mauropenasa said:
Mauro,
I did not mean that negatively. I was under the impression that you advocated a current-output amp to suppress the EMF *in the amp*, which will work. But then you end up with the EMF signal being present at the point where the amp connects to the speaker, and that will do interesting things between the EMF signal and the wanted signal. It is the same result you get with low damping factor. You keep the EMF out of the amp, but get worse signal at the speaker. Hence my question: do you want low EMF in the amp or good sound.
But you can have a voltage output amp with high damping factor using a transconductance output stage, and that would of course, as you say, be different from my description.
Jan Didden
janneman said:
Just go to the post you want to link to. In the upper right is a "Post #xxxx" link. Click on it and that post will become the current location on its page. Its URL will end up in the address bar of your browser. Highlight and copy the URL to the clipboard. Then do the following:
{url=http://put.url.here}This text will become a link.{/url}
but replace the four curly braces above with square brackets.
Well, since overbiasing goes towards class A, it MUST sound better, no?
Or as Floyd Toole once said, "Believing is hearing"
.
andy_c said:
Sorry, couldn't refrain from suggesting a faster one (not to correct andy_c)
Instead of opening the "Post #xxx" link, right click and select "Copy shortcut".
Then you can also (a matter of taste) type as suggested by Andy_c or press the "http://" button in the messge form above the text area. The first text control allows to type the link text. Press [Enter] and in the second text control paste (or press Ctrl-V) to insert the previously saved link, press [Enter] and done. A word of caution, the link html is placed at the bottom of the message, if you were editing in the middle of text, you must cut and paste it where it belongs.
Rodolfo
PS. About the sims. There is something worring me, in that the simulation conditions are for 0 output voltage. This should not matter if device behavior were irrelevant of output voltage, but for second order effects as we are interested, I'm, afraid it is not the case.
Just 2 cents.
ingrast said:
I just used your technique below. It's way faster and better than mine, thanks.
ingrast said:
I did discuss this in my original post. My original text read:
Here's a plot with the output voltage also stepped from -30V to +30V in 5V steps.
Attachments
Now, here we have something that's really interesting.
Let's suppose a VCCS output - very high Zout. But speakers require a voltage source, so feedback is used to lower effective Zout.
As stated, the EMF current wouldn't have anywhere to go - but to the feedback path, or to the amp input.
Let's suppose we have two feedbak amps, equal damping factor at LF, one with a high internal loop Zout, the other one with a quite low internal loop Zout, which would be the better one?
My vote goes for the second.
Let's suppose a VCCS output - very high Zout. But speakers require a voltage source, so feedback is used to lower effective Zout.
As stated, the EMF current wouldn't have anywhere to go - but to the feedback path, or to the amp input.
Let's suppose we have two feedbak amps, equal damping factor at LF, one with a high internal loop Zout, the other one with a quite low internal loop Zout, which would be the better one?
My vote goes for the second.
Speaker model
As promised in earlier posts, I've finished a first rough analysis of an actual speaker load, in this case a high end 2 way reference box.
The model includes to a fair approximation both electrical and electromechanical elements, and the so called "back EMF" results simply from normal complex loading.
Not wanting to derail the current thread direction, if somenone feels interesting to go ahead whith this issue, please advise and we can start and independent thread.
Attached is the PDF with the analysis.
Rodolfo
As promised in earlier posts, I've finished a first rough analysis of an actual speaker load, in this case a high end 2 way reference box.
The model includes to a fair approximation both electrical and electromechanical elements, and the so called "back EMF" results simply from normal complex loading.
Not wanting to derail the current thread direction, if somenone feels interesting to go ahead whith this issue, please advise and we can start and independent thread.
Attached is the PDF with the analysis.
Rodolfo
.
amplifierguru said:
Greg,
Of course. I was merely referring to the tendency of people to connect things like:
"Overbiasing goes toward class A. Class A sounds better. Therefore, overbiasing sounds better."
And, as almost anybody who understands a little of human nature knows, such connections tend to influence the outcome of listening tests.
Jan Didden
Nelson Pass said:
Actually, I believe that if it is really a current source, then the
EMF won't do a thing to the speaker, since it would require
a place for the current to go. No current, no force.
... but it does have a way to go! Not into the amp (that's infinite Z, ideally) but into the speaker itself. The speaker acts as a voltage source, generates the EMF at its terminals. There it finds the signal from the amp (actually, the current the amp tries to force into the speaker, which generates a voltage across the complex speaker-cum-xover imprdance). So at the speaker terminals you find the resulting signal, and that causes a current through the speaker and ultimately the sound of course.
Jan Didden
No, no, no!
These are not subject to democratic decision making - these are precise definitions.
A controlled CURRENT source produces a current across it's output terminals in function of it's control signal present on it's input terminals.
This is the definition.
And this implies that
-It does not take care of the tensions present at it's output terminals. This means that in case of any voltage forced upon [between] it's output terminals it will ACTIVELY modify it's output VOLTAGE until the resulting current flow will respond only and uniquely to the control signal on it's inputs.
-It does not take care of the load changes, as well, so it's output current will again be unchanged even in case of changing loads.
Please note that the high output Z was not included in the definition, it's a simple consequence of the definition!
If any foreign signal [EMF] present at the output terminals would be able to produce any kind of CURRENT CHANGE IN THE LOUDSPEAKER, that would be a violation of the definition, that is, it would be NOT a current source, that is, NOT high Z.
So, no currents caused by EMF on the output of an [ideal] transimpedance amplifier. In case of an amplifier with real output impedance, these currents will be present, but less - and more controlled - than in case of O [zero] output impedance.
This is a refrain constantly returning. Could somebody elaborate on it for me, please?
As an elementary first step, I can see that the coil acceleration [=sound pressure?] is uniquely depending on the coil current. Then, as Ingrast had explained [ehm, with Thiele -Small] in case of some special circumstances [constant impedances] this can be translated into a voltage across terminals - sound pressure equation, though complex and containing some variables..
So, when applying voltage to a current transducer, we are simplifying the equation [leaving out some variables]. Can this be translated into "speakers REQUIRE a voltage source"?
I have tried a simple spectrum analysis test. I knocked on the membrane of a loudspeaker. Listening closely, it can be clearly heard the difference between the terminals unconnected; connected to a low impedance output amp; shorted; connected to a 7 ohm impedance output amp.
First of all, the resulting THUMP has a clear character, which is defined by the actual speaker. This character changes in the BASS region, when one changes termination, but not in the upper spectrum. In the bass it becomes more dead, in case of a low Z termination. Interestingly it becomes more dead with the amp that with a short.
So, what I saw is that attenuating COIL movement has it's effects until the full cone moves like a piston. Then we can forget about it..
Then, undamped, the resulting sound is something coloured, but not distorted. If you bang on a contrabass, you don't get anything distorted - it resonates.
So, all this upheaval and hundred amps output stages only for eliminating that bass resonance? Can it not be done also other ways? [Also, I said "eliminating" but it's rather "modifying"]
Excuse me for the long rant!
Ciao, George
These are not subject to democratic decision making - these are precise definitions.
A controlled CURRENT source produces a current across it's output terminals in function of it's control signal present on it's input terminals.
This is the definition.
And this implies that
-It does not take care of the tensions present at it's output terminals. This means that in case of any voltage forced upon [between] it's output terminals it will ACTIVELY modify it's output VOLTAGE until the resulting current flow will respond only and uniquely to the control signal on it's inputs.
-It does not take care of the load changes, as well, so it's output current will again be unchanged even in case of changing loads.
Please note that the high output Z was not included in the definition, it's a simple consequence of the definition!
If any foreign signal [EMF] present at the output terminals would be able to produce any kind of CURRENT CHANGE IN THE LOUDSPEAKER, that would be a violation of the definition, that is, it would be NOT a current source, that is, NOT high Z.
So, no currents caused by EMF on the output of an [ideal] transimpedance amplifier. In case of an amplifier with real output impedance, these currents will be present, but less - and more controlled - than in case of O [zero] output impedance.
This is a refrain constantly returning. Could somebody elaborate on it for me, please?
As an elementary first step, I can see that the coil acceleration [=sound pressure?] is uniquely depending on the coil current. Then, as Ingrast had explained [ehm, with Thiele -Small] in case of some special circumstances [constant impedances] this can be translated into a voltage across terminals - sound pressure equation, though complex and containing some variables..
So, when applying voltage to a current transducer, we are simplifying the equation [leaving out some variables]. Can this be translated into "speakers REQUIRE a voltage source"?
I have tried a simple spectrum analysis test. I knocked on the membrane of a loudspeaker. Listening closely, it can be clearly heard the difference between the terminals unconnected; connected to a low impedance output amp; shorted; connected to a 7 ohm impedance output amp.
First of all, the resulting THUMP has a clear character, which is defined by the actual speaker. This character changes in the BASS region, when one changes termination, but not in the upper spectrum. In the bass it becomes more dead, in case of a low Z termination. Interestingly it becomes more dead with the amp that with a short.
So, what I saw is that attenuating COIL movement has it's effects until the full cone moves like a piston. Then we can forget about it..
Then, undamped, the resulting sound is something coloured, but not distorted. If you bang on a contrabass, you don't get anything distorted - it resonates.
So, all this upheaval and hundred amps output stages only for eliminating that bass resonance? Can it not be done also other ways? [Also, I said "eliminating" but it's rather "modifying"]
Excuse me for the long rant!
Ciao, George
Hi Rodolfo, Thanks of your work (ldc6.pdf).
Hi Jorge, you set a problem that for me it is important. How to all, you have a conclusion. I have an opposite conclusion, but I can be in error.
I that have been criticized because I sustained that NFB has un great limits of: phase- coherence, V/I coherence, etc... If it is true, that NFB works in "nearly ideal" way ( and I make a mistake me ), theoretically the presence of back-EMF in input (VCCS) allows to the NFB to elaborate the present THD in Back-EMF and reduce of fact the THD of the speaker , exactly as happens for the thds of the output stage.
Hi, George
I want write what it has written you, but doesn't know as traslate in so effective way. It does me like that the base electrotechnics theories is valid at least for a lot of members...
Ciao
Mauro
Hi Jorge, you set a problem that for me it is important. How to all, you have a conclusion. I have an opposite conclusion, but I can be in error.
I that have been criticized because I sustained that NFB has un great limits of: phase- coherence, V/I coherence, etc... If it is true, that NFB works in "nearly ideal" way ( and I make a mistake me ), theoretically the presence of back-EMF in input (VCCS) allows to the NFB to elaborate the present THD in Back-EMF and reduce of fact the THD of the speaker , exactly as happens for the thds of the output stage.
Hi, George
I want write what it has written you, but doesn't know as traslate in so effective way. It does me like that the base electrotechnics theories is valid at least for a lot of members...
Ciao
Mauro
Joseph K said:
George,
No, no, no! I'm sorry to disagree!
You are right with the definition etc of the current source. The current, going into the load (speaker), as a result of the amp input voltage, is independent of whatever happens at the amp output. But note, I said: as a result of the input voltage to the amp. Being H-Z, you can add any other signal from another source to the amp output terminal, for instance from the EMF. The resulting current flow through the speaker is a combination of the current resulting from the amp input signal, and the current as a result of the generated EMF. After all, that speaker is just a voltage source with some series- and parallel impedance. And just as a voltage source, the current delivered by the source through the impedance external to the source also flows internal through the source.
Jan Didden
Joseph K said:
I believe you agree that a high damping factor is necessary for good bass reproduction. High damping factor means the amp has a low output impedance.
A voltage source is equivalent to an ideal voltage generator (zero output impedance) with a low impedance in series.
A current source is equivalent to an ideal voltage generator (zero output impedance) with a very high impedance in series - or a very low damping factor (and the bass box will boommmm)
Jan,
I we would always agree how could we discuss..
This is why I emphasised,
That is, in my view, the speaker is a kind of voltage source [with it's ~ 6ohm series resistance] , yes. It's coil's movement in the magnet field produces TENSION across it's terminals. If we don't close the path across these terminals, there is no current generated, no forces awakening due to this current, no coil movement damping. Without an amplifier connected, and the speaker terminals shorted or not, it's clearly distinguishable.
Now, if we connect a 1Mohm output impedance amplifier, the tension generated by EMF will cause ~ microamper currents flowing, I would say, with no effect. This same amplifier in the same time is supposed to have it's own one million volt driving voltage input, so as to be able to drive through several amps. So, our speaker's terminal's are effectively shorted by a 1Mohm load; in the same time there is 1 amper signal current flowing. I would say the boss in the house is the million volt driving voltage...
Ciao, George
I we would always agree how could we discuss..
This is why I emphasised,
That is, in my view, the speaker is a kind of voltage source [with it's ~ 6ohm series resistance] , yes. It's coil's movement in the magnet field produces TENSION across it's terminals. If we don't close the path across these terminals, there is no current generated, no forces awakening due to this current, no coil movement damping. Without an amplifier connected, and the speaker terminals shorted or not, it's clearly distinguishable.
Now, if we connect a 1Mohm output impedance amplifier, the tension generated by EMF will cause ~ microamper currents flowing, I would say, with no effect. This same amplifier in the same time is supposed to have it's own one million volt driving voltage input, so as to be able to drive through several amps. So, our speaker's terminal's are effectively shorted by a 1Mohm load; in the same time there is 1 amper signal current flowing. I would say the boss in the house is the million volt driving voltage...
Ciao, George
Hello, Mauro
I've built a model of your transimpedance output stage, with some resistors values carefully adjusted so the Howland pump (output amp) has a really high output impedance under no feedback (on the order of 500K ohms alt low freq).
Then I applied a 1V voltage generator + 5 ohms resistor as a back EMF generator.
mauropenasa said:
I've built a model of your transimpedance output stage, with some resistors values carefully adjusted so the Howland pump (output amp) has a really high output impedance under no feedback (on the order of 500K ohms alt low freq).
Then I applied a 1V voltage generator + 5 ohms resistor as a back EMF generator.
Attachments
Detrhoning back EMF
Gentlemen:
We seem to be chasing after that entity - back EMF - and pondering on its efects (or lack thereoff) on sound reproduction as if it were some loudspeaker related peculiar phenomena.
Truth is by any well accepted circuit theory tennets, this back EMF is nothing more than the regular extension of Ohm's law to complex impedances. As much as a resistor "generates" a back EMF R*I.
The only difference is that a speaker - like an electric motor or other transducer - contains impedance components that reflect in an electrical fashion underlying mechanical phenomena. This difference is not crucial per se, but eventually in what implies as added elements to transduction accuracy.
So far, I belive our model is fairly accurate and useful. The discussion should be rather on how well we are controling the end variable - sound pressure - given the intervening undesired and unavoidable extra elements in the signal path.
If we accept voice coil velocity equals sound pressure, then we can focus on overall amplifier voltage output - voice coil velocity transfer function and then identify what topological features have impact on it.
Rodolfo
Gentlemen:
We seem to be chasing after that entity - back EMF - and pondering on its efects (or lack thereoff) on sound reproduction as if it were some loudspeaker related peculiar phenomena.
Truth is by any well accepted circuit theory tennets, this back EMF is nothing more than the regular extension of Ohm's law to complex impedances. As much as a resistor "generates" a back EMF R*I.
The only difference is that a speaker - like an electric motor or other transducer - contains impedance components that reflect in an electrical fashion underlying mechanical phenomena. This difference is not crucial per se, but eventually in what implies as added elements to transduction accuracy.
So far, I belive our model is fairly accurate and useful. The discussion should be rather on how well we are controling the end variable - sound pressure - given the intervening undesired and unavoidable extra elements in the signal path.
If we accept voice coil velocity equals sound pressure, then we can focus on overall amplifier voltage output - voice coil velocity transfer function and then identify what topological features have impact on it.
Rodolfo
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.