I'm having trouble wrapping my head around the idea that a back-EMF is a voltage source.
It is all in the mindset of the model. You can think of a driver as a motor in one moment and as an out of phase generator the next.
You can also think of Back EMF as an impedance to current flow.
^^Look for Faraday's law about electromagnetic induction. Its basic principle how such speaker driver works at all, how electric motors and generators work.
Last edited:
There is nothing basic about it.
https://purifi-audio.com/wp-content/uploads/2019/12/PURIFI-AES-9607.pdf
I recommend checking out the Faraday's law, couldn't be more basic. Its basis for such loudspeaker driver, and basis for the stuff in the paper you linked to. This thread is a bubble.
If it is basic, please explain blocked (non-motional) impedance of the coil and the impedance caused by the motion of the voice coil, a.k.a. the back-EMF.
This is basic Faraday's law kind of stuff.
When it is blocked it has an impedance as any other coil on your desk. In motion, its a microphone and sends a signal as the conductor moves in a magnetic field - aka B-EMF.
QED.
//
QED.
//
Yeah this is silly, alanhuth wonders how come backEMF is voltage and I answer Faraday's law explains it. Why I'm now obliged to explain Faraday's law is basic? Its fundamental stuff, electromagnetism fundamentals taught in college, if you haven't seen it you can find the info from the web. The paper builds on electromagnetism, builds on the Faraday's law among others.
I'm no scientist and the paper is too much to suck in at this moment but from other sources the EMF voltage is defined as Em = BlV, where Bl is the force factor, magnetic flux density times wire length in the magnetic field and V is voice coil velocity. Impedance is defined by voltage divided by current, also the ones you ask for, impedance due to backEMF voltage is defined by Zm=Em / I.
What makes the impedance vary is you ask? The Em varies because of the motion, because B and l in Em=BlV vary. I think this is what the paper is demonstrating, force factor modulation. Now, if you had constant current in the circuit the Zm would vary with the Em, with the force factor modulating and you'd get varying Zm, varying impedance due to varying backEMF voltage. To dig deeper on this you could read the references on the linked paper. I will some day, I have holes in the knowledge what actually actually happens.
Here is how I intuitively process it without knowing all the small details that go under it: think the backEMF as voltage source like it is, which makes current in the circuit over the impedance that is its load. Current affects the magnetic flux and all that, eddy currents, what ever, changing the parameters as it plays out. Basically one applies force to move the voice coil and the voice coil resists as per Faraday's law in simplest, all kinds of side-effects and non-linear parameters affecting modulating the outcome. The key point is that all of it happens simultaneously and what manifests to acoustic domain is the current in the voice coil, what the force moving the cone is including the modulating force factor. Current in the circuit varies, acoustic output varies, force factor varies, acoustic output varies.
What we can do about it, what it all means? We'd like to apply force to the cone to move accelerate exactly like the input waveform would indicate to get distortion free acoustic output. All this driver related parameters varying around with the motion is distortion to the acoustic output. Two things come to the force accelerating the cone, force factor and current. Raise the circuit impedance outside the driver to dilute effects of the voice coil movement to the circuit impedance, when the impedance is stable the current is stable. Use better motor to reduce the force factor modulation (and effects to current). These are the key elements in the electric domain.
Its very complicated stuff and I think there is no need to understand all of it. I suggest everyone to get some sort of simplified intuition to the subject to be able to use the knowledge as tool for better sounding systems. I feel having it, I'm analyzing the circuit impedance from drivers perspective and abstracting any power amplifier away by replacing it with its output impedance, like Thevenin's theorem, college level circuit analysis.
I don't have to exactly know what makes into it (the impedance), or how big or bad is it, all I need to know how and where to increase or decrease impedance in the circuit and why. I'm thinking only current generated by the driver voltage source(s) to simplify the situation. Increase circuit impedance to reduce distortion current (generated by the driver backEMF), decrease to increase current at driver main resonance if needed for electronic damping (generated by the driver backEMF). Use best driver you can afford with least distortion to begin with. Oversize the system to reduce motion, reduce current, reduce problems.
I'm no scientist and the paper is too much to suck in at this moment but from other sources the EMF voltage is defined as Em = BlV, where Bl is the force factor, magnetic flux density times wire length in the magnetic field and V is voice coil velocity. Impedance is defined by voltage divided by current, also the ones you ask for, impedance due to backEMF voltage is defined by Zm=Em / I.
What makes the impedance vary is you ask? The Em varies because of the motion, because B and l in Em=BlV vary. I think this is what the paper is demonstrating, force factor modulation. Now, if you had constant current in the circuit the Zm would vary with the Em, with the force factor modulating and you'd get varying Zm, varying impedance due to varying backEMF voltage. To dig deeper on this you could read the references on the linked paper. I will some day, I have holes in the knowledge what actually actually happens.
Here is how I intuitively process it without knowing all the small details that go under it: think the backEMF as voltage source like it is, which makes current in the circuit over the impedance that is its load. Current affects the magnetic flux and all that, eddy currents, what ever, changing the parameters as it plays out. Basically one applies force to move the voice coil and the voice coil resists as per Faraday's law in simplest, all kinds of side-effects and non-linear parameters affecting modulating the outcome. The key point is that all of it happens simultaneously and what manifests to acoustic domain is the current in the voice coil, what the force moving the cone is including the modulating force factor. Current in the circuit varies, acoustic output varies, force factor varies, acoustic output varies.
What we can do about it, what it all means? We'd like to apply force to the cone to move accelerate exactly like the input waveform would indicate to get distortion free acoustic output. All this driver related parameters varying around with the motion is distortion to the acoustic output. Two things come to the force accelerating the cone, force factor and current. Raise the circuit impedance outside the driver to dilute effects of the voice coil movement to the circuit impedance, when the impedance is stable the current is stable. Use better motor to reduce the force factor modulation (and effects to current). These are the key elements in the electric domain.
Its very complicated stuff and I think there is no need to understand all of it. I suggest everyone to get some sort of simplified intuition to the subject to be able to use the knowledge as tool for better sounding systems. I feel having it, I'm analyzing the circuit impedance from drivers perspective and abstracting any power amplifier away by replacing it with its output impedance, like Thevenin's theorem, college level circuit analysis.
I don't have to exactly know what makes into it (the impedance), or how big or bad is it, all I need to know how and where to increase or decrease impedance in the circuit and why. I'm thinking only current generated by the driver voltage source(s) to simplify the situation. Increase circuit impedance to reduce distortion current (generated by the driver backEMF), decrease to increase current at driver main resonance if needed for electronic damping (generated by the driver backEMF). Use best driver you can afford with least distortion to begin with. Oversize the system to reduce motion, reduce current, reduce problems.
Last edited:
Back EMF also occurs with regular fixed inductors in response to normal signal related flux variations.
Also mechanical parameters vary with excursion, the suspension system, cone area, the box, all of it changes with heat and so on, the room, the mood, everything is far from ideal like everything else in real world, in life. But still we manage to come along and many systems sound pretty nice 🙂 All we have to do is try to make best of it, try to get rid of the worst offenders, "problem free" to a target. Gotta remember many issues arise or magnify with increasing output level, for this reason the simplest way to get better sound is to have big enough system which pushes some audible issues to loud levels so that they are inaudible in typical listening situation. This is completely opposite to modern trend of smaller and smaller speakers.
Last edited:
Hello All,
I have been away from Sparkey’s Lab for several days.
It turns out that the 2016 Esa paper was here in the file for some time. Originally I had the impression that current distortion was measured across a current sensing resistor, I was mistaken. In the 2016 paper Esa measured distortion with a microphone placed on center, 20 CM or 8-7/8 inches away from the mounting plane of the driver. That is what I will also do.
The Classic P17WJ-00-08 driver is installed in a Denovo 0.55 cubic foot sealed enclosure with a fist full of Dacron pillow stuffing.
First off we will look at Frequency Response and Harmonic Distortion. Later add Two – Tone IMD tests.
Thanks DT
This would be problematic to distortion measurements as on-axis breakup peaks could exaggerate distortion numbers even for frequencies below breakup.
The on-axis region is only an insignificant portion of the driver output power. Output can vary significantly at different angles especially through breakup and also due to beaming/lobing.
Here (in red) is the on-axis region shown on a representation of a hemisphere..
Full wording is "back-electromotive-force. A force in electrical terminology is a voltage.
Absolutely!
Hence it is important to distinguish the difference between internal and external inductance. Keep internal inductance as low as possible and add some series inductance as a good thing. Indeed if you are able to achieve 5:1 ratio, then you effectively can have the benefits of current-drive whilst using a conventional voltage source.
Correct. A Sennheiser E815s dynamic microphone, quite inexpensive. I believe the E835s is the replacement, these mic can be had for around $100. This is a vocal mic and able to handle high SPL as some vocal performaners seems to almost swallow the mic.
Back-EMF is the key, it is not a stable impedance and cause impedance modulations hence current modulations of the amplifier.
Yes, varying impedance due to varying beck-EMF voltage, beautifully said.
People like Esa, and myself, has pointed to what he calls a "feedback"
Here is the full convoluted sentence from Esa's book "Current-Driving Loudspeakers" from page 57, it is the singular most important paragrapgh in the entire book, and I ask, does people understand what he is saying:
"A loudspeaker circuit operating on voltage drive exhibits a feedback effect where the EMF deriving from voice coil motion summates directly with the voltage applied to the driver, so that the resulting current is a mixture of the original signal and a spurious signal corrupted by the speaker's own mechanical, electrical, acoustic properties and circulated in the feedback process."
Lot to absorb in there as he has thrown everything at it, including the kitchen sink Please understand, the "feedback" he is referring to is really "feed backing" rather than how we use the feedback word.
So much of this comes down to language and how words are interpreted. What is clear is that he is talking about a mechanism by which current get corrupted (distorted) and not voltage distortion. Yet if you use the word "distortion" the mind to defaulting to voltage distortion. This is about current distortion and current is directly related to...
And it is the back-EMF that get impacted by a series of things as Esa says:
"... so that the resulting current is a mixture of the original signal and a spurious signal corrupted by the speaker's own mechanical, electrical, acoustic properties and circulated in the feedback process."
Note the word "circulated."
Yes, a looping backwards and forwards, circulating and smearing things that produces products that we know should not be there (they seem to be mostly odd order but there might also be other stuff), and they show up in Esa's acoustic measurement, but they are caused on the electrical side and then comes through on the acoustic side.
We know that hysteresis is a big problem. But the "feedback" thingy Esa talks about, circulating and a smearing in time, whilst hysteresis a magnetic property (of a magnetic field), here we have something on the electrical side that is not all that dissimilar?
That last bit, just thinking out loud...
Last edited:
It was what I was. You can reduce back-EMF issues (impedance modulations leading to current modulations) by having a larger and more stable external inductance. That is what I did in the Elsinore designs from Mk5 onward. The reduction in distortion (current) is then similar to what current-drive does. And you can definitely hear it.
This is the Elsinore thread and maybe should make a link between the discussion and the Elsinore design.
The earlier designs had additional corrections (LCR across the speaker terminals) in addition with the series inductor. Later, and it was a leap of faith, I tried to find drivers that allowed me to maximise the inductance as high as I could and avoid (leap of faith) to add any additional corrections. Yes, it was a challenge, but the high-pass that I developed (and has now been copied by other designers) was flexible enough to still get a reasonable result, particularly when the Elsinores are 15° off axis. Resist the temptation to throw as the baby with the bathwater. I know that other designers are not doing this... they are basically afraid of it.
Lars, if you are reading this, arguably the behaviour above 1KHz to near 5KHz made my job easier - the consistency between 0°, 15° and 30° is the best I have seen and I was thus able to used an increased series inductor value there. There is something you did to that cone, on top of the magnetics and the 'funky' surround, the way that I believe you distributed the mass has a lot to do with it. Just my guess.
Here is an example where 5:1 inductance would give you the same advantage as current-drive but without any of the downsides:
INDUCTOR-DRIVE - ALTERNATIVE TO CURRENT-DRIVE - JUST FIND THE RIGHT DRIVER:
Yes, the cone and thus its coil is trying to impress (feedback? not a good word here really) a voltage towards the driving amplifier as it moves thru the magnetic filed. But the amplifier has been given the task to also impress a voltage over the coil in the driver. It becomes a superimposing addition affair and at the same time a tug of war where the one with the least internal impedance (output impedance) will win. It is obvious as one know that a low output impedance amplifiers are not effected by the speaker amplifier (they win the tug) in the same way as e.g, a high output impedance amplifier (e.g. a typical tube amp or indeed a current drive amp) where the high output amplifier will have a distorted FR which follows the drivers impedance response. So there is something real in the concept of DF - damping factor. Its not mystic at all. Just simple physics and a bit of electricity... 😎 but I imagine the summation point in this inferno to be a quite interesting place of voltage, current, magnetics, phase and hysteresis... This place is of course in the coil.
The insight is naturally that the coil must be gone.
//
The insight is naturally that the coil must be gone.
//
Last edited: