dont use additional dummy load (like 16ohm), but just using this 11k for dummy load, will your SPLIF amp works? Or it needs lower ohm, like 16ohm, for SPLIF idea to work?
My amp has a feedback voltage divider of 10k+1k. This makes 11k in series.
What if I dont use additional dummy load (like 16ohm), but just using this 11k for dummy load, will your SPLIF amp works? Or it needs lower ohm, like 16ohm, for SPLIF idea to work?
AKSA said:Harking back to definitions, Charles' amplifier has:
1. Voltage feedback from the LTP input and VAS since these sections are shared between the working output stage and the dummy output stage.
2. Voltage feedback from the dummy output stage, which, being a unity gain current amplifier, and with suitable resistive loading, should take care of the largely current driven non-linearities in the typical output stage of any amplifier. Presumably this (global) feedback would be no more successful at scotching crossover errors than any other, however.
3. No feedback from the working output stage, which leaves the non-linearities driven by phase shift and the back emf uncorrected.
All this may be good, may be bad. From my limited experiments in this area, it will make the amp sound more organic, more twee; in short, a little like a tube amp since damping factor cannot be the same and thus we rely more on the mechanical damping of the speaker system. Incidentally, this amp might sound a tad better with the sorts of speakers which sound good with tube amplifiers.
An extremely interesting result of the SPLIF, however, is that since the global feedback loop has only to deal with a resistive, non-reactive load at its dummy output stage, the lag compensation regime to ensure stability can be much less limiting since there will be far less phase shift across the amplifier. Nonetheless, it goes without saying that the working output stage has to be stable in and of itself - not too difficult with base/gate stoppers and good layout - but potentially it should be possible to realize greater speed in this amplifier.
Should we call this 'dummy global feedback', Jan, or 'global dummy feedback'?
Cheers,
Hugh
No that's Global Feedback, Dummy. Sorry... couldn't resist.
Seriously, WRT to your analysis Hugh, it appears you are
one of the few here who has actually got their lateral thinking
cap on. As always, it's easy to look for and find, theoretical
flaws but it appears there are some real benefits.
However I still feel that linearising the OL OP stage and
running the whole thing without GF is the best approach.
This way miller compensation can be eliminated altogether.
I should have some answers to these theories this weekend
time permitting.
Cheers,
Terry
If the quality of you amplifier depends on how accurately the load represented by the speaker is duplicated, why not just use the real speaker? I think that's the idea behind feedback.
Drawing feedback from the output is not going to amplify the feedback. It's going to amplify the difference between the input(what the output should be), and the output (what it actually is.) If I'm not mistaken, amplifiers evolved to feeback designs to cure defficiencies with non-feedback designs. Why would anybody want to use many many more components to make a more poor quality amplifier?
Drawing feedback from the output is not going to amplify the feedback. It's going to amplify the difference between the input(what the output should be), and the output (what it actually is.) If I'm not mistaken, amplifiers evolved to feeback designs to cure defficiencies with non-feedback designs. Why would anybody want to use many many more components to make a more poor quality amplifier?
hummhoom said:
Some of us find ( regardless of cost ) listening to many solid state feedback amps a mildly painful experience...🙁 ..
..
Recently at a hi fi show as I went into many big name rooms my first thought was Wow, that sounds impressive. Then I would ask myself - But am I really enjoying it and in most cases the answer was no.
In one or two rooms the sound was sublime - not perfect but allowing the music flow and move me.
So the motivation to try things like this is to get some relief from that uninvolving, mechanical, unpleasant sound that so often accompanies solid state and feedback...🙂
I believe that it is possible for FB amps to sound very good - it's just that most of them don't.
If I had an amp that this mod could be done on I would try it immediately to see how it sounds.
My personal theory is that the speaker leads acting as RF aerials and amplifier stability are the main reasons that this splif modification may improve the subjective sound of many amps despite the down grading of the specs
mike
Mike,
I strongly suggest you find a good SS amplifier and do it - perhaps a Leach, or maybe a JLH mosfet, or even an AKSA!!
Cheers,
Hugh
I strongly suggest you find a good SS amplifier and do it - perhaps a Leach, or maybe a JLH mosfet, or even an AKSA!!
Cheers,
Hugh
lumanauw said:
Hi Lumanauw,,
I think I don't exactly understand what you mean. Do you have a schematic of that amp ?
Thanks, Charles 🙂
Re: Speaker EMF and other trivia
I am quite unsure if I am looking after this.
I think I am not looking after this at all.
See my graphical explanation below:
Conventional amplifier with feedback from speaker
Amplifier (i.e. SPLIF) with no feedback from speaker
See the difference ?
Just joking...
Charles 😉
ingrast said:
I am quite unsure if I am looking after this.
I think I am not looking after this at all.
See my graphical explanation below:
Conventional amplifier with feedback from speaker
Amplifier (i.e. SPLIF) with no feedback from speaker
See the difference ?
Just joking...
Charles 😉
hi lumanauw
look at fftshttp://www.diyaudio.com/forums/showthread.php?postid=519995#post519995
note harmonic distortion
if you think it won't destroy the sound quality then why losing efficiency???
look at fftshttp://www.diyaudio.com/forums/showthread.php?postid=519995#post519995
note harmonic distortion
if you think it won't destroy the sound quality then why losing efficiency???
I have tried doing that before. It does actually improve the sound of quite a few speakers, but only because of the increased output at resonance which is better accomplished with EQ.Ultima Thule said:
Re: Re: Speaker EMF and other trivia
Whether you - and others for sure - prefer unnatural reproduction of program source is up to your taste.
I for one, should rather have a 0 THD and 0 output impedance amplifier and preprocess the source to suit my tastes rather than let this be left to the vagaries of an uncontroled setup of poorly understood behaviour.
Rodolfo
ingrast[/i] To have a good control of the speaker voltage (eg. to make the amplifier output impedance as small a possible) is to have a good control (to a fair extent) of the voice coil speed said:
Whether you - and others for sure - prefer unnatural reproduction of program source is up to your taste.
I for one, should rather have a 0 THD and 0 output impedance amplifier and preprocess the source to suit my tastes rather than let this be left to the vagaries of an uncontroled setup of poorly understood behaviour.
Rodolfo
AKSA said:
Hugh,
I think you are by now aware that I have some serious reservations about the usefullness of this concept. 😉
But, assuming you want to try to get rid of as much of the output stage non-linearities in this peculiar way, from an engineering pov there may be a way to do it without the cost of the dummy output stage.
Instead of the dummy (heavy duty) output stage with a relative heavy load to mimic the currents in the real output stage, one could use a "dummy" output stage built from a couple of cheap small signal transistors, with a light load, possibly only loaded by the fb network.
The idea being that the non-linearities of a small-signal output stage with light load would mimic the non-linearities of a heavy-duty output stage with a heavy load. The effect would then be the same.
I am not sure whether this down-scaling (because that's really what it is) keeps the various parameter ratios intact sufficiently, but it may be a very cheap way to do this.
Jan Didden
As I think about it more, I would be inclined to think of
the SPLIF as a local feedback amplifier. The dummy
output stage and load mimic the actions of quite a
few other local feedback designs.
the SPLIF as a local feedback amplifier. The dummy
output stage and load mimic the actions of quite a
few other local feedback designs.
- Status
- Not open for further replies.