Don't forget that the cone is a moving object. An with that movement you get varying impedance from that cone movement. Just look at some Klippel test of drivers (*). So while a cone is playing low notes, making the cone move, it may simultaneously play mid frequencies too and due to the movement of the cone, the impedance of the driver will vary. Now apply that same voltage again and something has got to be different between voltage drive and current drive. As in both cases the same laws do still apply
. There will be differences between drivers in how they react to this due to their construction. Even passive filters have an influence, as has been mentioned before in this thread. Nothing new, nothing magic. Just following the same basic rules. If all drivers had a steady impedance even with cone movement your case would make sense. In our real world the impedance does vary with cone movement.So we don't have the same impedance, right? The added impedance (in series with the coil) of the current drive will swamp the varying impedance of the driver. Distortion as measured will be reduced.
(*) Don't look at the Purify drivers, they actually worked hard to keep the impedance steady with cone movement.
I was waiting for that sensible explanation. We discussed this a few times already here. A dynamic loudspeaker may behave linear, but you have to use quite an elaborate electrical model to simulate it, mainly because of the moving cone. The standard scheme isn’t enough. As Thiele (or Small?) pointed out: it’s small-signal. Not real world.
Hi Ramista,
Read Floyd Toole and anything with the Canadian NRC. There has been work done in that direction. I'm sure there will be lot's to add in the future.
Hi tmuikku,
The added components will interfere with the driver being driven. But sure, why not?
Hi wesayso,
The correction circuit being discussed is dynamic, so it corrects moment by moment. This is not a concern. At any rate, we are past that argument entirely and are discussing the measurements aspect only. Not how you drive the speaker or use active correction..
Read Floyd Toole and anything with the Canadian NRC. There has been work done in that direction. I'm sure there will be lot's to add in the future.
Hi tmuikku,
The added components will interfere with the driver being driven. But sure, why not?
Hi wesayso,
The correction circuit being discussed is dynamic, so it corrects moment by moment. This is not a concern. At any rate, we are past that argument entirely and are discussing the measurements aspect only. Not how you drive the speaker or use active correction..
I do not respond well to errant accusations that expressions of my knowledge are "baseless", and especially not to orders to "drop" anything from an accuser. This a public forum and any incorrect statements of scientific fact or misconceptions should be corrected. I am more than happy for you to correct me if you can find a single error in my posts here. So far you have failed to do so.
For all contributors, I will now respond with information that is highly relevant to the thread subject, and explains why certain loudspeaker distortion measures require further consideration if we are to relate them to subjective assessments. I refer everyone interested in understanding the subject to my original post (#24) before reading the added details here that relate specifically to distortion measurements.
WHY AMPLIFIER OUTPUT IMPEDANCE IS IMPORTANT
A moving-coil driver has two main sources of non-linear distortion that should not be confused. (There are other distortion sources too, just they tend not to be as significant). Of particular relevance here is how one of these two main distortion mechanisms manifests in measurements and subjectively - and exhibits a significant dependence on amplifier output impedance.
The most well-known distortion mechanism is that due to displacement dependent non-linearities, such as from variations in the "Bl" force factor due to gap/coil geometry, for example. Since displacement is greatest at low frequencies, these distortion mechanisms predominate in the low frequency region.
If driving from a low impedance (voltage) source, displacement distortions will be greater than if driving from a high impedance (current) source because the voltage -> velocity transfer function includes a term that is proportional to the square of the Bl force factor, where as for current drive, the current -> velocity transfer function is linearly related to the Bl factor.
Such measures are readily identifiable in distortion measurements, but largely insignificant subjectively. Further discussion of these aspects are then NOT especially relevant to this thread. By contrast, however...
The second prominent distortion mechanism arises from non-linear variations in the coil impedance. The most significant cause of such distortion is typically from non-linearities in the motor system. Specifically the distortion arises from eddy currents induced by the coil current, and is therefore most prominent at mid/high frequencies where the effects of coil "inductance" are most dominant.
Firstly and most simply, the dissipation of power losses due to the eddy currents causes distortion that increases proportionally to coil current (rather than displacement). These losses are reflected in the coil impedance, and hence there exists a significant difference in how the distortion manifests that depends on the amplifier output resistance...
If driving from a low impedance (voltage) source, the distortion of the coil impedance due to the reflection of the eddy currents will appear in the voltage -> velocity transfer function, and therefore appear in the driver output. By contrast, if driving from a high impedance (current) source, the coil impedance is irrelevant and plays no part in the driver output. Thus current drive affords immunity to this form of distortion.
Secondly, there exists a complicating factor that is significant subjectively. Where the eddy currents flow in (typically) steel pole pieces, we find the effects of magnetic hysteresis (with its characteristic non-linear "S" curve) reflected in the coil impedance too. Typically we observe spiky-like third harmonic distortion products, the exact form of which is effected by coil/gap geometry among other factors.
Since (in voltage drive) these distortion products are not completely correlated to the sound pressure, they can often be perceived more readily than displacement dependent non-linearities. Often the distortions are reported as a separate "noise" or a more general "glare" or "haze". (Personally I liken the subjective effect to that evident using poorly engineered class B amplifiers or undithered digital processing). And just to reiterate, with current drive these distortions are not present in the output.
It might be insightful for experimenters to measure this distortion in the impedance with a conventional low impedance amplifier, and then to repeat the measure with the driver motion clamped. With no coil motion, but current flowing, the second current-dependent distortion mechanism will still be evident in the clamped coil impedance measures.
Regardless, it remains well-established fact that current dependent non-linearities are not relevant to current driven loudspeakers, and that measuring with both voltage and current drive will likely be required to distinguish between the two distortion mechanisms if we wish to ascribe some subjective quality to our measurements.
WHY MOTIONAL FEEDBACK IS NOT IMPORTANT
Motional feedback is not a new idea. Here we have the long-held promise that by measuring voice coil velocity (directly or otherwise), we can apply negative feedback to compensate for driver distortion. (Notably we might simply measure the "back emf" generated by the coil motion that is nominally proportional to the coil velocity and save on any additional sensor).
Motional feedback is requisite in current driven loudspeakers to control the fundamental resonance damping (look up Birt's self-balancing bridge for an ingenious implementation). In voltage drive, this is not necessary because damping is dominated by the (dynamically variable) voice coil resistance instead (i.e. subject to more distortion too).
In the case of displacement dependent non-linearities, motional feedback will indeed reduce the distortion. If the force factor is reduced, for example as displacement increases, the driving signal (voltage or current) will increase to compensate for the reduction in coil velocity. This is negative feedback in action.
However, where there is distortion due to eddy currents, motional feedback with voltage drive will not be so successful. Here the feedback signal will likewise cause an increase in the applied voltage that will serve only to increase the coil current and the distortion it causes. Thus we have effectively positive feedback in action that will prevent the desired reduction in distortion at the output.
But if we choose instead a high amplifier output impedance, the distorted coil impedance will have no effect on the coil velocity: The current induced non-linearities can be as high as we can make them, but they will not be reflected in the driver output. The transfer functions with motional feedback for voltage and current drive do not then appear to be identical.
So have we identified some obscure physics that defies Ohm's Law? Not at all. As a cautionary note to any experimenter, if we derive the velocity feedback signal from the back emf, we will simply reintroduce the distortions we had eliminated by opting for current drive. But here at least the duality of voltage and current is restored: Ohm has not been toppled.
[This all leads to a very obvious engineering solution too... Since motional feedback is generally limited to low frequencies for reasons of stability, why not employ current drive and roll-off the MFB at higher frequencies, thereby leaving the advantages at both low and high frequencies? But now we are straying off-topic somewhat].
ANOTHER RELEVANT CONSIDERATION
Another often cited and relevant advantage of current drive is that of removing altogether the significant effects of coil resistance variations with temperature. It is foolhardy to suggest that the resulting thermal compression is audible with an individual driver because the thermal time constants are too long. The key factor here is instead how thermal compression in one driver effects the overall frequency response of a multi-way loudspeaker.
We then have another factor that is simply not evident if we stick to singular driver measurements. We might then do better to consider multiple measures, where notably we can measure voice coil temperature precisely by measuring its resistance.
CONCLUSIONS (Please feel free to append any of your own)
If we wish to link subjective assessments to loudspeaker measures, we should first be clear of what those measures include and what they do not. An understanding of the physics involved helps considerably. Furthermore, a singular measure of distortion is unlikely to suffice in such a quest, and multiple measures will usefully include those made with both high and low amplifier output resistances.
By all means, experiment away. But remember that you will be limited by the value of the series resistor you add. A low output impedance amplifier has (typically) an output impedance several orders of magnitude less than the coil resistance. If you add a series resistance several orders of magnitude larger than the coil resistance for a fair comparison, you will quickly run into headroom issues.
Hi Lojzek,
Measurements are the only way forward in understanding. Properly done they represent truth and will halt reckless advertising it it's tracks.
Example. In the 1970's, they came up with a power rating called "PEP" or Peak Effective Power. A car tape deck was rated at 40 watts. The truth, 2 watts per channel. Do you have any idea how many people were ripped off? So the industry came together and no longer were things like "Music Power" or PEP acceptable. They standardized on power developed across an 8 ohm resistive load using a sine wave measured in rms to give a true power level. A notable magazine took it a bit further and ran equipment for a length of time before measuring to ensure this rating wasn't measured on an intermittent basis. In other words, to make certain the equipment delivered the rated power in real terms without failing.
So you can thank measurements for both the advances in whatever technology you care to mention, and also a truthful report of performance. They may not be perfect, but they beat the heck out of claims made in the dark, and worse, subjective opinion that can be bought or influenced by any number of means.
-Chris
Measurements are the only way forward in understanding. Properly done they represent truth and will halt reckless advertising it it's tracks.
Example. In the 1970's, they came up with a power rating called "PEP" or Peak Effective Power. A car tape deck was rated at 40 watts. The truth, 2 watts per channel. Do you have any idea how many people were ripped off? So the industry came together and no longer were things like "Music Power" or PEP acceptable. They standardized on power developed across an 8 ohm resistive load using a sine wave measured in rms to give a true power level. A notable magazine took it a bit further and ran equipment for a length of time before measuring to ensure this rating wasn't measured on an intermittent basis. In other words, to make certain the equipment delivered the rated power in real terms without failing.
So you can thank measurements for both the advances in whatever technology you care to mention, and also a truthful report of performance. They may not be perfect, but they beat the heck out of claims made in the dark, and worse, subjective opinion that can be bought or influenced by any number of means.
-Chris
Hello soundbloke,
You have no idea of what my experience covers. That's why I said what I did. I am not interested in a debate.
I do understand all the concepts you brought up, however you were pushing the thread off-topic. You also had not completely explained where you were coming from. This explanation I can mostly accept, but it is solidly off-topic.
You have no idea of what my experience covers. That's why I said what I did. I am not interested in a debate.
I do understand all the concepts you brought up, however you were pushing the thread off-topic. You also had not completely explained where you were coming from. This explanation I can mostly accept, but it is solidly off-topic.
You have no idea of my experience or qualification either. But I am not interested in a debate, I am interested only in pursuing the truth as best as scientific enquiry permits. I trust that is a shared objective.
I also believe I have been clear all along - and that includes my assertion that the matter is not off-topic. My posts clearly and consistently refer to what is required to correlate subjective assessments with loudspeakers measurements, and the need to include several measures. Current dependent non-linearities happen to provide a very good example as to why this is so, and I hope I have at last managed to convey this information.
The first question should be then: sound ( from the end ) - drivers - measurements
What is sound ? It's a type of wave that happens to be in the human hearing band that is 20-20000 Hz
It's a slow ( relatively: 344 m/s) propagation phenomenon
Characteristic of the sound is that it has a direction ,a verse, an amplitude > it's a vector! Indeed we talk about sound field
A measurement shown on the graph shows the amplitude of the DUT, quite often. Read the ( self-made) Klippel thread to see ideas of how to circumvent the basic, scheletrical showing of SPL levels along a diagram.
Hmm, forgot the drivers. Those stupid machines!
No it is not!!!! Please read my posts and think some more. The thread title is :
Do measurements of drivers really matter for sound?
The answer is yes they do, but only when the appropriate measurements are properly qualified and understood. The discussion of current dependent driver non-linearities provides an outstanding example of how diligent analyses of driver measurements are required if they are to be useful in correlating with subjective assessments of sound quality. And amplifier output impedance is no less apart of that discussion than is coil resistance. To suggest otherwise is folly.Furthermore, I did not raise the issue, I only responded to correct some misinformation. If someone would like to suggest a better aspect of driver performance or measurement then I gladly discuss that aspect instead. But I am not going to refrain from adding highly relevant information to any of my responses without good grounds. I mean no offence in anything I write, and I seek only the truth.
I am not sure I have properly understood your question, but possibly an answer you seek is ti be found in that I referred to in my post (#24) re the bispectrum, whereby we can assign labels to objects we perceive.
If you are instead referring to sound field reconstruction, then unfortunately simple vectors are only apt until about 700Hz or so. After that we need higher (spherical/circular) harmonic information, but first order ambisonics is nevertheless a good starting point.