Unfortenatly the feedback is not so simple and good.janneman said:
It corrects not just the amplifiers distortion but try correcting
everything, cause it just looks the result at the output.
For example when the PS is weak, it cant be provide enough
power, and the output level Uout will not be Uin * Au,
but the feedback will try to correct this too.
When our load, the speaker talks back, the amp senses this,
and will try to correct this too.
But when we try to correct complex processes, we can easily do wrong, rather than good.
Cortez said:
I don't get you. Of course it corrects whatever deviations of Uout are there. It doesn't matter if they are caused by transistor beta droop, low supply, speaker back-EMF, high temperature, whatever. That is the PURPOSE of feedback. I have no idea why that would be bad?
Jan Didden
I dont know the right words in english, but there is a separate
profession what deal with process controlling and regulation.
We have learnd about this with simulations.
(I think MicroCap was the name of this program)
A simple example: our system has
- a motor
- an axle with a given torsion
- and a mass on this axle (this is the inertia of the load)
Our aim is to set (rotate) this mass to a fixed position (eg to 180°)
with the motor, and certainly with feedback controll, cause we have
a potentiometer on the axle.
Its a very serious and separate problem to configure this system.
The items in the feedback chain can be:
- P (Proportional)
- I (Integrator)
- D (Derivator)
This process runs in time so its 2 dimensional in this view.
When we try to achieve our aim, we will see, that its not easy,
and in general the system will have a resonance with a given
max amplitude and with a given falloff time, cause the motor
will stop the first time when the potentiometer says that 180°
is reached. But the axle has an inertia, and therefore it will not
stop at 180°, but rotate further. Then the feedback senses, that
the axle is not on 180° and starts to rotate it back, but every time
the inertia (and of course the torsion too) will confuse the system,
cause its not easy to controll the behaviour of such a complex chain.
I hope that shows the analogy to amplifiers, cause they are also
similar controll systems with the speaker. But also between two stages.
A P item is an ideal amplifier or attenuator, an I item is a capacitance,
and a D is an inductance in electronics.
This is the world of resonances, oscillation circuits, phase-distortions, etc...
Cortez said:
you just described a typical automation system where system stability is required.
What you didn't realize is that a well designed feedback system will "settle" down over a period of time. It is the designer's goal (one of them) to make sure that the system settles down in a given / short enough period of time.
There is no feedback system that will achieve instant settle-down. In an amplifier, the hope is that through minimization of non-resistive components (and I and D in your example), an amp with feedback will settle down fast enough for the "settle down" not to be audioble - ringing for example is a form of "settling down". and in most amps, it is dealt with effectively.
What you described is a theoretical possibility that really has no practical bearings in modern amplifiers.
Absolute phase, DC coupled amplifier wich can drive complex load, do you know one ?
Or an electronical stage does exactly what we expect from it ? No it does not!
If we have a resonance at the output, even if it will settle down, creates a "transient noise".
Therefore is the sine wave test too simple, cause these side effects settle down in a short time,
but in the music there is only transient signal, there is permanent changing and when the previous state
does not have falled off, it will modulate the current information.
But without feedback there are also important (phase and complex load) problems in an amp.
Otherwise maybe there is an effect like the jitter in the digital world cause problems to us.
Lets call it analog jitter, what means all kinds of time and phase problems across our chain.
Specially when a link shifts not with a fix value. Our system is a lot of spring and mass, and
we must to provide a faultless signal at the output from it...
I know but in my view its not enough, cause this falloff will contaminate our original signal.
Yes, partially tiny, dynamic, parasitic capacitances, and inductances are our enemies...
Well, i'am not sure. Unfortenatly when a trouble isnt direct, then its indeirect...
Why ? In your view, the speaker's cone move exactly like the signal at the amp's input ?
Or an electronical stage does exactly what we expect from it ? No it does not!
If we have a resonance at the output, even if it will settle down, creates a "transient noise".
Therefore is the sine wave test too simple, cause these side effects settle down in a short time,
but in the music there is only transient signal, there is permanent changing and when the previous state
does not have falled off, it will modulate the current information.
But without feedback there are also important (phase and complex load) problems in an amp.
Otherwise maybe there is an effect like the jitter in the digital world cause problems to us.
Lets call it analog jitter, what means all kinds of time and phase problems across our chain.
Specially when a link shifts not with a fix value. Our system is a lot of spring and mass, and
we must to provide a faultless signal at the output from it...
Re: Absolute phase, DC coupled amplifier wich can drive complex load, do you know one ?
I'm sorry, I am always willing to engage in a discussion, but you throw so many different things in one basket, I don't know where to begin. Most of it is speculation, and that can go on and on and on.
Maybe we should just take ONE issue like settling time in amplifiers. That is a real issue, but as said by 999etc this is no longer a practical consideration in competently designed present day amplifiers. If you test them with fast square waves or pulses, you may see ringing, but that's not a realistic music signal. It is easy to calculate the max rise time of a max output level 20kHz signal and design the amp with a comfortable safety factor for that. Problem solved.
Jan Didden
Cortez said:
I'm sorry, I am always willing to engage in a discussion, but you throw so many different things in one basket, I don't know where to begin. Most of it is speculation, and that can go on and on and on.
Maybe we should just take ONE issue like settling time in amplifiers. That is a real issue, but as said by 999etc this is no longer a practical consideration in competently designed present day amplifiers. If you test them with fast square waves or pulses, you may see ringing, but that's not a realistic music signal. It is easy to calculate the max rise time of a max output level 20kHz signal and design the amp with a comfortable safety factor for that. Problem solved.
Jan Didden
Sorry, my intention was rather thought-provoking with that.
Sorry, but i doesnt agree. Present day amplifier doesnt mean perfect.
We can certainly stop with developing our amps, but i dont want this.
You are right, roughly our amps works, but now its time to concentrate
to phenomenons in the time dimension, whereof i writed above.
The fact, it is heavier to show these problems on music signal,
doesnt mean that they doesnt exists in our system. Therefore
can be helpful to measure our amps with heavy transient signals,
cause they can show the real inner time and phase problems.
When it doesnt disturb you, that the majority of the amps isnt as clear as they could be...
Re: Absolute phase, DC coupled amplifier wich can drive complex load, do you know one ?
you are absolute right on both fronts. However, is there anything in the world that is exactly what we expect it to be? Can you ears tell the differrence between a perfect signal and a less than perfect but still good enough signal?
If your ears cannot, isn't it foolish to improve the reproduction beyond the point where you can tell the difference?
I think most rationale people would say that is a waste of our precious resources.
Again, you have taken a theoretical approach to a very practical problem.
Cortez said:
you are absolute right on both fronts. However, is there anything in the world that is exactly what we expect it to be? Can you ears tell the differrence between a perfect signal and a less than perfect but still good enough signal?
If your ears cannot, isn't it foolish to improve the reproduction beyond the point where you can tell the difference?
I think most rationale people would say that is a waste of our precious resources.
Again, you have taken a theoretical approach to a very practical problem.
Cortez said:
Well, I said that I agree that if you test with fast square waves and/or pulses you may see ringing or slew limiting. I know the existence of these problems. They have nothing to do with audio, it is straight forward electrical engineering. My point is that it is easy to design an amp that with music signals never gets into these problems. So why continue to stress that the problem *may* exist, if you use unrealistical, never appearing in practise, test signals. Do you suggest we design amps for a freq range from DC to light, for infinite to zero load impedances and infinite input levels just because otherwise they have an "internal problem"?
Jan Didden
No, but that is enough reason to stop exploring new directions ?
My ears can tell, that there is a lot of thing ahead need to develop.
I dont think that the present day amps have reached the quality-level,
that fully indulge the sensitivity and resolution our ears.
This is a DIY audio forum, this is the goal, isnt it ? :-]
Its not a constraint, but when there is a little hope to find something new
to a better sound quality, why wouldnt we go that way ?
Yes, because untill there is no solution, we must to presume theories.
Why should we spare on this ? Why we should have a worse solution, when we can achieve a better one ?
A better car is also better in a lower loading (ok, maybe not in parking
not only on havy stress.No, not ad infinitum, but as far as we can lets make it stable,
even on havy load and on fast transients, etc...
Dont interpret my posts as attacks, rather ideas, maybe new views on old things.
To view this hole thing with an other glasses. Specially the smaller phenomenons, the details.
The point is, that we should map the smaller effects too, like resonances and similar phenomenons
like the analog jitter. Lets measure even more details, the smaller ones too!
A few example:
- amp turn on/off (maybe a better transient test, than a continuous square wave...)
- little knocks with your finger to te cables, parts, pcb-s etc (meanwhile viewing the scope)
- moving, rotating the pseudo-load (lot of resistors, how do you say this ?)
- square wave with capacitive load, watch the input (!) with the scope
...
More stability, less unwanted parasitic resonances.
The amplification devices must be in our focus (no matter that its a tube, a FET, BJT...)
And obviously our tool is the circuit around it.
This whole process can be viewed also as an energy-flow.
Its operation is correct, when no other energy-chaning happens just the amplification.
Maybe the thermal-distortion can be a good thought-provoking thing.
The energy balance will be there injured, cause it heats up our amplifier device,
and the energy-stealing/storing will modify the devices operation.
And have we thinked to this earlier ? Not exactly. I know thats not a big thing,
but now small phenomenons are in my view. This is a very expressive example i think.
But thats just one thing, in my view there are lot of similar areas to explore...
Greets !
To all!
It's funny how these kind of discussions almost always boils down to one of the hottest subjects: "what happens in the feedback node/feedback sees this n' that.../with the feedback it will do this 'n that..."
The feedback is kind of an evil wheel, when this kind of discussion were up earlier this year(I think) in another thread I posted a link to Leach who talks in one of hes paper about the signal from output coming "too late" to the feedback node.
Cheers Michael
It's funny how these kind of discussions almost always boils down to one of the hottest subjects: "what happens in the feedback node/feedback sees this n' that.../with the feedback it will do this 'n that..."
The feedback is kind of an evil wheel, when this kind of discussion were up earlier this year(I think) in another thread I posted a link to Leach who talks in one of hes paper about the signal from output coming "too late" to the feedback node.
Cheers Michael
I also used to think that the problem with Feedback amp is the feedback signal comes "too late" compared to input signal.
"too late" here is actually "group/phase delay" occurs due to Capacitance and AC electricity.
But the truth is that the problem with global feedback is happening because of the contrary cause.
Every transistor has a capacitance inside, no matter how small (pf) it is. Where there is Capacitance and AC (audio is AC, right?), there will be group/phase delay. The more stages in 1 audio amp, there will be more delay. The VAS transistor is submitting its delay, the Pre-Driver transistor is also submitting its delay, the Driver transistor is also submitting its delay. Actually it happens in differential pair transistors too (but lets say it is 0 delay here for now).
If you make an open loop amp, say, at 10khz, the output will be "phase delayed" by 30deg. This means if we use dual trace scope, the output will be delayed by 30deg from the input trace.
When you CLOSE the loop, the differential pair will not allow ANY delay, because if you see the differential pair, the left base is fed by input signal, its emitors is tied with right's emitors, right's base is connected to output stage (by voltage divider). This structure doesn't allow any phase delay, otherwise the phase of signal (in 1 transistor/differential half) at the base will be not-in-phase with the signal at its Emitor/Collector (which is impossible).
So, what happens to the 30deg that "should be happening" in the rest of the amp cct? (but disabled by the differential pair when the loop is closed by global feedback)
This delay is a natural thing, should be happening naturally. But it is not happening because it is not allowed by the differential pair.
What happened if "something natural" is forced to "not happening"?
Is it transforming its "finite/some delay+low order/no harmonics" into "no-delay but+high-order-harmonics"?
"too late" here is actually "group/phase delay" occurs due to Capacitance and AC electricity.
But the truth is that the problem with global feedback is happening because of the contrary cause.
Every transistor has a capacitance inside, no matter how small (pf) it is. Where there is Capacitance and AC (audio is AC, right?), there will be group/phase delay. The more stages in 1 audio amp, there will be more delay. The VAS transistor is submitting its delay, the Pre-Driver transistor is also submitting its delay, the Driver transistor is also submitting its delay. Actually it happens in differential pair transistors too (but lets say it is 0 delay here for now).
If you make an open loop amp, say, at 10khz, the output will be "phase delayed" by 30deg. This means if we use dual trace scope, the output will be delayed by 30deg from the input trace.
When you CLOSE the loop, the differential pair will not allow ANY delay, because if you see the differential pair, the left base is fed by input signal, its emitors is tied with right's emitors, right's base is connected to output stage (by voltage divider). This structure doesn't allow any phase delay, otherwise the phase of signal (in 1 transistor/differential half) at the base will be not-in-phase with the signal at its Emitor/Collector (which is impossible).
So, what happens to the 30deg that "should be happening" in the rest of the amp cct? (but disabled by the differential pair when the loop is closed by global feedback)
This delay is a natural thing, should be happening naturally. But it is not happening because it is not allowed by the differential pair.
What happened if "something natural" is forced to "not happening"?
Is it transforming its "finite/some delay+low order/no harmonics" into "no-delay but+high-order-harmonics"?
lumanauw said:
don't we have to establish that having 30 degree phase shift is "something natural" first before we can answer the above question?
I mean, is it established anywhere that our ears are used to listen with a certain amount of phase shift / delays?
If the answer is "no" or "we don't know", maybe your question above is moot.
if you were to step back a little and recognize that nothing in this world is entirely evil or entirely perfect, you would have very quickly realized that criticizing feedback blindly, as some of us do, doesn't make sense.
Because no one has established so far that having no feedback is a better alternative.
Especially when one is dealing with mass production of open loop electronics,
Because no one has established so far that having no feedback is a better alternative.
Especially when one is dealing with mass production of open loop electronics,
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.