Hi Lumunauw,
"This is the question. Is the output node of the power amplifier forming another input for the amp system or not. In feedback sytem, YES, because the signal at the output node will be feeded back to the whole system by feedback system.
So, back EMF is another "distoriton source" for feedback power amplifier, even it comes at the output node. Back EMF is giving "not-music-related" energy at the output node, that the amp has to synchronize with signals at the inputs and amps output. (whew, how difficult it is tobe the amplifier )"
Feedback theory dictates that any intrusion is reduced by the amount of OL gain before the intrusion point, being an input referred qty (e.g. PSRR,...) AND then amplified to the output by the CL gain. So any intrusion at the output will be reduced by the full OL gain of the amplifier and reappear x CL gain.
Accordingly, increasing the OL gain will reduce such intrusion all things being equal. Feedback (theory) generally works as advertised!
Cheers,
Greg
"This is the question. Is the output node of the power amplifier forming another input for the amp system or not. In feedback sytem, YES, because the signal at the output node will be feeded back to the whole system by feedback system.
So, back EMF is another "distoriton source" for feedback power amplifier, even it comes at the output node. Back EMF is giving "not-music-related" energy at the output node, that the amp has to synchronize with signals at the inputs and amps output. (whew, how difficult it is tobe the amplifier )"
Feedback theory dictates that any intrusion is reduced by the amount of OL gain before the intrusion point, being an input referred qty (e.g. PSRR,...) AND then amplified to the output by the CL gain. So any intrusion at the output will be reduced by the full OL gain of the amplifier and reappear x CL gain.
Accordingly, increasing the OL gain will reduce such intrusion all things being equal. Feedback (theory) generally works as advertised!
Cheers,
Greg
amplifierguru said:[snip]So, back EMF is another "distoriton source" for feedback power amplifier, even it comes at the output node. Back EMF is giving "not-music-related" energy at the output node, that the amp has to synchronize with signals at the inputs and amps output. (whew, how difficult it is tobe the amplifier )"[snip]Greg
Greg,
To be clear: I am a feedback advocate, and I agree with your post above. But if we consider the amp output as an input (which is valid), what is then the output node for that input signal? If it is the same node, (IOW, input node = output node) then no matter what, input will always be = to output. Is this a problem for the amp?
What is the objective, is it to make this signal as small as possible? That can be done with the lowest possible Zout. With feedback or topology inherent. If we do it with feedback, there may be substantial signals within the loop, but the speaker is connected to the output, not within the loop. The only requirement then is to make sure that the amp can handle those signals internally. And it is my view that if an amp can handle normal full-level input signals in the whole audio band, it will only laugh at those puny EMF signals that the limited-range vibrating speaker cone throws at it. EMF in a competently designed amp is a non-issue.
Jan Didden
There is not only back-EMF, speaker cables are also aerials.
And _all_ feedback designs are low feedback designs at "high" frequencies.....
I like Mauro's work very much. Instead of saying "cannot be" he investigates and tries to arrive at a correlation between what he hears and what he measures.
Buon lavoro, complimenti Mauro.
Ciao, Tino
And _all_ feedback designs are low feedback designs at "high" frequencies.....
I like Mauro's work very much. Instead of saying "cannot be" he investigates and tries to arrive at a correlation between what he hears and what he measures.
Buon lavoro, complimenti Mauro.
Ciao, Tino
Hi amplifierguru, this is perfect only in the teoria. If the load that produces the error is reactive, the (voltage) NFB net works only on the part of the error in phase with if same ( and with the input signals). The level of the error depends from the phase-shift and from the value of Zout inside OL.
Hi Jan, I don't agree with some yours expositions:
Perhaps yes perhaps no. For my the opportunity to think all the non ideals behaviours deserve of be studied. From electronic developer, I am of accord in the act that be necessary gather only on the problems it embodies, but in audio it doesn't exist a line drives things cut off by establish on does gather. Among the "historians" audio planners am more frequent the opposite theories that convergent...
The presence of a (internal) Zout inserts (on principle) a constant of error in the loop NFB contingent upon the load current. Therefore part of the (load) current signal influences the NFB loop...
A part of me has it was "base" convictions of a lot of members. My curiosity me handed to appraise all the point of view on this matter. Jan asks: "-Which is the objective...-" I say to Jan: To have an objective affair leaves from a problem to resolve. I have not still "recluse" the problem "truthes" of the amplifiers, and am looking for to do it ( with my errors ). For me the fact that Hiraga has taken in the Back_EMF consideration as an element to analyse ( with her setup ) is enough to verify his thesis. and for you?
Ciao
Mauro
mauropenasa said:
Mauro:
While I hadn't still time to devote to the simulation promised some posts back, I feel necessary to make some precissions.
If we are dealing with a voltage amplifier - that is it tries to generate a scaled copy of the imput voltage - and it is well designed, NFB will do whatever is necessary to achieve this end.
If the load presents a back EMF, no matter whether it is in phase or not with input, then it will try to correct with an opposing output to compensate and ensure the load voltage is still a copy of the input voltage. **No matter the phase, not even the waveform for that matter, the offending EMF is presenting**.
It is simply treated as an error and compensated for.
For this to be true, it must comply with 2 basic requisites (besides high, flat OL gain).
1. The whole loop must be fast. This implies the closed loop phase shift must be negligible in the useful frequency range. (Many misbehaviors attributed to global NFB have much to do with overlooking this condition).
2. The output stage must be able to obey with whatever amount and phase shift current is demanded to comply with the desired voltage waveform, for the presented load. (Again, many misbehaviors have to do with good amplifiers for resistive loads, that cannot cope with terrible loads at certain frequencies).
I have a feeling after poring over hundreds of posts in this forum on the about 9 months I've been here, a sizable amount of controversy with respect to global NFB results from the murky situation that results when the previous conditions are not clearly defined or understood.
Rodolfo
Hi Jan,
So I wonder what output *impedance* our competently designed amplifiers have ?
The more the OL gain the lower the output *resistance*,
R = 1 / gm;
but the higher the open loop gain the greater the stability problems, and thus the lower the OL -3dB roll-off frequency must become to ensure stability.
SY and Greg concur that phase shift cannot be discounted.
The amplifier effectively becomes inductive, with
L = 1 / ( 2 x 22/7 x gm x f ) 'f' being the -3dB OL roll-off frequency.
The higher the OL gain the lower the inductance; but this is simultaneously countered by the lower stabilised OL roll-off frequency.
That inductance develops a parasitic output terminal voltage wrt differential error control and this is related to loudspeaker current flow - not input, though it is the combination of input and back-EMF induced error that result in the generation of a non-musical reproduction 'sound' characteristic.
So simply increasing the open loop gain to decrease output resistance does not mean that the amplifier will *output* more accurately when driving a reactive loudspeaker, hence the sonic characteristics often ascribed.
My experience is that low frequency back-EMFs can occasionally be considerable, and not only draw much greater internal currents than expected due to maximium amplitude acting upon a basic load, but cause potential scenarios for internal cross-conduction due to output device storage at bias crossovers in class-AB designs where the zero current crossovers become phase shifted away from zero alternating voltage crossover.
Cheers ......... Graham.
So I wonder what output *impedance* our competently designed amplifiers have ?
The more the OL gain the lower the output *resistance*,
R = 1 / gm;
but the higher the open loop gain the greater the stability problems, and thus the lower the OL -3dB roll-off frequency must become to ensure stability.
SY and Greg concur that phase shift cannot be discounted.
The amplifier effectively becomes inductive, with
L = 1 / ( 2 x 22/7 x gm x f ) 'f' being the -3dB OL roll-off frequency.
The higher the OL gain the lower the inductance; but this is simultaneously countered by the lower stabilised OL roll-off frequency.
That inductance develops a parasitic output terminal voltage wrt differential error control and this is related to loudspeaker current flow - not input, though it is the combination of input and back-EMF induced error that result in the generation of a non-musical reproduction 'sound' characteristic.
So simply increasing the open loop gain to decrease output resistance does not mean that the amplifier will *output* more accurately when driving a reactive loudspeaker, hence the sonic characteristics often ascribed.
My experience is that low frequency back-EMFs can occasionally be considerable, and not only draw much greater internal currents than expected due to maximium amplitude acting upon a basic load, but cause potential scenarios for internal cross-conduction due to output device storage at bias crossovers in class-AB designs where the zero current crossovers become phase shifted away from zero alternating voltage crossover.
Cheers ......... Graham.
Guys,
Let's cut the **** and put figures on this so called problem to define the exact miniscule nature of it. And let's include filtering out of any say 100KHz + perturbations/instant start simulation effects and deal with REAL problems not imagined ones. Like the old TIM when you took out the out of band spikes that just aren't present in nature - TIM just faded into insignificance!
Hi Graham,
Sure every feedback amplifier has an inductive component to it's Zout - we deal with that. It's not the end of the world as we know it. And it's of little to no detriment in a well designed system.
Cheers,
Greg
Let's cut the **** and put figures on this so called problem to define the exact miniscule nature of it. And let's include filtering out of any say 100KHz + perturbations/instant start simulation effects and deal with REAL problems not imagined ones. Like the old TIM when you took out the out of band spikes that just aren't present in nature - TIM just faded into insignificance!
Hi Graham,
Sure every feedback amplifier has an inductive component to it's Zout - we deal with that. It's not the end of the world as we know it. And it's of little to no detriment in a well designed system.
Cheers,
Greg
The Story so far!
The Back EMF is natural phenomena and encountered by every amplifier which is connected to reactive load or real world speaker......It increases with increase in voice coil inductance , especially in case of low frequency transducer such as a subwoofer.... ...
To every action there is equal and opposite reaction---Newton Says
low output impedance Playing the Vital Role
The low impedance output of amplifier acts as a short circuit to the back EMF generated , hence the speaker ringing is eliminated and much of the effect of BACK EMF is reduced hence what you get is tight speaker cone movement...The Inductive output of every amp is related to the realworld parameters which co-exist with inductive and capacitive nature of components and thus the composite actually refers to all the reactiveity to deal with.....
regads,
Kanwar
The Back EMF is natural phenomena and encountered by every amplifier which is connected to reactive load or real world speaker......It increases with increase in voice coil inductance , especially in case of low frequency transducer such as a subwoofer.... ...
To every action there is equal and opposite reaction---Newton Says
low output impedance Playing the Vital Role
The low impedance output of amplifier acts as a short circuit to the back EMF generated , hence the speaker ringing is eliminated and much of the effect of BACK EMF is reduced hence what you get is tight speaker cone movement...The Inductive output of every amp is related to the realworld parameters which co-exist with inductive and capacitive nature of components and thus the composite actually refers to all the reactiveity to deal with.....
regads,
Kanwar
Hm.. since when is ringing in an LC tank circuit is eliminated by RAISING the Q?
And I am NOT playing with words, I am serious. And what tight cone movement is guaranteed by lowering the amp out Z to .0001 ohm, when the cable/ crossover/ speaker sum DC R is ~ 8 ohm? RAISING the output Z to 8 ohm would mean a heavy 6dB loss in effectivity of the damping! [that is, the half]
Ciao, George
Hi Ingrast, I accept always gladly the precise statements. I have un verifications from behaviour, me reserve to motivate my affirmations in way more exact ( if succeed ). 
Hi amplifierguru, am a lot of happy if draw up a list me of " real-problem " to resolve to build an "ideal" amp. You save up me a lot of work!
I don't say it to take you in turn, am serious. I accept all the "demonstrable" theories that improve the "musical" performance ( that electric not are of interest me much ).
Hi George, in 2 words you has dismounted the theory of the " cone damping ". If I had your ability of synthesis...
Ciao
Mauro
Hi amplifierguru, am a lot of happy if draw up a list me of " real-problem " to resolve to build an "ideal" amp. You save up me a lot of work!
I don't say it to take you in turn, am serious. I accept all the "demonstrable" theories that improve the "musical" performance ( that electric not are of interest me much ).
Hi George, in 2 words you has dismounted the theory of the " cone damping ". If I had your ability of synthesis...
Ciao
Mauro
Graham, you've confused me. On the one hand, you talk about the rising output impedance with frequency as being inductive (which is a valid model). OTOH, without taking a breath, you talk about back EMF which is mostly a low frequency issue, where the inductive nature of output Z is insignificant. I'm having trouble connecting the dots here.
Jorge said:
Hi Jorge
well, more precisely it is electro-mechano-acoustical transducer and concerning relatively low low velocity of sound ('bout 340 m/s), than equivelent electrical circuit must involve transmission lines (lossy) with huuuuuuge propagation delay. I guess I don't have to add this is tricky load to drive in terms of stability. And this can produce delayed back EMF, which may be also high freq., not only close to woofer's resonance freq.
best regards
Relevance of acoustic part of EMF
While working my way currently in the T-S simulation, and listening to some concepts rised once and again, I feel compelled to remark something.
The purely acoustic component of back EMF - i.e. that resulting from electrical to acoustic transduction and from ambient pickup - are small in comparison with other EMF sources (in fact frequently neglected in modeling).
Microphones are notoriously known for being feeble generators. Speakers working backwards are better due to comparatively larger area, but no match to actual amplifier output levels.
By far the most important speaker (mechanical) EMF sources are the driver's own electromechanical system (moving mass, suspension) and to a lesser degree the enclosure loading.
As said, all this is accurately reflected in the electrical equivalent model provided by the T-S analysis.
Rodolfo
While working my way currently in the T-S simulation, and listening to some concepts rised once and again, I feel compelled to remark something.
The purely acoustic component of back EMF - i.e. that resulting from electrical to acoustic transduction and from ambient pickup - are small in comparison with other EMF sources (in fact frequently neglected in modeling).
Microphones are notoriously known for being feeble generators. Speakers working backwards are better due to comparatively larger area, but no match to actual amplifier output levels.
By far the most important speaker (mechanical) EMF sources are the driver's own electromechanical system (moving mass, suspension) and to a lesser degree the enclosure loading.
As said, all this is accurately reflected in the electrical equivalent model provided by the T-S analysis.
Rodolfo
Hi Greg,
Yes every amplifier has separately acting internal output impedance, but, not all amplifiers have impedance that causes a damping phase change at low/mid audio frequencies such that all damping above that frequency is in quadrature, so I do not know what you say is 'dealt with'.
Okay so all amplifiers have output impedance ... and what happens at the output of an inductive feedback loop ? ... there is a leading error voltage before current correction is capable of acting ... the error increasing with frequency or transients.
Thus it is possible for back-EMF to fractionally reverse bias and cause phase shifted output stage crossover distortion before the NFB loop can minimise the sensed error.
Even Rod Elliot shows something similar to this on his Distortion Analyser postings, and I believe that Mauro's spectral investigation could highlight these effects.
It is also possible for an amplifier with lesser damping factor to generate less back-EMF induced error than one with a high damping factor if the former has phase coherent damping, as with JLH class-A and the latter does not. This is due to the numerical Sine value of the damping angle being a multiplicant.
The higher the damping factor the lower the output error and the greater the cone control to the limit of system resistance, but also the more dynamically induced ringing there will be with cable/crossovers etc.; this leading to a change in reproduced tonality.
Hi SY,
Back-EMf arises according to the cable/loudspeaker system. It may be at any frequency where there is electrodynamic activity and electrical reactivity, especially charging 'C's in series with 'L's. With a low output impedance amplifier series inductance can lead to mid frequency back-EMF inducing a higher frequency error; you can hear it through a tweeter but it won't show via any steady sinewave investigation ... it is caused by changing waveforms inducing the ringing.
Hi Darkfenriz,
I think Ben Duncan made some investigations that showed cable reflections in the 100kHz range .... where few amplifiers have coherent damping capabilities.
Hi Rodolfo,
If you examine a loudspeaker system only at the (steady sinewave) frequency at which it is driven, then your argument holds. But what happens after say a bass cone has compressed the cabinet air spring due to a large sub-resonant frequency pulse caused by an additive musical wave beat between two instruments ... the restitutional back-EMF arises at a different frequency and can significantly modify intended on-going drive.
Cheers ......... Graham.
Yes every amplifier has separately acting internal output impedance, but, not all amplifiers have impedance that causes a damping phase change at low/mid audio frequencies such that all damping above that frequency is in quadrature, so I do not know what you say is 'dealt with'.
Okay so all amplifiers have output impedance ... and what happens at the output of an inductive feedback loop ? ... there is a leading error voltage before current correction is capable of acting ... the error increasing with frequency or transients.
Thus it is possible for back-EMF to fractionally reverse bias and cause phase shifted output stage crossover distortion before the NFB loop can minimise the sensed error.
Even Rod Elliot shows something similar to this on his Distortion Analyser postings, and I believe that Mauro's spectral investigation could highlight these effects.
It is also possible for an amplifier with lesser damping factor to generate less back-EMF induced error than one with a high damping factor if the former has phase coherent damping, as with JLH class-A and the latter does not. This is due to the numerical Sine value of the damping angle being a multiplicant.
The higher the damping factor the lower the output error and the greater the cone control to the limit of system resistance, but also the more dynamically induced ringing there will be with cable/crossovers etc.; this leading to a change in reproduced tonality.
Hi SY,
Back-EMf arises according to the cable/loudspeaker system. It may be at any frequency where there is electrodynamic activity and electrical reactivity, especially charging 'C's in series with 'L's. With a low output impedance amplifier series inductance can lead to mid frequency back-EMF inducing a higher frequency error; you can hear it through a tweeter but it won't show via any steady sinewave investigation ... it is caused by changing waveforms inducing the ringing.
Hi Darkfenriz,
I think Ben Duncan made some investigations that showed cable reflections in the 100kHz range .... where few amplifiers have coherent damping capabilities.
Hi Rodolfo,
If you examine a loudspeaker system only at the (steady sinewave) frequency at which it is driven, then your argument holds. But what happens after say a bass cone has compressed the cabinet air spring due to a large sub-resonant frequency pulse caused by an additive musical wave beat between two instruments ... the restitutional back-EMF arises at a different frequency and can significantly modify intended on-going drive.
Cheers ......... Graham.
Note: My computer crashed some time ago, and I did not reload LTSpice, so at the moment I can't draw a circui or run simulations...
But as ingrast posted, the back EMF of a speaker is feeble compared to amp output (a typical speaker is less than 1% efficient). Most of the output will be at the ressonance freq as at ressonance efficiency rises significantly.
If one considers there is an air path between boxes, the speaker acting as a microphone will produce a very feeble output (comparable to a microphone - how large is the gain of a mic preamp?)
Now, in some circuits this output will be sent to the amp input due to high intrinsec output stage impedance (that was made lower by significant ammount of feedback).
Other circuits already have an intrinsecally low output impedance (before feedback) and will 'short circuit' this signal, so only a fraction of it will be sent to the input.
But as ingrast posted, the back EMF of a speaker is feeble compared to amp output (a typical speaker is less than 1% efficient). Most of the output will be at the ressonance freq as at ressonance efficiency rises significantly.
If one considers there is an air path between boxes, the speaker acting as a microphone will produce a very feeble output (comparable to a microphone - how large is the gain of a mic preamp?)
Now, in some circuits this output will be sent to the amp input due to high intrinsec output stage impedance (that was made lower by significant ammount of feedback).
Other circuits already have an intrinsecally low output impedance (before feedback) and will 'short circuit' this signal, so only a fraction of it will be sent to the input.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.