Quite so.There are three different types of back-EMF. One is motional and creates the motional impedance we see as peaks in the box alignment, but most easy to understand when just looking at the free air Fs resonance of the driver. It is caused literally by motion in the magnetic gap. The other that you identify is inductive EMF, this is caused by the current itself passing through an inductor. In a driver, it is non-linear because at 1KHz and 10KHz the driver does not measure the same inductance, hence low inductance is very important. But the last one is perhaps the most interesting, microphonic back-EMF, because it modulates the existing back-EMF and makes the current that the Re part of the impedance sees as unstable. Like an obvious cone resonance (and a host of other things, every flaw of the driver) makes the whole back-EMF impedance wobble and now the current from the amplifier will produce undesirable current fluctuations in sympathy with those flaws. That current is fed to the speaker and it responds to that less ideal current and modifies it further, now it becomes cyclic. Esa Merilainen also refers to this mechanism in his book on current driving (but I am not in favour of current driving) and likens it to a feedback loop. The thing feeds on itself.
When speaking to Dan (Max Headroom) on the phone and describing it, he could see it and called it a merry-go-round.
Think about it for a second. EQ the current, so that the amplifier produces the same current at all frequencies, the current phase angle will sit around zero, now you have a way to make it very difficult for the amplifier to respond by creating bad current, reactive current. Our cyclic or 'feedback' event get nipped in the bud. The flaws of the driver we shall still hear, they will not be removed, but they will not cause a secondary issues either. The mechanical flaws of the driver may well prove to be less irritating to the ear that electronic problems and distortions of the amplifier.
Finally, since most people use Class AB amplifiers, and when switching transition between A and B, that voltage and current are not in sync, does that not spell trouble? I am not the only one saying this, here is a quote:
"When the current through the load is not zero because of a phase shift between the two, one of the output transistors needs to deliver this current. But this is not possible unless an error voltage at the output can open one of the output transistors via the feedback loop! So distortion is introduced and this phenomenon tends to enhance cross-over distortion which is well-known for its highly annoying properties because it has the wide spectral distribution of harmonics as shown... A class-A amplifier is far less sensitive to this phenomenon (simply because there is always current flowing through the power transistors)..." Hans van Maanen from Temporal Coherence (Temporal Coherence - Natuurlijk geluid dat je raakt... - Home
See his schematic representation as attachment below.
It isn't phase shift perse' that causes the problems. This isn't about the ear being sensitive to such things. Just that is a separate issue. It has taken me a long time to think 'current first' while not ignoring the voltage. We are talking about current devices, that means that the voltage across the speaker terminals is not what a driver responds to, it is the current. If the voltage does not change and the current doubles, then the dB-SPL goes up by 3dB. That's it!
Attachments
Last edited:
I guess that I must be a "minority hooligan", because I see this whole discussion as inane.
Well Earl I'm a major hooligan probably on several ignore lists here, these discussions have been inane for years if not down right pathological.
Have I done measurements, yes. But I am not sure of how to answer the question re 'impact' because I don't know what you mean.
Again I have to say that this is not about the flat impedance as such. That is just the obvious thing that stands out. This is about what the current is doing, and what the voltage of an amplifier does and what the current of the amplifier does, are two very different things. What a flat impedance does is make the amplifier produce the same current at all audio frequencies. The current phase angle goes flat, near zero degrees. In the same way that Neville Thiele used 'constant current' to measure the impedance of a driver. He used a 1K series resistor. That source impedance means zero current phase angle no matter how 'reactive' the load is. Thiele measured the pure impedance with no reactive current coming from the amplifier. This suits those into current drive very well. But use a voltage source, EQ the current flat at all frequencies, then you have also stopped the amplifier from producing reactive current into reactive loads.
And if you can hear it, and that it is obvious, then if you are onto a good thing, do it. But why does stopping the amp from producing reactive current make for an obviously heard better sound? Because the amplifier is no longer dancing to the tune of what the speaker does in a 'reactive' way. Voltage across the driver terminals do not make sound. The current through the voice coil does, and it has to get to the Re part of the impedance. In series with that you have the back-EMF impedance. Re is not 'reactive' and the back-EMF is highly reactive to EVERYTHING. If there is a V/I conversion taking place, it is across the Re. I have measured the Re and the voltage there, both measured and modelled it, and the voltage and current across Re is what gets translated to the dB-SPL, measured with a real physical microphone 1 metre away. Then look at the change in current and the dB-SPL changes accorsingly. Not the voltage, the current. The voltage is coincidental and the one that matters is across the Re and not the driver terminals. Neither the physical measurements, nor the modelling, are lying. They tell the same story.
BUT...
Something else happens when you make the current the same at all frequencies. The source impedance of the amplifier largely disappears. No, it does go away, but the FR deviations goes away. The box alignment no longer needs to see a zero Ohm impedance. A sealed Butterworth alignment Q=0.707 stays the same. The crossovers also gets locked in, they too no longer needs to see a zero Ohm impedance looking back. A pure voltage source is now no longer a must. Why were we not told that?
And yes, I have both measurements and modelling in agreement.
So if you have doubts about other matters, then the very fact that we have a practical means to eliminate the need for ultra-low impedances, is that at least good to know? One day, you might want to do just that.
And what does that say about so-called 'Damping Factor' that we hear about. Makes it kinda obsolete, doesn't it? BTW, the amplifier can only see a single impedance at any frequency, so if the output impedance is 0.1 Ohm and the speaker is 8 Ohm, all the speaker sees is 8.1 Ohm because they are all in series. That means the idea was never on in the first place.
Sure. But EQ the current here as well. But if we do that to small bandwidth drivers, there is going to be a lot of wasted heat I suspect. Even the tweeter has to produce the same current down to many octaves below its crossover frequency - not sure if I would want to do that. BUT it has been done, by Temporal Coherence in the Netherlands, this is what they do. But I think that for most of us, the conventional 2-Way and 2,5-Way are the most applicable in a practical way (as that is what I have done), and 3-Way could be a challenge, haven't done that yet. It is all about producing an impedance that can be easily EQ'd and I have developed such a crossover that makes it relatively easy. It keeps the impedance as reasonable high as possible and a predictable look. Now add the EQ components before the crossover, but seen directly by the amplifier.
I like the results and so are others, that is what it is all about.
Cheers, Joe
Maybe just a little bit too hasty?
BTW, I am a huge fan of waveguides - took me a little time too.
When you qualify this statement (and you need to because forcing constant current at the source has a contrary effect to this), then what remains is simply that conjugating your load prevents response variations from a finite Zo.
Besides, the load (into which the current is constant) includes the conjugate..?
Not saying this is a bad thing..
No to the first and yes to the second part - and there are no serious negative affects at all, quite to the contrary. Not if you do it right. Even you could do it.
The greater output of a current source is because the back-EMF impedance cannot reduce the current.The output in dB-SPL is directly proportional to that current, the voltage is then is purely coincidental. It is caused by the back-EMF impedance in series with the Re of the driver.
The 'trick' works with both voltage source amps and current source amps, because we have cancelled out the 'effect' of the output impedance. I works the same in both cases. This is not just theory, I am doing it, I can demonstrate it to anybody who comes through my front door.
For years, Adam Hutnik, the former owner of Pymble Hi-Fi, would preach the gospel of high damping factor. He came over and I put my Zo 270 Ohm transconductance amplifier on the Elsinore Mk6 speakers and played Joe Morello's Take Five drum solo at full concert level. It was explosive to listen to and can do that same demo for anybody, even today. After I played it, asked he he heard any lack of damping. No! Then I asked "If the output impedance is 270 Ohm, where is the damping factor?" And it finally dawned on him. "How have they then being getting away with it?"
He never again has brought up the topic. The best bass I have ever heard was from an amplifier with 5 Ohm Zo. Astonishingly clear bass, musical, powerful, no overhang, seems to go deeper but not due to any FR improvement.
The damping of the bass in any speaker is down to the alignment and nothing else. If the Zo changes the alignment and you hear that, it is the alignment that caused the change you heard. The best 'teacher' I had who taught me that was Richard H. Small in 1975. A amp with 'poor' damping factor can have GREAT bass, just keep an eye on the alignment being blown out. I am pretty sure Lynn here will agree with me.
What I could have pointed out as well, I could have asked Adam "does it sound like the crossover has gone out of wack with a 270 Ohm source impedance?" Because it too was hanging in there even though it too was not seeing a near zero source impedance.
Make the amplifier produce the same current at all frequencies and both the box alignment and the crossover will love it.
It just simply works a treat. The only other question:
Does it also improve the sound? Yes, it does, try it. A win-win situation.
Cheers, Joe
Beyond the Ariel
Joe,
What does the frequency and phase response look like when you do a system measurement? Would it be linear phase instead of minimum phase?
Currently I use digital processing to create linear phase. This effectively does change the voltage phase and current phase in the process.
Joe,
What does the frequency and phase response look like when you do a system measurement? Would it be linear phase instead of minimum phase?
Currently I use digital processing to create linear phase. This effectively does change the voltage phase and current phase in the process.
Last edited:
Good question. But imagine that you had a driver where you only saw the Re f the voice coil, it would look like a resistor to the amplifier. It would look like 6 Ohm and flat with frequency. The current phase angle would not change, the Re would see fairly 'clean' current. But when you got a back-EMF impedance, that would mean, pick a frequency like 1KHz, that would sit above 6R. It might be 10R. Now you know that the back-EMF impedance is 4R. But the current phase angle will now also be changed by the FR. Then I would expect the current phase that is shifted and seen by the Re part of the impedance, that is what we would hear. It is the current that makes the sound.
So if the voltage is a 1KHz square wave and you don't see that same sqaure wave across the Re, guess what you end up listening to? And going active will not fix that. EQ the current will. Well, sort off. It gets a bit complicated, but again I have actual measurements done. The conjugate will straighten up the current, but it sits in parallel, so the back-EMF phase shift cannot be avoided, only the amplifier current phase shift can. But in an active speaker, Class AB will benefit the most, but other benefits is that the back-EMF cannot destabilise the current, like resonances in the driver etc can. This is a good fix, not a perfect one. And you can hear it.
I know, this stuff is difficult to get one's head around, it was for me and took a long time to figure it out. But nothing new here, this has been happening all along, just didn't look?
The voltage tells us what it should be doing, the current tells us what is actually happening. The voltage is actually correct, it is our indicator, but if the current deviates, then that is what he hear - put a microphone in front of the speaker and you measure what the current has done. So will any change in current show up there? You bet!
Cheers, Joe
So if the voltage is a 1KHz square wave and you don't see that same sqaure wave across the Re, guess what you end up listening to? And going active will not fix that. EQ the current will. Well, sort off. It gets a bit complicated, but again I have actual measurements done. The conjugate will straighten up the current, but it sits in parallel, so the back-EMF phase shift cannot be avoided, only the amplifier current phase shift can. But in an active speaker, Class AB will benefit the most, but other benefits is that the back-EMF cannot destabilise the current, like resonances in the driver etc can. This is a good fix, not a perfect one. And you can hear it.
I know, this stuff is difficult to get one's head around, it was for me and took a long time to figure it out. But nothing new here, this has been happening all along, just didn't look?
The voltage tells us what it should be doing, the current tells us what is actually happening. The voltage is actually correct, it is our indicator, but if the current deviates, then that is what he hear - put a microphone in front of the speaker and you measure what the current has done. So will any change in current show up there? You bet!
Cheers, Joe
Last edited:
I was responding to what Joe had said.
I have worked on many aspects of the analog reproduction section. In each case, I need to look at some sort of measurement to make sure I am on the right track. Up to now, I have not been able to do that once between two interconnects I designed, and I do intend to get to the bottom of it some day.
But for loudspeakers and amplifiers, I have been able to do so up to now. So I am curious to see what measurements Joe can show in an end to end measurement. I do understand how important it is to control current. Even in the layout of an amplifier. But lots of things are involved whereas Joe is not describing it in an organized and understandable way.
I have worked on many aspects of the analog reproduction section. In each case, I need to look at some sort of measurement to make sure I am on the right track. Up to now, I have not been able to do that once between two interconnects I designed, and I do intend to get to the bottom of it some day.
But for loudspeakers and amplifiers, I have been able to do so up to now. So I am curious to see what measurements Joe can show in an end to end measurement. I do understand how important it is to control current. Even in the layout of an amplifier. But lots of things are involved whereas Joe is not describing it in an organized and understandable way.
OK Joe, let me see if I understand this.
You have an amplifier with a high output impedance, i.e. a constant current source, and you put a loudspeaker on this amp and it "sounds better". Is that correct? Or do you also have to do something with the speaker? I'm not clear on that part. And if so, What?
You have an amplifier with a high output impedance, i.e. a constant current source, and you put a loudspeaker on this amp and it "sounds better". Is that correct? Or do you also have to do something with the speaker? I'm not clear on that part. And if so, What?
That is fair enough, but right now I am travelling in Europe. And to collate it into a post here with all the necessary explanations, that is not so simple. I have written a discussion paper and getting comments from people I know can be trusted and so that it is tested properly. It includes a large number of actual physical measurements and computer modelling, but if you say it has to be anchored in actual physical measurements, absolutely, that is a must.
The centre piece is an "equivalence" test, based on actual test with a microphone and confirmed physical measurement that can be repeated. It can also be taken into SoundEasy and further modelled there. All I can say, I am confident it hangs together and already have on physicist who has passed it.
But I am just being a little careful and not a bit wary, based on what can happen on social media and diyaudio.com has not entirely proved itself immune to that. What I can say is this, that the "equivalence" test has been passed by one physicist and being examined by another. One is also a speaker designer of international reputation and the other with a great interest in speakers, but not directly part of his work. He has already written a paper that describes how the output impedance of an amplifier affects the loudspeaker and describes it in terms of a voltage-divider. The "equivalence" test is more about the current. So what reveals more? Looking at things from a voltage perspective (as is almost universally done, but a generalisation I know) or will a current perspective reveal additional useful things? I believe it will.
So patience is a virtue. Stay with me and that test will come your way too, and you will be as free to comment as anybody. But right now, I am testing the waters first.
Cheers, Joe
Last edited:
Hi Joe,
may be useful
https://www.edn.com/design/consumer...periority-of-current-drive-over-voltage-drive
regards
ivica
Hi Ivica
That link sounds to me to be identical to Joe's claims - or maybe I just don't get it.
It is true that a constant current amp does lower some kinds of nonlinearity, but it creates a real problem with crossovers and LF tunings. One has to know what the impedance of the source is to design a good crossover. Hence constant current could be workable is the amp and speaker are sold together, but by the two as seperates is pretty much out of the question.
And then there is the age old problem of what distortion does one actually hear in a loudspeaker. If it is a well designed loudspeaker then it's not any that constant current will lower. So it seems to me that this is a lot of talk about something that is 1) hard to implement and 2) doesn't make a difference anyways.
KEF made speakers with embedded constant current amps decades ago. It didn't go anywhere and was soon abandoned. So it's not new, just impractical and ineffective.
Makes for good copy though. One can always get a piece out of digging up this long dead horse.
That link sounds to me to be identical to Joe's claims - or maybe I just don't get it.
It is true that a constant current amp does lower some kinds of nonlinearity, but it creates a real problem with crossovers and LF tunings. One has to know what the impedance of the source is to design a good crossover. Hence constant current could be workable is the amp and speaker are sold together, but by the two as seperates is pretty much out of the question.
And then there is the age old problem of what distortion does one actually hear in a loudspeaker. If it is a well designed loudspeaker then it's not any that constant current will lower. So it seems to me that this is a lot of talk about something that is 1) hard to implement and 2) doesn't make a difference anyways.
KEF made speakers with embedded constant current amps decades ago. It didn't go anywhere and was soon abandoned. So it's not new, just impractical and ineffective.
Makes for good copy though. One can always get a piece out of digging up this long dead horse.
Hi Ivaca, yes and I have the book too. But I am not a proponent of current drive, I am in favour of putting the voltage amp into 'constant current' mode and when you do that, you can use any drive, even current drive. This is because the output impedance virtually gets cancelled out. It also, like current driving, makes it difficult for the amplifier produce current that deviates from zero degrees and hence suppressed reactive current. The beauty of it is that it is compatible with every amplifier. Cheers, Joe
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.