Negative feedback - right or wrong to use?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
As i started in title things are interesting to think about....at least for me.If we have some amp (no matter which topology is used) and a speaker box with three speakers in it (bass,middle and tweeter) situation is far away from simple.At first place lets assume that crossover design is done as standard quality one.We will have different impedances for some frequency ranges.If we have constant gain and perfect frequency responce (flat line) what we will hear will NOT be flat frequency responce on speakers.....On some freq. ranges speaker box will pull more current (constant output voltage from amp on all frequencies) and on some it will pull less no matter of how flat output from amp is.So,some frequencies will be louder and some quieter.Why we need then negative loop?On other side if we dont have any feedback from output situation will be same but almost worse becouse output voltage will fluctuate too.This will happen with single speaker on output without any crossover design too.So,my main question is : Is there some other concept around all this that can match output with load better than all its done till now?Any ideas?
 
Administrator
Joined 2007
Paid Member
Commercial speakers are designed on the following assumptions:

1/ The speaker design itself is designed to be driven from a constant voltage source... which neatly leads on to:

2/ That the amplifier driving the speaker will maintain its constant voltage output over all conceivable load impedances.

This means that even though the speaker impedance may vary by a factor of 20 or more over the audio band, the amplifier is not bothered by this and always delivers the same voltage at its output.
 
Commercial speakers are designed on the following assumptions:

1/ The speaker design itself is designed to be driven from a constant voltage source... which neatly leads on to:

2/ That the amplifier driving the speaker will maintain its constant voltage output over all conceivable load impedances.

This means that even though the speaker impedance may vary by a factor of 20 or more over the audio band, the amplifier is not bothered by this and always delivers the same voltage at its output.
Assumptions are fine in theory but not in practice in the audio realm IME.
20:1 impedance variation is too much for most amplifiers, and the result is a smattering of distortion products. `
Zobels across individual drivers plus across overall crossover network markedly reduces subjective distortions.
Cost is appreciative but the benefits are priceless.

Dan.
 
AX tech editor
Joined 2002
Paid Member
As i started in title things are interesting to think about....at least for me.If we have some amp (no matter which topology is used) and a speaker box with three speakers in it (bass,middle and tweeter) situation is far away from simple.At first place lets assume that crossover design is done as standard quality one.We will have different impedances for some frequency ranges.If we have constant gain and perfect frequency responce (flat line) what we will hear will NOT be flat frequency responce on speakers.....On some freq. ranges speaker box will pull more current (constant output voltage from amp on all frequencies) and on some it will pull less no matter of how flat output from amp is.So,some frequencies will be louder and some quieter.Why we need then negative loop?On other side if we dont have any feedback from output situation will be same but almost worse becouse output voltage will fluctuate too.This will happen with single speaker on output without any crossover design too.So,my main question is : Is there some other concept around all this that can match output with load better than all its done till now?Any ideas?

There is no definite link between how much current speaker pulls and how loud it is.

For instance, at resonance a driver is high impedance thus pulls less current but is as loud as at other points due to resonance.

Also, xover may have dips and peaks of impedance which may or may not be louder or less loud. No direct link.

Jan
 
So there ARE variations......Constant voltage at all frequencies from amp means SAME output level on all frequencies.And that my friends will NEVER happen.You WILL HAVE constant voltage and absolutely wrong output levels on speaker.If we assume no feedback at all then ABSOLUTELY speaker design matters and ABSOLUTELY anyone will make difference between any two speaker boxes.This puts all speakers and companies at same testing level and this is real comparison of the speaker boxes qualities......further more with no feedback and "IDEAL" speaker box design where impedance is always same on any frequency (not done yet) result is ideal.So MATCHING circuitry between load impedance and output amplifier impedance is whats needed.....or nothing if speaker box is properly designed.Most of companies makes amplifiers that measures "PERFECT" and sound is pure crap....On other side there are designs that measures very bad but sounds amazing.I tested few of that kind and developed my own....So any ideas of possible matching real time network :) ?

Cheers,
 
AX tech editor
Joined 2002
Paid Member
So there ARE variations......Constant voltage at all frequencies from amp means SAME output level on all frequencies.And that my friends will NEVER happen.You WILL HAVE constant voltage and absolutely wrong output levels on speaker.If we assume no feedback at all then ABSOLUTELY speaker design matters and ABSOLUTELY anyone will make difference between any two speaker boxes.This puts all speakers and companies at same testing level and this is real comparison of the speaker boxes qualities......further more with no feedback and "IDEAL" speaker box design where impedance is always same on any frequency (not done yet) result is ideal.So MATCHING circuitry between load impedance and output amplifier impedance is whats needed.....or nothing if speaker box is properly designed.Most of companies makes amplifiers that measures "PERFECT" and sound is pure crap....On other side there are designs that measures very bad but sounds amazing.I tested few of that kind and developed my own....So any ideas of possible matching real time network :) ?

Cheers,

Well, with so much ABSOLUTES, not much left for an intelligent discussion.

Jan
 
Things are simple.Having feedback makes amp to be power source which doesnt care about load but about it self only.If i measure 7498 WITHOUT any load on 80% of max output voltage at certain range i will have huge distorsions....if i have load it will be same thing but about 1-10% in comparison with previous situation.It has feedback as many amps do_On other side i made my first design without feedback and i noticed same thing .... from 3khz to 7khz and latter from 17khz to 25khz i had effect on output like resonance and voltage boost and huge distorted sine wave.So i worked last few years on that and i made feedback less design which doesnt suffer any more from this issues.....it acts like AB class amplifier ....with or without load i have same output amplitude on any output voltage level including max one.And i have linear response on output without feedback on pure resistive load (zero inductance - hand made load)....and i DONT have linear response on different speaker boxes.BUT i have on some :) which was very interesting for me.Thats why i decided to discuss about feedback.Feedback less sound is uncomparable with feedbacked sound....Feedback should be discussed as term here.What feedback should be and what it should represent and do to make amp sound better ??? or to make constant power output on any frequency ???? or both ??? or something else.With simple graphic equalizer i am able to tune amp to have linear output on any speaker box (not idealy flat but 98%) ....so do we need feedback to kill all good and have "great" amplifier or we need feedback but at first ....something else that we will call feedback and not whats present in books?I expected to hear some new idea for feedback that we all can benefit from.I dont have new idea about it but i decided that its time to start working on it which i will do no matter how long it will take me to reach the goal.
 
AX tech editor
Joined 2002
Paid Member
The issue you are experiencing is due to variations in the amplifier output impedance and is a well-kown and described phenomenon. The output impedance largely determines the interaction of the amplifier with the speaker.

As Marcel mentioned, feedback can be used to manipulate that output impedance and thus the interaction of the amp with the speaker.
But it is not the feedback directly that determines the interaction.

You can change the amplifier output impedance directly with external resistors and a low-output impedance amp (except negative Zout) , and that will allow you to investigate, without having to worry about feedback.

Jan
 
I made D class amplifier which has very small output impedance already.I can manipulate it through voltage,curent or mixed type feedback ...i know that.Overal response is almost linear even amp is feedback less....i am talking about response on speakers.I have small differencies on different speakers.All of them i can correct with graphic equalizer.So main question is do we need feedback at all in this case....and also question is what concept of feedback is ideal for this "issues" ?
 
Account Closed
Joined 2001
I'm not an expert, but I would suspect yes. No doubt there is some servo-control of various aspects of the cutting process. And there is certainly feedback used somewhere along the way before the music arrives at the vinyl recording stage.

I suggest to secure all devices in your everyday life that use feedback and see how that goes for you. :)

Dave.
 

ICG

Disabled Account
Joined 2007
Assumptions are fine in theory but not in practice in the audio realm IME.
20:1 impedance variation is too much for most amplifiers, and the result is a smattering of distortion products.

Er, no. Most amplifiers with clean, working design work fine even without any load without clipping, wave form changes/deformations or distortion. That means high impedance peaks are not a problem (well, except for impedance peaks of the tweeter which leads to frequency response peaks too at class D amps). Low impedances on the other hand are because the high current lets the voltage drop (limitations of the power supply i.e.) which also might cause clipping etc.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.