Hello Joe and All,
Joe you asked our thoughts about odd harmonics and time smear. For odd harmonics the top most likely and largest culprit is magnetic hysteresis. More voltage is consumed creating flux than is generated when flux decays. Is a lossy process, the generated back EMF waveform is asymmetrical. As a result Bl modulates. That is the stuff of much of the odd harmonic series of distortion. In terms of time smear, it does take time for voltage through a coil to build up flux and time for that flux to generate Back EMF. It is a feedback kind of thing, each time around the loop it takes time and the magnitude decreases. Time smear is not that technical of a term, how is it measured and what are the units. Typically when you see time quantified you see Degrees, Radians, seconds or some related unit of measure. You know, some kind of 2 Pi Radians type of factor.
About thought experiments:
Faulty thought experiments lead to faulty assumptions.
Faulty assumptions lead to faulty thought experiments.
This is an example of a faulty assumption:
“It is that simple. You cannot divorce the current from the force BLi - to be blunt, as far as the driver is concerned, the voltage doesn't even enter into the 'equation' and yes, there are equations.”
This is the equation from above F = i * Bl
This equation is absent from the assumption above i = (U - EMF) / Re . (This is Ohms Law.)
U and EMF are both voltages.
It is more complex than a “KISS” on the cheek will get you.
In my opinion the result of a “Thought Experiment” is a hypothesis that requires testing.
Thanks DT
Joe you asked our thoughts about odd harmonics and time smear. For odd harmonics the top most likely and largest culprit is magnetic hysteresis. More voltage is consumed creating flux than is generated when flux decays. Is a lossy process, the generated back EMF waveform is asymmetrical. As a result Bl modulates. That is the stuff of much of the odd harmonic series of distortion. In terms of time smear, it does take time for voltage through a coil to build up flux and time for that flux to generate Back EMF. It is a feedback kind of thing, each time around the loop it takes time and the magnitude decreases. Time smear is not that technical of a term, how is it measured and what are the units. Typically when you see time quantified you see Degrees, Radians, seconds or some related unit of measure. You know, some kind of 2 Pi Radians type of factor.
About thought experiments:
Faulty thought experiments lead to faulty assumptions.
Faulty assumptions lead to faulty thought experiments.
This is an example of a faulty assumption:
“It is that simple. You cannot divorce the current from the force BLi - to be blunt, as far as the driver is concerned, the voltage doesn't even enter into the 'equation' and yes, there are equations.”
This is the equation from above F = i * Bl
This equation is absent from the assumption above i = (U - EMF) / Re . (This is Ohms Law.)
U and EMF are both voltages.
It is more complex than a “KISS” on the cheek will get you.
In my opinion the result of a “Thought Experiment” is a hypothesis that requires testing.
Thanks DT
It would be nice if it would fit in wavegides wich are made for standart 1" compression Drivers. There would be a lot of horns and waveguids out there wich could be used. And for some who already have speakers with 1" drivers it would be easy to give the new Drivers a try.
Besides that I would opt for some intrgral waveguids wich can house one ore two 6.5" Drivers like this one https://www.limmerhorns.de/630-bc1/
I did read the posts. But they are not focussed on the real issue.
This is definitely about back-EMF-related distortion.
That's exactly what I have been on about. I am despondent that is not obvious.
There are certain concepts that clearly are not getting across. Down here in Oz some of us are having this discussion. We are putting the pieces together and then figuring out what we need to fix and what are red herrings to avoid. Outcomes is what matters!
Sorry, but that statement cannot be entirely correct.
The driver is the load, it does not generate power. It does not generate distortion. It is not unrelated to distortion, however. The load triggers current and back-EMF interferes with the nature of that current, and that is what causes distortion.
But not voltage distortion. The voltage of the amplifier has not changed. But the current has.
It is current distortion.
Where does the current come from?
Answer: The amplifier.
OTOH: If the amplifier is a current source, the back-EMF of the load just produces a voltage only that cannot interfere with the current. Due to the amplifier having a series impedance of infinity, the current cannot be interfered with because the current is fixed. No current distortion is the result. At least in theory.
The driver will still produce distortion under current drive. Current drive does not magically stop that. Only that additional distortion due to the current corruption is being prevented.
If only we had perfect drivers. No driver distortion. No back-EMF. Then no back-EMF induced current distortion.
One problem triggers another.
There you have it in a few sentences the difference between voltage and current drive. The advantage of current drive is simple, it prevents the driver from inducing more distortion triggered by its own imperfection. It is not exactly rocket science.
So back-EMF related, it really comes down to a single question: If we are to use voltage source amplifiers, what can we do to reduce back-EMF induced current distortion?
The answer to that question is what drives me.
Please help me, am I getting across?
You have entirely missed the point, so I am not even going to try.
The KISS principle isn't about reducing the number of complexities, it is a pragmatic approach to enable a person to focus on what matters to get a practical outcome.
I would have thought that was obvious.
Hi,
yes you are getting across, I understand what you say but the message you have is confusing to me.
Lets make thought experiment from the quote above: back-EMF is voltage source, right? The same as the power amplifier we are having, a voltage source, that makes two voltage sources in the same circuitry. Now, the power amplifier makes current into the circuit over its load impedance as you say, over impedance that is there for it in the circuit, right? What the other voltage source does? Of course the same, makes current over its load impedance, why would it be any different? Different physics? Its all in the same circuit, happening at the same time. Electromagnetism.
Moving coil in magnetic field is current generator, same as moving magnetic field in static coil, that is basic physics phenomenon all generators work from record player stylus to microphones to Hoover Dam. If the circuit is open its only voltage induced in the conductor, until circuit is formed which enables to extract power from the system, current flows. Why would it be any different in amplifier / driver system?
What makes this particular system we inspect here, amplifier and driver, confusing is that there is the power amplifier in the circuit making the movement for the generator and catching the attention of us inspectors. But the power amp is clear as day regards to distortion in comparison to what the driver does. Take a look at the system again, forget about the power amplifier and think there is somebody using a stick to make the cone move. Moving driver cone will generate current in the circuit as long as its a closed circuit.
Check out Thevenin and Norton equivalents to analyze the circuit for each source / load.
Now, replace the voltage power amplifier in the circuit with a short or very small resistor and analyze whats the load impedance for error current mechanism, load for back-EMF voltage source? What if there was current drive amplifier instead, replace it with a big resistor or open circuit, what is the load impedance for driver now? What if it was voltage amplifier and an inductor in series? What about shunt components, any effect?
I get what you are saying that amplifier operation is severely disturbed by the back-EMF, and it might be so, which makes driver distort. By severely I mean why we measure ~1% distortion with microphone when the amplifier has 0,001% of it, what makes the rest? Its because of the driver as we both say. Its clear how to reduce driver distortion current just by increasing load for it, and you can measure that. Driver distortion current fixed, amplifier distortion fixed, right? I think you are meaning exactly same thing with your explanation about the system, its just that the explanation seems far fetched. Its much simpler to just look the system driver as generator and come to same conclusion, and this is easy to inspect and fix just by analyzing the system driver as generator. To answer your question what we can do is we can increase driver load impedance to reduce distortion. Explanation why that happened is irrelevant in the end as long as it gives power for us designers to keep distortion at bay.
Last edited:
I assume there is lot of readers and I see lots of confusion in posts like I had as well, also not many is coming forward and say they would understand what you try to say with current distortion. I try not to underestimate anyone and also try be fearless about making discussion on the topic even though I'm very much apprentice here. It wasn't until checking out the system from driver perspective that made all things clear to me, how the distortion mechanism delivers into acoustic domain. All fairy dust and unknowns gone, simple system with simple explanation, simple inputs and outputs, and it seems to hold water so kind of trying to provide an alternative way of looking at it for everyone. Peace, I'm out !
No, you are not getting across, you are still barking up the wrong tree. Back EMF is not some weird monster that crawls back through your speaker wires to attack your amplifier. Please stop putting these unfounded theories into people's heads.
The simple truth about back EMF is that it is the voltage that results from the movement of the voice coil through the magnetic field of the driver. This voltage is opposite to the voltage delivered by the amplifier that put the voice coil in motion in the first place. The effect of this is that the amplifier sees an increase in resistance from the driver (impedance would be a more correct term) around resonance. From the amplifier side, this is great. Almost all amplifiers produce less distortion with lighter loads.
And so it is with EMF. At speaker resonance, EMF increases, impedance increases, and the amplifier sees a lighter load, which is good for distortion.
It is quite easy to force this impedance peak around driver resonance down with a coil, a capacitor and a resistor, like what you do, but it does not help your amplifier. To the contrary. You will just add some extra resistance which will cause almost all amplifiers to distort more.
To the general public a word of caution. There are many instances where vendors float funky theories in order to sell their merchandise. This is one of them imho.
It's nice to engage in thought experiments and some theory, but only practical listening tests will yield the answers. Life is about practical experience, listening to music is about pleasant emotions from sound. So far, you seem to have pleasant emotions, especially when watching nice looking graphs. Joe, so do you have your theories verified by ear or not? It seems that you don't, and you only care about the beautiful straight lines in the graph of the impedance and electrical phase of your speakers and the satisfaction of the mind that lives in theoretical ideas.
Thank you for asking that question.
I remember the first time I tried it on a speaker and I got this electrical outcome:
The reaction was instantly obvious: What is happening here?
This was the first speaker I tried it on:
It was a remarkable experience. I asked a friend to come around, he has very good high-end ribbon speakers. He spent three hours listening to them and made the comment that this was the first time he had heard a dynamic loudspeaker that blind-folded he could have sworn he was listening to a ribbon speaker. Among ourselves, the late Allen Wright had his Apogee Acoustats ribbons and we listened to them all the time. We knew the sound well. So this was high praise indeed.
Even ribbons have their plus and minuses. But what they do have, within their usable passband limitations, a very resistive load.
The back-EMF current distortion is very low due to ribbons generally having much lower inductive back EMF.
I was getting a bite of the cherry using dynamic drivers. Note the waveguide. Look at that current phase angle. Flat! I instinctively realised that was the key and not just the flat impedance. So it had to do with the current.
Benchmarking? I will let other people do that. They are welcome.
Please, loving pretty straight lines? That's a rather condescending thing to say.
I don't show graphs to give me flushes of happiness... sheesh...
And I know what live music sounds like!
I have a family full of musicians on both my side and my wife's side. Amateurs and professionals. I don't think I go without a day hearing a real piano. If you can't get a system or loudspeaker to sound like a proper piano, then it's probably because it is not that good. Have you heard the intro to St James Infirmary by Hugh Laurie on piano on a system that does it justice?
I have several friends who run top-class recording studios and for whom I have made equipment. Their hearing is scary good. My middle son is a piano teacher. My oldest son Mikael plays real classic guitar even though he has a severe mental handicap (no, that does not make him dumb, just problematic) and studied for nearly ten years to be able to do that. His guitar is worth a fortune, but what a sweet sound comes out of it. Early I posted my son Beau playing a Beethoven piece and maybe I should get a recording of Mikael playing a Spanish piece called "Recuerdos de la Alhambra" by Franscisco Tarrega.
Here is somebody who has done the same piece on YouTube. One day I will record Mikael doing it.Beau's performance of the opening Allegro movement (featuring a so-called 'Mannheim Rocket' phrasing). Hope you like it. Beau is improving all the time:
https://www.customanalogue.com/Beau_Recital.mp3
The photo below is from our Music Room. Yes, we have a Music Room in our house and there is no Hi-Fi equipment in it. But it has a real Yamaha piano.
It is plainly wrong to both minimise and maximise an issue.
Based on past history, you will understand that I shall not respond to your continued baiting, so have a good day, sir!
Sorry for the above post, hope you understand.
I was already in the process of preparing the installment below, so here goes.
Others will be happy that this next installment contains plenty of intriguing curves and no straight lines. lol
May I borrow from Esa is electrical equivalent circuit of a driver. I have sneakily edited a few things in Red in order where they show up in the impedance plot of the driver.
We have all seen bog standard impedance plots and they don't show anything special, right?
Actually, that would be very wrong. Let us dissect what an impedance plot reveals.
Here RC is the DC Resistance of the voice coil. Let us say 6 Ohm.
Ei is the inductive Back-EMF.
Em is the motional back-EMF.
Es is the microphonic back-EMF.
Rc or Re is the impedance formed DCR, the stable part of the impedance.
The first three are E in the sense they are electro-motive forces.
That also makes them voltages. Electro-motive forces are voltages. Why are so many resisting that? (There is a pun in there, did you spot it?)
Let us post our example impedance plot and identify a few things:
Let us make it a bit more obvious and match up with Esa's equivalent circcuit:
Ahah! Is it starting to make sense? All four of them are right in there in our impedance plot.
Now we see where all the elements are, Ei, Em, Es and Rc.
EVERYTHING ABOVE THE BLUE LINE IS BACK-EMF.
We will make the back-EMF Purple. In fact there is more Purple than Blue. It swamps it:
Only the light Blue is resistive and everything above that is caused by back-EMF.
There is a reason that this is not called front-EMF.
And there is a reason that it is EMF, Electro-Motive Forces = Voltage Sources!
But it shows up vertically and as Ohm.
___________________
But here is something the impedance plot can only hint at.
IT IS MUCH WORSE:
Back-EMF induced distortions must take something into consideration, that the impedance of an average but good driver looks something like this:
You do not have a stable impedance. Only the Red one above is known by you. In actual fact, when music with multiple instruments are playing... you do not know the impedance is at any moment in time!
Is any of this being understood by you guys?
Excursions!
Everything vertically causes current by the amplifier. If you pick any frequency where you find the impedance to be 12 Ohm, produce a voltage of 12V at that frequency and the driver will draw 1 Ampere. The current is inversely proportional to the impedance. The impedance goes up and the current will go down, and vice versa.
Excursions!
There is your wild card.
Cause significant excursions at low frequencies and what will happen to our stable current of 1 Ampere at the frequency where we thought the impedance was 12 Ohm?
I mean, this is not exactly rocket science.
The truth is that our impedance plot is not nice and clean and has a fuzzy character. That leads to fuzzy current and that leads to more distortion. This is the distortion that current drive is supposed to prevent.
But most of us are going to be listening and staying with voltage sources.
I know the answer! We need perfect drivers without Ei, Em, and no Es would be nice.
___________________
Finally, I want to go back to why the KISS principle is so important.
Yes, you are listening to the current of the amplifier. But is that all? No, of course not.
"Bli"
i is the current.
l is the length of the voice, also in fact the gap (not the entire length of the voice coil).
B is the magnetic strength in the gap.
This is keeping it as KISS as possible. Why is that so important?
Consider this: We feed a nominal 2.83V RMS across the driver terminals and we measure the sound pressure level with a microphone at 1 Metre.
Let us say that we just happen to measure 90dB-SPL.
Now here is where KISS comes in. Whatever gave us that outcome, That is something we rightly could have endless discussions about. So many things got us to arrive at that 90 dB-SPL and that is all very fine.
But we could emerge ourselves long and hard and yet miss an important fact.
And to me, it is the most important fact.
It is not how we got 90dB-SPL at 1 Metre that interests me, it is how we get from 90dB-SPL to 96dB-SPL.
Yeah, think about it.
It's late here. I will let you guys think some. Did you get that last point? To me this is the key.
I was already in the process of preparing the installment below, so here goes.
Others will be happy that this next installment contains plenty of intriguing curves and no straight lines. lol
May I borrow from Esa is electrical equivalent circuit of a driver. I have sneakily edited a few things in Red in order where they show up in the impedance plot of the driver.
We have all seen bog standard impedance plots and they don't show anything special, right?
Actually, that would be very wrong. Let us dissect what an impedance plot reveals.
Here RC is the DC Resistance of the voice coil. Let us say 6 Ohm.
Ei is the inductive Back-EMF.
Em is the motional back-EMF.
Es is the microphonic back-EMF.
Rc or Re is the impedance formed DCR, the stable part of the impedance.
The first three are E in the sense they are electro-motive forces.
That also makes them voltages. Electro-motive forces are voltages. Why are so many resisting that? (There is a pun in there, did you spot it?)
Let us post our example impedance plot and identify a few things:
Let us make it a bit more obvious and match up with Esa's equivalent circcuit:
Ahah! Is it starting to make sense? All four of them are right in there in our impedance plot.
Now we see where all the elements are, Ei, Em, Es and Rc.
EVERYTHING ABOVE THE BLUE LINE IS BACK-EMF.
We will make the back-EMF Purple. In fact there is more Purple than Blue. It swamps it:
Only the light Blue is resistive and everything above that is caused by back-EMF.
There is a reason that this is not called front-EMF.
And there is a reason that it is EMF, Electro-Motive Forces = Voltage Sources!
But it shows up vertically and as Ohm.
___________________
But here is something the impedance plot can only hint at.
IT IS MUCH WORSE:
Back-EMF induced distortions must take something into consideration, that the impedance of an average but good driver looks something like this:
You do not have a stable impedance. Only the Red one above is known by you. In actual fact, when music with multiple instruments are playing... you do not know the impedance is at any moment in time!
Is any of this being understood by you guys?
Excursions!
Everything vertically causes current by the amplifier. If you pick any frequency where you find the impedance to be 12 Ohm, produce a voltage of 12V at that frequency and the driver will draw 1 Ampere. The current is inversely proportional to the impedance. The impedance goes up and the current will go down, and vice versa.
Excursions!
There is your wild card.
Cause significant excursions at low frequencies and what will happen to our stable current of 1 Ampere at the frequency where we thought the impedance was 12 Ohm?
I mean, this is not exactly rocket science.
The truth is that our impedance plot is not nice and clean and has a fuzzy character. That leads to fuzzy current and that leads to more distortion. This is the distortion that current drive is supposed to prevent.
But most of us are going to be listening and staying with voltage sources.
I know the answer! We need perfect drivers without Ei, Em, and no Es would be nice.
___________________
Finally, I want to go back to why the KISS principle is so important.
Yes, you are listening to the current of the amplifier. But is that all? No, of course not.
"Bli"
i is the current.
l is the length of the voice, also in fact the gap (not the entire length of the voice coil).
B is the magnetic strength in the gap.
This is keeping it as KISS as possible. Why is that so important?
Consider this: We feed a nominal 2.83V RMS across the driver terminals and we measure the sound pressure level with a microphone at 1 Metre.
Let us say that we just happen to measure 90dB-SPL.
Now here is where KISS comes in. Whatever gave us that outcome, That is something we rightly could have endless discussions about. So many things got us to arrive at that 90 dB-SPL and that is all very fine.
But we could emerge ourselves long and hard and yet miss an important fact.
And to me, it is the most important fact.
It is not how we got 90dB-SPL at 1 Metre that interests me, it is how we get from 90dB-SPL to 96dB-SPL.
Yeah, think about it.
It's late here. I will let you guys think some. Did you get that last point? To me this is the key.
Joe,
When you try to make it simple is when you confuse your readers.
In your last post you spell it out pretty well. You add up all the constituent parts that make up Back Emf. Then you show that Back EMF opposes the voltage applied by the power amplifier. Pretty simple, and understandable so far.
Then you toss out "Bli",
"Bli" is not so simple at all, it leaves out all the moving parts.
In your next breath you include 2.83V RMS. This is the point I have been making all along (and others), no volts no current. The concept of Volts is what makes the distortion effects of Back EMF understandable.
“Bli” is too simple, It is not a complete model.
This is my idea of a simple model, 2 equations:
F = i * Bl (1)
i = (U - EMF) / Re (2)
If you want more complex look up the L2 R2, Wright, or Leach models.
Hats off to Lars Risbo and his AES Paper https://purifi-audio.com/wp-content/uploads/2019/12/PURIFI-AES-9607.pdf
Thanks DT
When you try to make it simple is when you confuse your readers.
In your last post you spell it out pretty well. You add up all the constituent parts that make up Back Emf. Then you show that Back EMF opposes the voltage applied by the power amplifier. Pretty simple, and understandable so far.
Then you toss out "Bli",
"Bli" is not so simple at all, it leaves out all the moving parts.
In your next breath you include 2.83V RMS. This is the point I have been making all along (and others), no volts no current. The concept of Volts is what makes the distortion effects of Back EMF understandable.
“Bli” is too simple, It is not a complete model.
This is my idea of a simple model, 2 equations:
F = i * Bl (1)
i = (U - EMF) / Re (2)
If you want more complex look up the L2 R2, Wright, or Leach models.
Hats off to Lars Risbo and his AES Paper https://purifi-audio.com/wp-content/uploads/2019/12/PURIFI-AES-9607.pdf
Thanks DT
Last edited:
Joe - why so obsessed to describe the innards working of your product? Most of your buyers don't understand this anyways. Do you feel you need to do it to make fancy advertisement? Brag about you technical skills? And you claim the amp choice is non critical - so no need to talk currents and voltages - advertise that!
Why not talk about the sound instead? Shouldn't that suffice?
//
Why not talk about the sound instead? Shouldn't that suffice?
//
@Boden,
While Joe sleeps.
You may want to download the Purifi AES paper and search the term " position ".
The impedance of the voice coil is dependent on its' position in the magnetic gap.
a-symmetry in the suspension plays a smaller role.
Thanks DT
https://purifi-audio.com/wp-content/uploads/2019/12/PURIFI-AES-9607.pdf
The way I understand this from Joe is:
Play a low note, close to the impedance peak of the driver. The cone moves in the frequency of this lower note to create the sound, low note is a slow pace. Play a higher note simultaneously, while the lower note is playing. Now the higher note moves way faster in and out, barely moving the cone, while the lower note does move the cone.
As the lower note holds the cone in inward position, the higher note moves in and out creating it's sound... but it's position yields a different impedance than when the lower note moves the cone outward again, and the high note again swings trough it's own faster motion while the cone is largely outward.
The graph that predicts this for a driver is that Inductance graph:
So when the low note is keeping the cone at -2.5 mm position and it is simultaneously playing the higher note, Le is about 0.135 mH. Now the cone is swinging outward, and reaches +2.5 mm, while still swinging back and forth in a faster movement to create that higher note, In that position it is seeing an Le of 0.105 mH. (original graph and story here)
An IMD two tone test should show this difference, playing a low note that really moves the cone, while simultaneously playing the higher note, you should see a difference in the distortion figures between current drive and voltage drive. Unless you use one of them fancy drivers from Purify. Note that the cone really needs to move due to playing that lower note or you will not see the difference. Examples of current drive vs voltage drive here and here. Not every driver will show similar results, certainly not the Purify driver. And not all amplifiers are created equal either.
Next, Joe wanted to find a way to make a Voltage drive amplifier have a similar result. That's where the series coil and series/parallel woofers come into play. The series/parallel woofers limit excursion (less change in Le because of less movement) The coil? Well, we've read that here multiple times. Basically it has the same function as the high output impedance of the current drive amp. The impedance compensation? That does make sure that any amp's output impedance has no real life effect on the resulting frequency curve you get. Well, almost no effect, a very small difference remains at the port. Not sure yet about any other benefit of that compensation, that part is unclear to me. We do let the amplifier make about an equal level of voltage and current at each frequency this way, as it largely sees a resistive load.
Last edited:
Joe, thanks for your response. But I wanted to know what change in sound you heard when comparing the Elsinore's compensations connected and disconnected. Instead, we read about a friend who felt like he was listening to ribbons speakers when you played him some previous boxes with compensation. As I mentioned before, I did a listening test with speakers with a very ugly impedance curve, and with compensations the sound was no better. He was only slightly different and rather worse. So I didn't find a correlation between beautiful pictures with straight lines and better sound. That's why I see this whole effort of yours as a useless solution to theoretical matters.