Jan,
I am not quite sure that this "back-EMF injection" makes any problem to a properly designed amp.
I am not quite sure that this "back-EMF injection" makes any problem to a properly designed amp.
There could be a logic trail here worth noting.
1. Global negative feedback amps would appear to sound different to local only or zero feedback amps.
2. There are diverse views about why, but it would appear to come back to phase shift from the reactive speaker load being passed to the feedback node.
3. Charles' idea uses two otherwise identical output stages, one feeding the reactive speaker load and the other a resistive load. Clearly this dummy load will elicit no phase shift at the load, though it will account for device characteristics within the other output stage because the components used are identical.
4. Ergo, the usual load induced phase shift presented to the feedback node in a conventional amplifier is therefore removed, thus eliminating the need for complex lag compensation regimes to ensure stability in the more conventional design. In my experience lag compensation largely determines the sonics of any amplifier, and lag compensation is normally de rigeur in a global negative feedback circuit block.
Charles holds a patent on this idea, which is novel and original to my mind, so detailed analysis and discussion here should be possible since the purpose of a patent is to bring good ideas to the fore. Perhaps a naysayer with a little time on his hands should build it and measure it to within a micron of its root Hertz to verify one way or the other?
Besides, any man who improves a Harley Davidson ignition system can't be all bad..........
Cheers,
Hugh
1. Global negative feedback amps would appear to sound different to local only or zero feedback amps.
2. There are diverse views about why, but it would appear to come back to phase shift from the reactive speaker load being passed to the feedback node.
3. Charles' idea uses two otherwise identical output stages, one feeding the reactive speaker load and the other a resistive load. Clearly this dummy load will elicit no phase shift at the load, though it will account for device characteristics within the other output stage because the components used are identical.
4. Ergo, the usual load induced phase shift presented to the feedback node in a conventional amplifier is therefore removed, thus eliminating the need for complex lag compensation regimes to ensure stability in the more conventional design. In my experience lag compensation largely determines the sonics of any amplifier, and lag compensation is normally de rigeur in a global negative feedback circuit block.
Charles holds a patent on this idea, which is novel and original to my mind, so detailed analysis and discussion here should be possible since the purpose of a patent is to bring good ideas to the fore. Perhaps a naysayer with a little time on his hands should build it and measure it to within a micron of its root Hertz to verify one way or the other?
Besides, any man who improves a Harley Davidson ignition system can't be all bad..........
Cheers,
Hugh
janneman said:
Jan, you still look at it as if it was a feedback amplifier.
But it is a no-feedback amplifier. Of course there is no control over the voltage at the speaker. If there was, it would be a conventional feedback amplifier.
BUT IT IS NO FEEDBACK. NO FEEDBACK -> NO CONTROL ABOUT VOLTAGE AT SPEAKER POST !!!!!!!!
You keep on asking me where to fill in gasoline, but its an electric motor !!!
You cannot measure distortion at the speaker output terminal. This is a short-sighted nonsense belief.
Drop it.
If you want to measure distortion you have to do it after the speaker with a microphone.
You do not listen to the speaker output, you listen to sound.
Also: there are different types of loudspeakers. I agree, that some flabby bass speakers will sound even more flabby with a no-feedback amplifier. That is not a fault of the amplifier. It's a fault of the speaker.
On the other hand, there are speakers that refuse to sound good with ANY conventional feedback amplifier, but work perfectly with no-feedback amps.
My amplifier topology is not designed to make mediocre would-be-hifi sound even more mediocre.
Charles 😉
Charles said:
Charles,
You can try using all capital letters, but that doesn't change the facts. You use identical output stages, one, driving a resistive load, wrapped within a global fb loop, and another driving a speaker. Because the output stages are identical, driven from the same pre-stage, it cannot be else than that the global feedback influences (if you have something against the word 'control' ) the speaker driving output as well. It is not zero feedback.
I agree that there is no back-EMF injection in the feedback loop, other than the feedthrough back to the driver stage. But also, the feedback does less to linearize the speaker driving output than it does to the resistive load. So, you win at one place and lose at another. You clearly feel that the outcome is positive, I don't agree simply because I think the back-EMF is not really a problem in a competently designed amp, while the forward non-linearity of an open loop output stage in class (A)B is.
But please, don't insult our intelligence by calling it non-feedback.
Charles said:
You really have to get your stuff together Charles! You listen to speakers which are electro-mecahnical contraptions designed to convert an electrical signal into a mechanical movement so that the mechanical vibrations can be transmitted through a medium , generally air, to be sensed by hearing mechanisms. These vibrations, produced by said electro-mechanical converter, are generally referred to as 'sound'.
Have a nice weekend,
Jan Didden
hallo guys
i think you should stop arguing and realize why charles' amps are not exactly
the same as conventional amp +1 extra (redundant) O/P stage
the point is (charles isn't very precise i feel)
some of you treat a speaker as a (passive) load itself...
but in fact speaker is not only the passive electrical device
because of some mechanical issues (machanical resonances, playing as a microphone-
it actually is a dynamic microphone, etc.) the speaker is rather a (reactive)
passive load combined with (serial to??) a hard to determine voltage source
in conventional configuration all these voltage signals are connected to the
- input of a diffamp (charles says:the speaker speaks back) and are totaly
damped. total control of the speaker. high 'fidelity' indeed.
in splif configuration there is no damping (except the damping factor of the amp, which is a
very low output impedance device) of these signals. the wind blows- the speaker's membraine
is going to move a bit...
the fidelity is lower
'the amp is not trying to take full control of the speaker'- charles would say
thus => the speaker is making some things more (or less or different) than
driving electrical signal forces thus => the speaker distorts
it is exactly the definition of distortion: not exactly the same
signal
so if splif does sound better to enybody's ear, the point is:
what (magic) types of distortion (linear, non-linear: symetric or not, resonances, microphone-like
behaviours, reflected waves co-incidence, etc...)
does a 'non-fedback' speaker make which makes the sound better????
... much...much...confusion...
???
what does it mean that music is free or released???
maybe some acoustic measurements would show the vital mystical difference
between all these non-global-feedback and global-feedback amps
i think you should stop arguing and realize why charles' amps are not exactly
the same as conventional amp +1 extra (redundant) O/P stage
the point is (charles isn't very precise i feel)
some of you treat a speaker as a (passive) load itself...
but in fact speaker is not only the passive electrical device
because of some mechanical issues (machanical resonances, playing as a microphone-
it actually is a dynamic microphone, etc.) the speaker is rather a (reactive)
passive load combined with (serial to??) a hard to determine voltage source
in conventional configuration all these voltage signals are connected to the
- input of a diffamp (charles says:the speaker speaks back) and are totaly
damped. total control of the speaker. high 'fidelity' indeed.
in splif configuration there is no damping (except the damping factor of the amp, which is a
very low output impedance device) of these signals. the wind blows- the speaker's membraine
is going to move a bit...
the fidelity is lower
'the amp is not trying to take full control of the speaker'- charles would say
thus => the speaker is making some things more (or less or different) than
driving electrical signal forces thus => the speaker distorts
it is exactly the definition of distortion: not exactly the same
signal
so if splif does sound better to enybody's ear, the point is:
what (magic) types of distortion (linear, non-linear: symetric or not, resonances, microphone-like
behaviours, reflected waves co-incidence, etc...)
does a 'non-fedback' speaker make which makes the sound better????
... much...much...confusion...
???
what does it mean that music is free or released???
maybe some acoustic measurements would show the vital mystical difference
between all these non-global-feedback and global-feedback amps
darkfenriz said:
I think of it in several ways:
Just because a speaker is a nonlinear impedance, it will not perform better, if we hold the amp's output voltage in constant (gain) ratio to the input voltage (through feedback).
Who has started claiming this ? I think nobody has every claimed it, but it slipped into our thinking.
Of course you can specify an amplifier's performance by saying how good it can keep the output voltage stable with varying and reactive loads.
Maybe this would be of importance in power supply applications, where you can switch on several light bulbs and don't want that the other light bulbs go dim.
We are not listening to the voltage at the speaker post, but to the sound that comes out of the speaker. So any definition of distortion that is derived by looking at the amp's output voltage under dynamic load is not so directly related to fidelity as one may think.
It may be true for some speakers, but it is definitely untrue for many speakers.
Another aspect of a conventional feedback amplifier is that you build up a highly reactive antenna.
Think about it this way: The feedback line goes from the output stage to the input stage diff-pair.
In fact it goes through your whole listening room, from the diff-pair out of the amp to the speaker through the crossover and into the voice coil. This long line picks up not only any mechanical speaker back movements and decaying oscillations (for correction) but also electric fields.
It all ends up at your input diff-pair which tries to computate speaker movements that stem from long gone excitations plus voltages induced by electrical stray fields with fresh music that comes in.
And this is the point where 'flow' enters the playfield.
The Altmann SPLIF amp has a switch provision, that lets you switch from a completely conventional amp to a Split-Feedback amp with zero feedback derived from the speaker.
If you just play a single piano-, guitar- of whatever note through this amp and listen to it in both switch settings (conventional and no-feedback) you know the difference 🙂
And it truly is a big difference.
If you have an old guitar tube amplifier with a presence knob, you can try it out. Presence was coined as the opposite word for feedback. The presence pot gradually disconnects the feedback line from the output transformer.
Feedback enabled -> Presence zero.
Feedback disabled -> Presence full on.
Listen to the difference in flow.
Charles 🙂
Charles said:
I don't think this slipped at all.
When tubes were the only means of amplifying an audio signal it became quite obvious that the *power* delivered to the load would *vary* with the load impedance in two important ways.
First, as the impedance went above the *optimal* matching load, the power would drop.
Second, as the impedance went below the *optimal* matching load the power would drop.
This would yield a quite non-flat response - anywhere in the band where this occurred.
Feedback helped to fix this issue somewhat.
No slip.
_-_-bear
bear said:
I don't think this slipped at all.
When tubes were the only means of amplifying an audio signal it became quite obvious that the *power* delivered to the load would *vary* with the load impedance in two important ways.
First, as the impedance went above the *optimal* matching load, the power would drop.
Second, as the impedance went below the *optimal* matching load the power would drop.
This would yield a quite non-flat response - anywhere in the band where this occurred.
Feedback helped to fix this issue somewhat.
No slip.
_-_-bear
Hi bear,
if you put constant power into a speaker this means in no way, that you get a flat response.
As I suggested earlier, there are speakers that were designed to work with no-feedback amplification.
Take for example the Altec 416 bass driver in a VOTT system).
It virtually refuses amplifier control (that's at least my experience). It was designed to sound great on its own.
A real high quality speaker driver construction will bring excellent results with no-feedback amplification.
Take a Lowther driver and try to use it with a feedback amplifier... or better try not 😉
Charles 🙂
JC has a good solution on this. He put MOSFET for driver. It blocks what happened at the output from feeding back to the front end.
Just a dummy question. If the feedback is taken in the VAS (not after the final stage) and it uses mosfet as drivers, will it sounds like full non feedback?
lumanauw said:
Luminauw,
I'm not sure who this question is directed at, but I think the first part of the quote was from me.
There are no absolutes in audio design, as nowhere in life. Having a low output Z as output stage driver will minimise reverse feedthrough of speaker influence.
If you take the fb from the Vas, you will get a Vas voltage that is very low distortion (because of the *global Vas fb*), and that drives an open loop output stage. So, output stage non-linearities are not corrected. Especially in class (A)B stages, these non-linearities increase in relative amplitude with lower output signals. To use a term of the one and only, the First Watt gets much more corrupted than the 20est Watt. That's the output stage itself, but the speaker is also non-linear in impedance, so that also influences the output voltage.
Now, if you wrap the whole amp (including the output stage) in a fb loop, you also linearise the output stage, and you isolate it from speaker non-linearities. If the speaker behaves non-linear, the fb will try to keep the output voltage at the correct value irrespective.
To do that, the fb has to adapt the error signal driving the input, so you will find the speaker's influence in the feedback signal. That is no shortcoming as such; you have to tell the corrector what is wrong so it knows what to correct, so to speak.
Now, there is a school of thought that says this is a Bad Thing, and that it actually makes the amplifier sound worse than without global feedback. This is not founded in measurements; open loop non-linearity is measurable (much) higher than closed loop residual non-linearity. But many people report differences in listening tests, and say that the non-global fb sounds better.
As you may know, I am very sceptical in this: firstly, I am still searching for controlled tests that unequivocally show that indeed there are differences. If so, we can look at which is preferred, which is not necessarily the better one in terms of more faithfully reproducing the original signal.
Now Charles idea, which I agree is pretty shrewd, is somewhere in between. If the speaker was identical with the resistive load, it would correct for the amplifier non-linearities, and the loads being the same, the non-feedback output stage would act exactly as the feedback-output stage. But, that means we can connect the two output stages (as they have identical signals) and this shows that in effect we just have global feedback on both outputs!
But, the speaker is non-linear. So, the speaker effects on the open-loop output are still there. Some aspects of the amp are corrected, but the speaker influence is not. It is somewhere between global and zero feedback. I personally think it isn't worth the extra effort, as the non-linear speaker influence on the open-loop stage will swamp any small feedback-back-influence you would have if you used full global feedback.
Jan Didden
Charles said:
Charles,
I don't mean to be picky because this thread has evolved in
an interesting direction but the comment WRT presence knob
on tube amps is not strictly correct.
Marshall (and early Fender?) GTR amps had 'presence' control &
which did reduce feedback but also contoured frequency
response to give a significant upper mid / low treble boost.
This why they called it presence because it allowed the tone
to cut through a bit more.
However, I have built GTR amps and compared various amounts
of (flat) FB back to phase splitter and I also find they sound
better open loop.
GTR amps are not such a good example though since many
things such as high leakage inductance, low bandwidth OP
transformer also sound good, especially when amp is "cranked
up" ie OP driven into distortion.
Cheers,
Terry
(former Marshall 'plexi' super lead 100 owner)
Terry Demol said:
Hi Terry,,
early bassman and superlead have a presence pot that gradually connects a grounded capacitor to the cathodes of the phase splitter, providing an adjustable AC shunt to ground.
Presence knob full on -> full AC shunt to ground -> no feedback.
Presence knob full off -> feedback derived from secondary through 27k/5k voltage divider.
I once loaded the schematics from www.schematicheaven.com.
Maybe later models used some tone-control in conjunction with the precence knob, however not in the early application.
I just find it significant, that (maybe Leo Fender) coined the term 'Presence' by just listening to what it does, and really if you gradually introduce feedback the presence of the tone is reduced...
Charles 🙂
Charles said:
Charles,
I don't know why you would find this 'significant'? Presence means that the mid/ low-treble region is lifted wrt the rest of the freq response. It creates the illusion that the main performer, often human voice, is closer because it uses the well known effect that higher level makes us believe the source is closer. Presence is nothing more than linear distortion introduced by freq dependent amplification.
Introducing more feedback makes the freq response more flat, reduces the non-linear distortion that caused the presence.
Elementary.
Jan Didden
janneman said:
Hi Jan,,
yes, the tube amp will have a little more harmonic distortion with the feedback line AC-grounded.
But something else is also happening, that I refferred to as 'flow' in earlier posts.
In order to obtain some real-world experience on this issue, I recommend building a Splif amplifier 🙂
You can insert a switch that lets you toggle between completely conventional amp and no-feedback. You have the same amp, everything the same, and then you switch feedback off.
Then listen ...
... and tell me about it.
Charles 🙂
Hi Charles,
would you like to answer some questions from my post #356? and perhaps perform a test as proposed?
Do you have any own amplifier prototype with any measured data?
BTW, all of you intersted in the topic discussed right know,
here's an 2½ year old thread with interesting comments concerning the SPLUF topology!
Cheers! 😎
would you like to answer some questions from my post #356? and perhaps perform a test as proposed?
Do you have any own amplifier prototype with any measured data?
BTW, all of you intersted in the topic discussed right know,
here's an 2½ year old thread with interesting comments concerning the SPLUF topology!
Cheers! 😎
Ultima Thule said:
Hi Ultima,,
you are completely right in your remarks that voltage and current are different at the feedback- and speaker- nodes.
With a bipolar output stage (not running in class A) I would expect audible crossover distortion. But a bipolar output stage would also introduce feedback from the speaker through VAS talkback via base-current, and therefore would not be the premier choise for making a SPLIF amplifier.
This will however not happen to this extent with a properly biased Mosfet output pair.
When switching feedback off on my personal Splif amp, I would notice any crossover distortion if it would be prominent to any significant amount, but this is not the case. Also since Audio-Technik has made quite a lot of SPLIF modifications to consumer hifi amplifiers they would also have informed me, if there would be any problem, but this does not seem to be the case at all.
If this had been the case, I would have made some measurements on my own and maybe discarded the idea of making the SPLIF concept public.
But hey, it sounds great, so why should I bother ?
Charles 🙂
Charles,
sorry to say this but you really slipped away... didn't you?
Crossover distortion was mentioned in my post #350, not in post #356 to which I reffered in my previous post, so let us forget about the crossover thing at least for the moment!
Why repeating yourself, you have allready talked about the "positive points" with FET's.
You may have whatever subjective opinion about what is sounding good, I did not ask your subjective opinion yet!
So, back on a technically discussion basis, for your information I was asking for information about the damping factor and whether you would perform a X/Y-ing on oscilloscope, see #356 again. 🙂
Would you provide it to make your self belived as an engineer?
Another question, what kind of disturbances exactely do you belive is picked up of loudspeaker cables that can make the FB circuitry behave "ill" in a normal global FB amplifier?
BTW, you are talking about speakers as something "bad" with it's "unlinear" nature, have you ever seen a linear B/H curve?
I am on a learning curve,
Cheers! 😉
PS: I should mention that I still think the SPLIF is absolutely an interesting invention, such it's "unique"! 😎
sorry to say this but you really slipped away... didn't you?
Crossover distortion was mentioned in my post #350, not in post #356 to which I reffered in my previous post, so let us forget about the crossover thing at least for the moment!
Why repeating yourself, you have allready talked about the "positive points" with FET's.
You may have whatever subjective opinion about what is sounding good, I did not ask your subjective opinion yet!
So, back on a technically discussion basis, for your information I was asking for information about the damping factor and whether you would perform a X/Y-ing on oscilloscope, see #356 again. 🙂
Would you provide it to make your self belived as an engineer?
Another question, what kind of disturbances exactely do you belive is picked up of loudspeaker cables that can make the FB circuitry behave "ill" in a normal global FB amplifier?
BTW, you are talking about speakers as something "bad" with it's "unlinear" nature, have you ever seen a linear B/H curve?
I am on a learning curve,
Cheers! 😉
PS: I should mention that I still think the SPLIF is absolutely an interesting invention, such it's "unique"! 😎
janneman said:
not exactly jan
did you read my last post??
speaker is a dynamic microphone so every mechanical motion (own mechanical resonances included) are transduced to voltage
so it isnt just a 'load'
- Status
- Not open for further replies.