maybe it helps to lok at the Thevenin or Norton equivalents of a circuit https://en.wikipedia.org/wiki/Thévenin's_theorem
any linear source (eg amp) can be modelled by a voltage source in series with an output impedance (thevenin) or a current source shunted by an impedance (Norton). This also applies to the amplifier plus xover network - the Thevening impedance of this is what matters for how the distortion back EMF (varying inductance, hysteresis etc) is converted to current (which is converted to sound and shows up as distoriton)
any linear source (eg amp) can be modelled by a voltage source in series with an output impedance (thevenin) or a current source shunted by an impedance (Norton). This also applies to the amplifier plus xover network - the Thevening impedance of this is what matters for how the distortion back EMF (varying inductance, hysteresis etc) is converted to current (which is converted to sound and shows up as distoriton)
There are reasons for avoiding those steeper slopes and use the right drivers that don't need it.
Also, EQ the current and you can drive resonances without any loss of stamping and it does not matter if you are using current or voltage 'drive' as it works with any source impedance. Look at the Elsinores as that is what they are doing. You can drive them with a current source and not lose any damping.
Make the amplifier produce the same current at ALL frequencies and both LF alignment as well as crossovers get locked IN. No, I am not kidding!
Red is voltage. Green is current. The slight deviation at 20 Hertz is because this is a vented box. If this was a sealed box design, even that would not be there. As it is, above at low frequencies, the power handling will only be slightly less.
Even the crossover does not change:
Voltage first:
Now current:
Try this with conventional design. The above is the Elsinore ULD.
Also, EQ the current and you can drive resonances without any loss of stamping and it does not matter if you are using current or voltage 'drive' as it works with any source impedance. Look at the Elsinores as that is what they are doing. You can drive them with a current source and not lose any damping.
Make the amplifier produce the same current at ALL frequencies and both LF alignment as well as crossovers get locked IN. No, I am not kidding!
Red is voltage. Green is current. The slight deviation at 20 Hertz is because this is a vented box. If this was a sealed box design, even that would not be there. As it is, above at low frequencies, the power handling will only be slightly less.
Even the crossover does not change:
Voltage first:
Now current:
Try this with conventional design. The above is the Elsinore ULD.
Hello,
You are on target. It is the distortion that counts.
It is time for some measurements in this thread.
Thanks DT
Hello Joe,
Saying it does not make it so.
It is time to post some measurements of load impedance, phase, transconductance amplifier voltage output and current output.
Perhaps to start with you can put your transconductance amplifier into a simulator and show us the results.
The MATLAB transconductance amplifier model I am looking at does not show output voltage or impedance being bullied around by output current.
Voltage is the independent variable. Current does what it is told to do.
It is a voltage controlled current output amplifier.
Yes I do understand 99%+ of the amplifiers used are voltage sources.
Thanks DT
Experiments done by me and others don't show current drive to produce less distortion per se. It is highly dependant on the driver and the frequency the measurements are done. If this were the right tree, you would find many more dogs barking against it.
Any half competent engineer can turn a voltage controlled amp into a current controlled one. I am just 1/4 competent and even I could make a couple where you could dial in the desired output resistance, and one where for example on the high end it functions as a current source, and a voltage source on the low end. All to do measurements and find out for myself.
It ain't worth the bother.
Any half competent engineer can turn a voltage controlled amp into a current controlled one. I am just 1/4 competent and even I could make a couple where you could dial in the desired output resistance, and one where for example on the high end it functions as a current source, and a voltage source on the low end. All to do measurements and find out for myself.
It ain't worth the bother.
Last edited:
That is disappointing.
It seems you want to talk about transconductance amplifiers. This is not a forum about amplifiers.
This forum is about a speaker, in fact, Elsinores. It has some interesting design features, that means it can be used with both kinds of amplifiers.
Note how the Elsinores do something that no commercial speakers do.
EQ the current flat: The speaker FR is locked in, the bass alignment is almost perfectly locked in, and finally, even the crossovers are locked in.
Here it is below, can you spot which is voltage source or current source?
That is pretty impressive, don't you think? What is the key to making this happen, it is simple:
Make/force the amplifier to produce the same current at ALL frequencies, from zero Hertz and up.
If you want to have a discussion about transconductance amplifiers (voltage-controlled current sources), there is already a forum here:
Link: "Trans-Amp" - 40 Watt Transconductance "Current Amplifier"
The Elsinore's can be driven equally well with both voltage sources and current (transconductance) sources. It does not matter which. I don't use my trans-amp for listening, I use it as a tool.
My actual listening amplifier has an output impedances of 3 Ohm on 8 Ohm tap and 1.5 Ohm on 4 Ohm tap,
This is 3 Ohm output impedance compared to 0 Ohm:
Can you tell which is which? Note that there is no reduction in damping or otherwise the response would be up in the 80-100 Hertz and it is not.
IMPORTANT: YOU CAN GET GOOD BASS DAMPING WITH A CURRENT SOURCE!
They say you can't, but the above is proof that you can.
Let us compare when we take away the Current EQ below 100 Hertz approx:
Once again, there is a widespread belief that you cannot get good damping by using a current source. But the above proves otherwise. EQ the current and you can use it with any amplifier without loss of damping. BTW, the alignment in the Elsinores is similar to a Bessel and not Butterworth, which means a more damped alignment than Butterworth, and this damping is kept even when using a current source.
The Elsinores can be driven by any amplifier.
Now as for me just saying so does not make it so. Well, I have not changed I=VR so that does make it so.
Below I have changed the wording slightly, but the thought is the same:
A voltage source controls the voltage across the voice coil, it has no control over the current (through the voice coil).
A current source controls the current through the voice coil, it has no control over the voltage (across the voice coil).
V=I*R
With a voltage source, V is constant, and the I is determined by R.
With a current source, I is constant, and the V is determined by R.
So with a voltage source you can only control the voltage. The current is determined by an unpredictable load. I=V/R
With a current source, you can only control the current. The voltage is determined by the load. V=IR
But with a current source, the R cannot impede the current (because the source impedance is infinitely high).
This is all pure Ohm's Law.
If you want to talk about trans-amps, use the above link.
If you want to discuss loudspeakers as pertaining to Elsinore's design, please, ask any question you like.
For example, how does current EQ make it possible to use current drive even in the bass? That is if you want to. But I use an amplifier that does not have a low enough output impedance to be a voltage source and is nowhere near high enough to be a current source. But my trans-amp has an output impedance of 270 Ohm. But I don't use it. It is a tool to me.
Crossovers too. Once again, the convention is that they need a voltage source to work and any added series impedance will cause the Crossover to function incorrectly. Yet if you EQ the current of the amplifier, this does not happen.
So these design features have some desirable and practical use. Do they also improve the sound? Oh yes, I believe so and so do others.
I agree.
Particularly about the driver.
Have you tried to EQ the current and then done measurements?
But this forum is not about amplifiers. They are about a pair of speakers that does work with a current source if that is your wish.
It is not mine. But it is simply proof that this speaker design does in effect cancel out the output impedance of the amplifier, no other claim is made, beccause it does what it does!. Does it improve the sound? I think it does, but ultimately the proof is the speakers and it is up to those who have built them.
But again, DT seemed to think I wanted to talk about current source power amps. Actually, not really. They are old hat to me.
Hello Joe,
When you say EQ the current flat as in making the amplifier produce the same current at all frequencies are you describing how a zobel network works? Thanks https://en.wikipedia.org/wiki/Zobel_network
Cheers
When you say EQ the current flat as in making the amplifier produce the same current at all frequencies are you describing how a zobel network works? Thanks https://en.wikipedia.org/wiki/Zobel_network
Cheers
Much more than a Zobel, which is usually connected across the driver terminals. In this instance below, there is a Zobel, but you can see that it is effectively across the amplififier terminals, and it's a total or eight components in the crossover below:
Yes, it does take computer modeling to get the right component values.
Yes, it does take computer modeling to get the right component values.
Hello Joe,
You were the one to bring up the topic and post the amplifier schematic.
What you say about transconductance (current output) amplifiers does not agree with the computer models. https://www.mathworks.com/help/physmod/sps/ref/operationaltransconductanceamplifier.html
My impression it is about equalization; passive as you show in earlier posts today or active equalization as Esa Merilainen shows in his book. Linkwitz does a lot of active equalization of his Open Baffle speakers.
It is time to see your equalization and measurements of the reference Vifa P17WO-00-08 driver.
Thanks DT
Hello All,
This is what Joe said about amplifiers in this very thread. This is the context:
“When an amplifier controls the current it cannot control the voltage. It is a current source.
When an amplifier controls the voltage it cannot control the current. It is a voltage source.
Something has to give. You can only control one of them, choose which one. It is really that simple.”
What I said:
“Saying it does not make it so.”
Current-Voltage feedback amplifiers (current amplifiers) precisely control output current into a variable impedance load by actively controlling output voltage. This is textbook stuff.
So yes, current amplifiers actively control both output voltage and output current. A voltage controlled current source.
Thanks DT
This is what Joe said about amplifiers in this very thread. This is the context:
“When an amplifier controls the current it cannot control the voltage. It is a current source.
When an amplifier controls the voltage it cannot control the current. It is a voltage source.
Something has to give. You can only control one of them, choose which one. It is really that simple.”
What I said:
“Saying it does not make it so.”
Current-Voltage feedback amplifiers (current amplifiers) precisely control output current into a variable impedance load by actively controlling output voltage. This is textbook stuff.
So yes, current amplifiers actively control both output voltage and output current. A voltage controlled current source.
Thanks DT
I think a translation of Joe's words could be written like this; To keep the current steady, the voltage has to vary with the load using a current source amplifier.
That the idea is to do it precisely in the amplifier itself by "controlling" the voltage in this process is beside the point. The voltage has to vary with the load.
That the idea is to do it precisely in the amplifier itself by "controlling" the voltage in this process is beside the point. The voltage has to vary with the load.
It's simple, you make a choice, what do you want to control, the voltage or the current? You can't have both, this is basic Ohm's Law and I am surprised that this should be controversial.
The fact that the amplifier might be a "voltage controlled current source" really has nothing to do with it. The voltage on the input, on the line level side, has nothing to do with it. That is a different voltage.
Indeed the input of a current source output can be either voltage or current. Most likely it is voltage, yet has little to do with it. So I think that DT has just misunderstood it.
Precisely.
Right. The voltage has to vary with the load when the current come from a current source. It's not rocket science.
But let us look at it differently, all amplifiers can be modeled as a voltage source and an output impedance, even current sources can be modeled this way:
If the impedance is zero, then the voltage source V has total control. The voltage is fixed by the output and not the input.
If the impedance is infinite, then the voltage source V has no control. over the volage seen by the load. In fact, it is no longer a voltage source at the load terminals, it has become a current source.
I will give a practical example: Neville Thiele, who started off the whole Thiele-Small parameters here in Sydney, Australia, used a current source to measure the impedance of [bass] drivers, in order to calculate the Fs, Qe, Qm and VAS of drivers.
Study the above carefully. This is the setup that I was told that Thiele used to measure driver parameters. The above uses a fixed 10V output and the 1000R resistor used here, if shortened to ground, would produce a fixed 10mA of current. Hence we have set up a 10mA current source.
We can now measure the impedance of the driver by putting it in series with our 10mA current source and we can now determine the impedance by reading the voltage. We read 60mV and we know it will be 6 Ohm.
The current source is stable, provided that the source impedance to load impedance is very high. Ideally it should be an infinite ratio, but in reality, in the real world, the source impedance will have be high but not infinite. In our case the ratio in the above scenario is 1000R/6R = 166:1 and that was good enough for Thiele.
But the above scenario can help us to measure the motional back-EMF impedance as well. We vary the frequency and finding the highest point and around 30 Ohm is not unsual. The accuracy does dcrease a bit since the ratio is now 33:1 abut still acceptable. We have now found the Fs of the driver. We can also measure the actual DC resistenace of the voice coil, and let us say that it is 6 Ohm DC. Now we also know the motional back-EMF impedance and that will be 30R-6R = 24 Ohm. Total impedance, 24R+6R = 30 Ohm.
It all works that way because that 10mA is a near constant and it stays that way provided the impedance you are trying to measure is many times lower than the source impedance.
But is also demonstrates that a current sources fixes the current and has no control over the voltage. The load does.
Hence the converse must also be true:
That voltage sources fixes the voltage and has no control over current. The load does.
It's really not rocket science.
So let us leave that one behind and discuss WHY the use of current EQ is good for EVERYTHING whether the amp is voltage or current.
The fact that the amplifier might be a "voltage controlled current source" really has nothing to do with it. The voltage on the input, on the line level side, has nothing to do with it. That is a different voltage.
Indeed the input of a current source output can be either voltage or current. Most likely it is voltage, yet has little to do with it. So I think that DT has just misunderstood it.
Precisely.
Right. The voltage has to vary with the load when the current come from a current source. It's not rocket science.
But let us look at it differently, all amplifiers can be modeled as a voltage source and an output impedance, even current sources can be modeled this way:
If the impedance is zero, then the voltage source V has total control. The voltage is fixed by the output and not the input.
If the impedance is infinite, then the voltage source V has no control. over the volage seen by the load. In fact, it is no longer a voltage source at the load terminals, it has become a current source.
I will give a practical example: Neville Thiele, who started off the whole Thiele-Small parameters here in Sydney, Australia, used a current source to measure the impedance of [bass] drivers, in order to calculate the Fs, Qe, Qm and VAS of drivers.
Study the above carefully. This is the setup that I was told that Thiele used to measure driver parameters. The above uses a fixed 10V output and the 1000R resistor used here, if shortened to ground, would produce a fixed 10mA of current. Hence we have set up a 10mA current source.
We can now measure the impedance of the driver by putting it in series with our 10mA current source and we can now determine the impedance by reading the voltage. We read 60mV and we know it will be 6 Ohm.
The current source is stable, provided that the source impedance to load impedance is very high. Ideally it should be an infinite ratio, but in reality, in the real world, the source impedance will have be high but not infinite. In our case the ratio in the above scenario is 1000R/6R = 166:1 and that was good enough for Thiele.
But the above scenario can help us to measure the motional back-EMF impedance as well. We vary the frequency and finding the highest point and around 30 Ohm is not unsual. The accuracy does dcrease a bit since the ratio is now 33:1 abut still acceptable. We have now found the Fs of the driver. We can also measure the actual DC resistenace of the voice coil, and let us say that it is 6 Ohm DC. Now we also know the motional back-EMF impedance and that will be 30R-6R = 24 Ohm. Total impedance, 24R+6R = 30 Ohm.
It all works that way because that 10mA is a near constant and it stays that way provided the impedance you are trying to measure is many times lower than the source impedance.
But is also demonstrates that a current sources fixes the current and has no control over the voltage. The load does.
Hence the converse must also be true:
That voltage sources fixes the voltage and has no control over current. The load does.
It's really not rocket science.
So let us leave that one behind and discuss WHY the use of current EQ is good for EVERYTHING whether the amp is voltage or current.
Hello Joe,
We are speaking of an chip amplifier, LM3875T as you show in your amplifier. The LM3875T has 2 voltage inputs. The + input connects to the line level voltage input. The - input connects to voltage feedback loop, the voltage drop across the sensing resistor.
So your amplifier, LM3875T, has 2 voltage inputs one from the line level input, the one that you claim is different and the other voltage input is voltage input from the voltage drop across the current sensing resistor resistor.
So we do have voltage input controlling voltage output that compensates for the variable impedance driver load.
This is a voltage controlled current source.
This is textbook current-voltage feedback that controls a current source.
I have marked up the feedback on your amplifier drawing.
I understand just fine.
Thanks DT
Hi DT
I am not having that discussion here. I know what a voltage-controlled current source is.
And you have clearly not understood something rather basic that I said. I am not talking about the conversion of volts to current. My discussion was about the speaker's interface with the amplifier, a totally different topic.
This forum is about speakers and not amplifiers.
So can we leave it there, please?
Joe
I am not having that discussion here. I know what a voltage-controlled current source is.
And you have clearly not understood something rather basic that I said. I am not talking about the conversion of volts to current. My discussion was about the speaker's interface with the amplifier, a totally different topic.
This forum is about speakers and not amplifiers.
So can we leave it there, please?
Joe
Hello Joe,
OkWell, like I said earlier it won't upset any amplifier, because it makes the load it sees purely resistive.
Cheers
Yes indeed. This is the discussion we should have.
But is that the only benefit? Please see post #4,947 above, can you see any other benefits?
Joe, what do you think of this page, has some interesting graphs comparing current, BL and displacement
https://jrkpost.medium.com/loudspeakers-the-secrets-of-distortion-cf24473c8be
https://jrkpost.medium.com/loudspeakers-the-secrets-of-distortion-cf24473c8be
Hello Joe,
You have known one thing and said something else all along the way. More than a little disappointing.
Current-voltage feedback is an active interface between speaker and amplifier. Not a different topic at all.
Now on to equalization and drivers.
Thanks DT
Most amplifiers are from that nether region somewhere between voltage and current source, especially with all that equalization in between source and driver.