JensRasmussen said:
Nice to see you Jens....
1. The real world acoustic transducer is already emulated here somewhere in these posts and is used as a reference to simulate the amp + emulated load which describes the reactive elements as well as non-linear and resistive perspectives of output of an amplifier..
2. Especially when we drive 18" professional subwoofer its back EMF is certainly many times bigger than driving a speaker of 6"..
yes you are right...
3. It is not that its less to worry, but yes it would not disturb the other speakers because the transducer is connected an individual amplifier comprising an active system...
regards,
K a n w a r
Re: Back EMF is real !
I agree with the galvano meter test. Old hands may remember that when transporting such a galvanometer we would short the terminals. That way, the movements of the coil & pointer were damped by the EMF, a sort of electromech brake. Works well.
Jan Didden
ashok said:
I agree with the galvano meter test. Old hands may remember that when transporting such a galvanometer we would short the terminals. That way, the movements of the coil & pointer were damped by the EMF, a sort of electromech brake. Works well.
Jan Didden
andy_c said:
mmmh... are you sure about that? 😕
Imagine the following (typical) scenario. While your cone is still moving in one direction, the driving signal reverse polarity. Your amplifier will try hard to suddenly reverse the direction of cone movement. But of course the speaker moving mass is non-zero and inertia will prevent it to instantly follow any such sudden change.
What will happen then?
I'm no speaker expert, and have done no calculation or simulation whatsoever but, by intuition, my guess would be that under voltage drive, the instantaneous current in the VC will increase quite a bit (that is, the apparent load impedance will decrease -possibly to a negative value!-) until the movement of the cone goes back "in phase" (same direction) with the driving signal.
Under current drive, ideally the current will not change but the voltage will... and nevertheless the net result will be the same, an increase in the instantaneous power sunk by the load.
After all, the moving mass of the speaker possess a certain amount of kinetic energy; if you want to reverse the direction of that movement, you need to supply an amount of energy which is > than that amount. The quicker you want this change to happen (as for "high DF") the higher the power required to supply that much energy.
(thus, in voltage drive, the higher the current and of course the lower the apparent instantaneous load impedance).
Am I wrong? If yes, why?
Once more, back EMF here we go
It is mildly amusing for me at least, to continue hearing arguments regarding this issue.
The physics behind, and the math that allows to formally handle it are both well known perhaps for over a century at this time, and the bottom line is a speaker behaves as a plain complex load, nothing more, nothing less. How comes this?
First, there are the normal electrical components we are familiar with, in this case voice coil inductance and resistance. So far so good.
The quirk - that may result confusing for some - is how electromechanical phenomena are accounted for.
It happens that both in electrical circuits and mechanical systems - as also in other fields of physics - the descriptive equations have the same structure as for electrical phenomena. This is no mere coincidece but unveils a deeper nature that as remarked, surfaces in seemingly unrelated fields like thermodynamics.
The consequence of this formal coincidence, is that the relation among force, position, velocity and acceleration with mass, friction and elasticity, is described by exactly the same equations that in the electrical field relate voltage, charge, intensity and derivative of intensity with inductance, resistance and capacitance.
The connection then is made using the speaker motor as a transformer of sorts, where the primary side handles the electrical magnitudes voltage and current, while the secondary side handles the mechanical magnitudes force and velocity respectively, whit the only catch that the "turns ratio" Bl relates Votage with velocity and current with force (in a stright transformer, the ratio should be voltage to force and current to velocity).
This transformation is what allows to - dropping the transformer altoghether - insert in the primary side the former electromechanical compnents and magnitudes this time as virtual equivalent electrical magnitudes to be combined with the purely electrical ones of voice coil resistance and inductance.
A more accurate model includes not only the strictly speaker related elements as mass, frictonal losses and suspension elasticity, but also enclosure elements and - for utmost precission - radiation and even room effects. These latter elements are of such low relevance (talk about speaker eficiency) that are usually neglected.
What matters is that the final equivalent circuit - compossed only of electrical elements - includes all relevant electrical and mechanical components as to what affects what the amplifier sees, nothing less, nothing more.
I do not mean to say the lengthy behavioral descriptions sprinkled here and there in the thread regarding as to what is happenning with voltages, currents, cone movement and so forth are wrong. Simply they are more or less accurate as a way of visualizing the physical phenomena, but the real money is in dealing with the equivalent circuit for the sake of analysis, prediction and design.
Rodolfo
It is mildly amusing for me at least, to continue hearing arguments regarding this issue.
The physics behind, and the math that allows to formally handle it are both well known perhaps for over a century at this time, and the bottom line is a speaker behaves as a plain complex load, nothing more, nothing less. How comes this?
First, there are the normal electrical components we are familiar with, in this case voice coil inductance and resistance. So far so good.
The quirk - that may result confusing for some - is how electromechanical phenomena are accounted for.
It happens that both in electrical circuits and mechanical systems - as also in other fields of physics - the descriptive equations have the same structure as for electrical phenomena. This is no mere coincidece but unveils a deeper nature that as remarked, surfaces in seemingly unrelated fields like thermodynamics.
The consequence of this formal coincidence, is that the relation among force, position, velocity and acceleration with mass, friction and elasticity, is described by exactly the same equations that in the electrical field relate voltage, charge, intensity and derivative of intensity with inductance, resistance and capacitance.
The connection then is made using the speaker motor as a transformer of sorts, where the primary side handles the electrical magnitudes voltage and current, while the secondary side handles the mechanical magnitudes force and velocity respectively, whit the only catch that the "turns ratio" Bl relates Votage with velocity and current with force (in a stright transformer, the ratio should be voltage to force and current to velocity).
This transformation is what allows to - dropping the transformer altoghether - insert in the primary side the former electromechanical compnents and magnitudes this time as virtual equivalent electrical magnitudes to be combined with the purely electrical ones of voice coil resistance and inductance.
A more accurate model includes not only the strictly speaker related elements as mass, frictonal losses and suspension elasticity, but also enclosure elements and - for utmost precission - radiation and even room effects. These latter elements are of such low relevance (talk about speaker eficiency) that are usually neglected.
What matters is that the final equivalent circuit - compossed only of electrical elements - includes all relevant electrical and mechanical components as to what affects what the amplifier sees, nothing less, nothing more.
I do not mean to say the lengthy behavioral descriptions sprinkled here and there in the thread regarding as to what is happenning with voltages, currents, cone movement and so forth are wrong. Simply they are more or less accurate as a way of visualizing the physical phenomena, but the real money is in dealing with the equivalent circuit for the sake of analysis, prediction and design.
Rodolfo
Rodolfo,
Finally a sober voice in the choir! Though your reassumption is precise, it should be added that when you flip between voltage / current drive, the thing what you turn on/ off is exactly this transformer model. And, given the fact that the Bl "turns ratio" is nonlinear, and Zm is nonlinear, in this way one is excluding additional nonlinear elements from the feedback loop. And this way make it more easy drive for the amp.
Please notice that this highly nonlinear transformer model described above is hanged on the amplifier output, and as such, is EXCLUDED from the feedback loop, even in the case of a high feedback voltage drive! I think at this point it is totally analog to the tube amp /output transformer case, where you take the feedback BEFORE the output transformer! With the only difference that an output transformer is a much more linear device.
This is why I would never dream of reducing speaker distortion with a feedback loop which is not closed around the speaker.
the max you can do is to apply "a sort of electromech brake". Which works. AND introduces a whole lot of new distortions.
Ciao, George
Finally a sober voice in the choir! Though your reassumption is precise, it should be added that when you flip between voltage / current drive, the thing what you turn on/ off is exactly this transformer model. And, given the fact that the Bl "turns ratio" is nonlinear, and Zm is nonlinear, in this way one is excluding additional nonlinear elements from the feedback loop. And this way make it more easy drive for the amp.
Please notice that this highly nonlinear transformer model described above is hanged on the amplifier output, and as such, is EXCLUDED from the feedback loop, even in the case of a high feedback voltage drive! I think at this point it is totally analog to the tube amp /output transformer case, where you take the feedback BEFORE the output transformer! With the only difference that an output transformer is a much more linear device.
This is why I would never dream of reducing speaker distortion with a feedback loop which is not closed around the speaker.
the max you can do is to apply "a sort of electromech brake". Which works. AND introduces a whole lot of new distortions.
Ciao, George
Hi all,
I would like to ad the following informative text found on Wikipedia regarding EMF.
I did an extensive search on the net, and everywhere were EMF related to electrical motors etc. and batteries, nowhere could I see anything saying an EMF could exist when a capacitor/capacitive alike load is connected to a source (eg. battery for instance).
However the following text from University of Alberta's, Department of Physics, gave a possibility to interpret that so could be:
The bolded text in the quote above means interpreted that as long as an voltage source have an increasing voltage, there will be a current flowing into a capacitor.
In a loudspeaker element we don't have any capacitor, just the mechanical one which will give a capacitive reactive load, in this case we can't just rise the voltage to infinity to maintain a current.
This is for DC however but it shows that a loudspeaker element is diffrent to a capacitor even if it has a capacitive frequency region extending from the resonance frequency to a point were the impedance is flattening out and becomming resistive and with frequency also inductive.
So how can we interpret the use of the EMF abreviation regarding loudspeaker elements, well loudspeaker elements are for sure quite complex load and I would tend to say that if we are talking about EMF then there's a "fuzzy logic" just grouping the whole loudspeaker element as a complex load without going into detail exact how the currents and voltages are looking for a particular frequency and drive level, and there's places for a complex current and voltage phase and linearity models too.
As I have mentioned before I suggested to use the phrase "reactive load", and maybe this is suitable if we are just trying to model up the complex load of a loudspeaker element into simple and plain inductors, capacitors and resistors were we can explain for any particular frequency if the driver is acting capacitively or inductively and find values to put the finger on.
To conclude it arises that we can see at the complex load and use both ways to express depending how it's fitted.
Back EMF/reactive load and amplifiers, I guess the last word is not yet said anyhow, theirs as far as I see more to find out regarding loudspeaker elements and their behaviour in combination with an audio amplifier is my guess, so far too many are simplyfying the loudspeaker elements equivalent load bahaviour.
But I will have to check this out first and make postings later.
Graham,
from your post #172:
focusing on you saying in above quote "...can be reactively delayed...", could you develope that part of the sentence, how do you see at it.
Ok then I understood you right, but what do you mean with "flat reverse phase response", where in an amplifier circuit can these arteffects arise?
More questions then comments to your post this time, just trying to follow you and understand your comments, will return later.
Cheers Michael
PS: Kanwar,
could you please stick to technical writings, you don't have to show such a need'a-group-belonging psychological behaviour and lacking on others not "belonging" to the same camp as you.
I would like to ad the following informative text found on Wikipedia regarding EMF.
I did an extensive search on the net, and everywhere were EMF related to electrical motors etc. and batteries, nowhere could I see anything saying an EMF could exist when a capacitor/capacitive alike load is connected to a source (eg. battery for instance).
However the following text from University of Alberta's, Department of Physics, gave a possibility to interpret that so could be:
The bolded text in the quote above means interpreted that as long as an voltage source have an increasing voltage, there will be a current flowing into a capacitor.
In a loudspeaker element we don't have any capacitor, just the mechanical one which will give a capacitive reactive load, in this case we can't just rise the voltage to infinity to maintain a current.
This is for DC however but it shows that a loudspeaker element is diffrent to a capacitor even if it has a capacitive frequency region extending from the resonance frequency to a point were the impedance is flattening out and becomming resistive and with frequency also inductive.
So how can we interpret the use of the EMF abreviation regarding loudspeaker elements, well loudspeaker elements are for sure quite complex load and I would tend to say that if we are talking about EMF then there's a "fuzzy logic" just grouping the whole loudspeaker element as a complex load without going into detail exact how the currents and voltages are looking for a particular frequency and drive level, and there's places for a complex current and voltage phase and linearity models too.
As I have mentioned before I suggested to use the phrase "reactive load", and maybe this is suitable if we are just trying to model up the complex load of a loudspeaker element into simple and plain inductors, capacitors and resistors were we can explain for any particular frequency if the driver is acting capacitively or inductively and find values to put the finger on.
To conclude it arises that we can see at the complex load and use both ways to express depending how it's fitted.
Back EMF/reactive load and amplifiers, I guess the last word is not yet said anyhow, theirs as far as I see more to find out regarding loudspeaker elements and their behaviour in combination with an audio amplifier is my guess, so far too many are simplyfying the loudspeaker elements equivalent load bahaviour.
But I will have to check this out first and make postings later.
Graham,
from your post #172:
focusing on you saying in above quote "...can be reactively delayed...", could you develope that part of the sentence, how do you see at it.
Ok then I understood you right, but what do you mean with "flat reverse phase response", where in an amplifier circuit can these arteffects arise?
More questions then comments to your post this time, just trying to follow you and understand your comments, will return later.
Cheers Michael
PS: Kanwar,
could you please stick to technical writings, you don't have to show such a need'a-group-belonging psychological behaviour and lacking on others not "belonging" to the same camp as you.
Ultima Thule said:
My initial objection to the "particular" (ab)use of the term "back EMF" is that it has been well understood for many years and it adds nothing but confusion if we insist on calling a white cat black - which the back EMF camp is effectively doing.
In addition, all the pile-on of irrelevant and irrational terminologies used in the descriptions of back EMF and how it affects amp/sound makes things worse.
What the back-emf crowd failed so miserably to do is in the department of explaining how this could affect sound AND demonstrate its real life impact.
My view of this whole thing is that as we have limited resources, we ough to focus on things that make a difference. A good engineer focuses his efforts on major issues and ignores the rest; and a bad engineer tries to cover every base. a resistive load can be viewed as a simple 1st order approximation of a real life speaker, and it is probably 80% accurate. a rlc network is like a 2nd order approximation of the speaker and is probably 95% accurate; introducing non-linear elements to the above will further improve our model and may be 99% accurate. The point is that each level of complication introduces marginal but diminishing gain. This is especially costly if you, like some in the back emf camp, do not have a clear understanding (intuition?) of how this impacts sound.
Joseph K said:
You are right, the previous post only dealt with the linear model which for most purposes is fairly adequate.
With respect to the nonlinear behavior, I am working (as time allows) in a sensitivity analysis starting with the voltage to velocity and current to velocity transforms as per Hawksford's paper cited some time back. Anyway I stress again this is of significance for low frequencies with emphasis on full range schemes, where displacements and temperature excursions have some impact.
Appart from these extremes, it has been confirmed by actual distortion measurements that nonlinearity is not that severe as may be implied looking at the Bl, Zm and Le curves with xm, and Re curve with T.
Perhaps a further comment with respect to the linear model, where it has been said delay is not taken into account, is that it is essentially not true. The electromechanical model is a lumped one in the sense it is assumed signal propagation is instantaneous. This is a fairly accurate assumption whenever the wavelengths involved are extremelly large in comparison with actual device dimensions. Only for radiofrequencies do physical dimensions matter, and we switch accordingly to a distributed element model.
Rodolfo
Hi Rodolfo
nice to see the discusion is more precise now.
Thank you (and others).
Of course the whole speaker system can be described by R L C components, more or less linear. But the point is most amps are tested on resistive load with steady state sine. And we really need to analyse what's going on in reality.
And back EMF DOES AFFECT the work of an amplifier.
Take a darlington output amp.
The output voltage with output current are phase shifted. Furthermore the output current affects the work of VAS, because it sucks a little fraction of VAS current (output current divided by product of betas).
So in the end voltage signal is distorted, which must be corrected by global feedback.
Take a transcondutance amp.
Open loop damping is zero.
All the work to keep output voltage stiff relies on feedback.
Take a mosfet output amp.
The transconductance of power mosfet is low, so backEMF causes voltage to be different on different loads. Again a work to do by feedback.
Problems of feedback are already very well known: intermodulation, harmonic distortion, phase shift and in many poor amps lowering feedback factor with frequency due to weak freq. compensation ( a dominant pole within audio range).
This all causes the amp to have times lower performance than suggested by thd numbers with steady sine/resistive load experiment.
And of course there are ways to prevent such behaviours.
......
regards
nice to see the discusion is more precise now.
Thank you (and others).
Of course the whole speaker system can be described by R L C components, more or less linear. But the point is most amps are tested on resistive load with steady state sine. And we really need to analyse what's going on in reality.
And back EMF DOES AFFECT the work of an amplifier.
Take a darlington output amp.
The output voltage with output current are phase shifted. Furthermore the output current affects the work of VAS, because it sucks a little fraction of VAS current (output current divided by product of betas).
So in the end voltage signal is distorted, which must be corrected by global feedback.
Take a transcondutance amp.
Open loop damping is zero.
All the work to keep output voltage stiff relies on feedback.
Take a mosfet output amp.
The transconductance of power mosfet is low, so backEMF causes voltage to be different on different loads. Again a work to do by feedback.
Problems of feedback are already very well known: intermodulation, harmonic distortion, phase shift and in many poor amps lowering feedback factor with frequency due to weak freq. compensation ( a dominant pole within audio range).
This all causes the amp to have times lower performance than suggested by thd numbers with steady sine/resistive load experiment.
And of course there are ways to prevent such behaviours.
......
regards
Rodolfo,
I would be really interested in your findings. And I agree that the numbers can change and the drivers can become more sophisticated - though the one investigated by Hawksford was one of the best of that time.
But for me all this does one thing: allows the technically inferior [the voltage drive] method to be applied with less detrimental effect.
This point of mine is based on the hawksford article, as an explanation of some personal listening experience. Until You can proove the contrary. 🙂
[Then will remain the listening experience to be explained.]
As regards extremes, for me the standard is the two / two &half way bookshelf / small floorstander, where all these effects are well exposed, end the extreme is the 30 -40 cm diam, 0.5 mm Xmax wide / full range, maybe in a horn, where these effects are mitigated.
I wonder what percent of us is actually listening to OB full rangers or B&W Nautiluses and the like. [though I would do that happily at any time.. 😀 The rest listens to something very similar to that Celestion in the article.
As for the delay, are you talking about electrical or acoustical wavelenghts, in the electromechanical model?
Ciao, George
I would be really interested in your findings. And I agree that the numbers can change and the drivers can become more sophisticated - though the one investigated by Hawksford was one of the best of that time.
But for me all this does one thing: allows the technically inferior [the voltage drive] method to be applied with less detrimental effect.
This point of mine is based on the hawksford article, as an explanation of some personal listening experience. Until You can proove the contrary. 🙂
[Then will remain the listening experience to be explained.]
As regards extremes, for me the standard is the two / two &half way bookshelf / small floorstander, where all these effects are well exposed, end the extreme is the 30 -40 cm diam, 0.5 mm Xmax wide / full range, maybe in a horn, where these effects are mitigated.
I wonder what percent of us is actually listening to OB full rangers or B&W Nautiluses and the like. [though I would do that happily at any time.. 😀 The rest listens to something very similar to that Celestion in the article.
As for the delay, are you talking about electrical or acoustical wavelenghts, in the electromechanical model?
Ciao, George
darkfenriz said:
And this is the bottom line. We should rather erradicate such terms as back EMF which might lead to endless discussions, and concentrate on finding a reasonable standard - or set of standards to cover a representative spectrum - of real world reactive loads on which to test amplifiers.
My hunch is it is possible to agree on a load exhibiting a moderate Q resonance in the 50 - 100 Hz, possibly with additional resonances at about 500 Hz and / or 2000 Hz or thereabouts. Results with this kind of load may not equal a particular box / driver combination but will stress the amplifier with operating conditions similar to real life and may be as good as it gets to uncover weaknessess.
Rodolfo
Joseph K said:
Electrical wavelengths. All phenomena included in the first level model are affected by electrical / electromagnetic propagation delays except enclosure loading, and this bears on low frequencies (so delays are less of an issue).
Other models including for example cone flexure modes etc. should include the corresponding physical effects only this time with acoustic wavelength being the criteria for establishing delay relevance.
Rodolfo
ingrast said:
You would think so, but when I measure amplifiers into such
loads, they usually get better, not worse, because the
impedance has gone up, and the performance is more sensitive
to the amount of current than the phase angle.
Others have tried this, but I have not seen any really usable
information as a result.
Nelson Pass said:
This is interesting. Others do not think load phase is secondary, while I do not have but a theoretical ground to score its importance.
Actual experience feedback like yours is most relevant.
Rodolfo
ingrast said:
Depends on what you measure and what you think is important
to measure. In the example I gave, we would be looking at
common harmonic or IM distortion.
This is interesting mr. Pass indeed
but it depends on what kind of amplifier you measured.
If it was zero global feedback mosfet as I would suppose, than it is clear that most problems related to reactive loads are gone. With bipolar output it could be something else, because output current affects the VAS.
regards
but it depends on what kind of amplifier you measured.
If it was zero global feedback mosfet as I would suppose, than it is clear that most problems related to reactive loads are gone. With bipolar output it could be something else, because output current affects the VAS.
regards
Nelson Pass said:
Barring special purpose applications like musical instrument amplifiers, I guess we all look after reproduction accuracy as what's important.
This translates to as little as practical deviations from flat frequency response (arguably phase linearity also) and amplitude linearity (harmonic and IM distortion products).
Rodolfo
darkfenriz said:
Certainly. A big factor is the characteristic of the load line
experienced by the gain devices. We will expect a triode
routinely optimized around a resistive load line to experience
more harmonic and IM distortion with a reactive load, as that
load line will evolve from a straight line to an ellipse.
A typical bipolar amp actually enjoys some benefit from simple
reactance, as it leaves more voltage across the devices
at the points of highest current. If I measure distortion into
a capacitor or inductor at a frequency giving the same (absolute)
load impedance, I often see lower THD at higher power levels.
Back EMF at play
THD of a well known Class D amplifier with a resistor load:
And the exact same amplifier, now with a typical 2 way speaker load:
Both conditions 10 W 4 Ohms load.
Note the scale difference!
(If you're gonna do this test at home, use ear protection!!!)
THD of a well known Class D amplifier with a resistor load:
An externally hosted image should be here but it was not working when we last tested it.
And the exact same amplifier, now with a typical 2 way speaker load:
An externally hosted image should be here but it was not working when we last tested it.
Both conditions 10 W 4 Ohms load.
Note the scale difference!
(If you're gonna do this test at home, use ear protection!!!)
- Status
- Not open for further replies.