I once did a test using DC to measure excursion:
https://www.diyaudio.com/community/...the-bl-curve-of-a-woofer.386167/#post-7034578
Not sure wether the results are very meaningful, though!
Except for very low frequencies (below resonance), maybe.
Edit: just realized it was your thread, @Boden !
https://www.diyaudio.com/community/...the-bl-curve-of-a-woofer.386167/#post-7034578
Not sure wether the results are very meaningful, though!
Except for very low frequencies (below resonance), maybe.
Edit: just realized it was your thread, @Boden !
Last edited:
Well, I've never tried it but I think you could just excite the driver at the resonance peak where current is lowest and place a small signal higher frequency sine on in, then filter out the LF tone and look how the valleys in the envelope develop vs LF tone level.
To avoid that jump resonance and other nasty stuff, the LF tone should better be low drive impedance.
They make a claim for which they provide no evidence at all.
And you parrot it.
Now, who takes the blame?
Jump Resonance definitely occurs the same with VC:
https://paforum.de/forum/index.php?...followed-collapse/&postID=1392548#post1392548
...that was in the year 2005
Those Tymphany guys ain't no idiots.
They examined the theory, modelled a driver and found the predicted behavior in the real driver. Full evidence.
As for the Dynamic DC offset (aka coil jump out and burn) which plagued the PA guys when ultra bass-heavy EDM entered the clubs and festivals, that's a problem which relates to BL(x) nonlinearity (and where electrical damping does not help) whereas the Tymphany papers deals with Kms(x) nonlinearity (where damping helps).
And for the thread topic, I've said it all in post #42. Self-fulfilling prophecy at work. The closer the source impedance approaches current drive (say, from >= 5:1 impedance ratio) the lesser the current distortion. At full current drive hence not distortion.
I'm withdrawing from the discussion.
Last edited:
Back to your posting before >KSTR< interrupted us.
If you leave the range of the semi-linear BL(x) with your dynamic transducer, then this transducer will end up in the garbage can. This will doom your project.
Nothing can be changed about that.
If you would like to hear the resulting distortion, please be very careful.
If you don't want this distortion, then VC (voltage source) won't help you. Not even a little bit.
My Drive Current Distortion Measurement shows a way to reduce distortion. My underlying project works fine. But the series resistance does not increase the range of the semi-linear BL(x).
The underlying distortion mechanism does affect the BL, but without the simple connection with (x). If I remember correctly, Wolfgang Klippel called it BL(i). (i) stands for current.
Assuming a simple case like a sine wave playing at the bass resonance frequency:
Acceleration (-sine) is 180° out of phase with displacement (+sine), and velocity (+cos) is 90° ahead of displacement.
Where I would expect the acceleration peaks to get 'clipped' due to sagging Bl, velocity crosses over zero anyway. Maybe it can get somewhat corrected 'near' the peak, but it looks like the velocity curve would get distorted in proportion to the clipped acceleration curve. I see no obvious mechanism for self-correction here.
The lumped mass is already storing the energy and working in the system's favour, as is the spring.
But, outside of the zero crossings, a distorted velocity curve would act to oppose the distorted acceleration curve which created it.
That's pretty much I would describe how (I think) it works and that it is most effective at resonance because there the internal velocity feedback is highest from the dominance of the velocity term (Back-EMF) in the driver terminal's voltage over the I*Re term, and both terms are in phase.
Sketch the graphs (sines and cosines representing displacement, velocity and acceleration) and see what happens. If the applied V and I are in phase (as they should be, at resonance), then the EMF would actually be proportional to the velocity and shifted 90°.
At those displacement and acceleration peaks where Bl drops off, it seems like the back EMF would skew the wave sideways.
At those displacement and acceleration peaks where Bl drops off, it seems like the back EMF would skew the wave sideways.
^
Hi,
back EMF is in phase, or actually opposite phase, at drivers resonance and lags 90deg above.
Snippet from https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/
"
The assumed control of cone motion
Suppose that in a voltage-driven speaker some unwanted mechanical disturbance force strives to exert the cone just at the resonant frequency. Here, the mass and spring forces (mA and kX ) cancel each other out, and thus all of the force first ends up to move the (mechanical) damper. Thus, the disturbance force translates into the velocity of the cone (F = bV ). This velocity now generates an EMF that is in phase with the velocity and hence also in phase with the disturbance force. This in-phase EMF in turn introduces an in-phase current through R c , which in turn effects a force that counteracts the original disturbance force. Thus, the disturbance-induced velocity becomes greatly reduced by the velocity-induced counterforce.
Suppose now that a similar mechanical disturbance force strives to exert the cone at some frequency well above the resonance region. Now, the force ends up accelerating the mass and thus translates into cone acceleration (F = mA ). The consequent velocity comes, by definition, 90° behind the acceleration and hence also behind the disturbance force.
Ignoring inductance, the EMF generated and hence also the resulting current and force are now perpendicular to the original force and therefore do not in any way counteract it.Thus, throughout the whole mid-frequency region, the motional EMF no longer damps or controls anything but merely acts as an uncontrolled interference source, wreaking havoc on the vulnerable V/I conversion performed by the voltage-driven speaker. (The so-called back-EMF is thus a back-EMF only near the fundamental resonance; elsewhere it is a perpendicular EMF.)
If the voice coil inductance is taken into account (and it should be), the action gets even worse. The inductance namely introduces yet an extra phase lag in the current caused by the EMF, meaning that the phase of the supposed damping force lags considerably more than 90°, and thus the whole mechanism actually turns to enhance the original disturbance instead of damping or suppressing it.
"
I like to imagine the whole thing like so:
at drivers resonance, even small force makes the cone move a lot, so relatively small current through voice coil results in great displacement and thus relatively big backEMF, this is the impedance peak. Now, the back EMF is function of the motor properties, and if any of those change like Bl, it would affect the impedance peak, which would in turn change how much current flows through the voice coil with constant voltage from amplifier.
So, to me it seems logical and plausible that as Bl drops with great displacement increase in current would compensate and keep acoustic distortion low. As the backEMF (impedance) drops with dropping Bl, more current flows and the F in the motor changes less than it could. This happens since force to the moving cone is F=Bli, as B drops the F would drop, but if i rises the F would not drop as much at least. Not sure how's the balance though, if i completely compensates for Bl, likely not, but some yeah.
If not already clear for everyone this works only at drivers resonance, where the impedance is motional impedance and involves the force factor Bl. Also assumption that it's voltage drive system so that driver impedance dominates circuit impedance to have great effect on circuit current. If there was extra impedance in series with the driver at resonance, like a resistor or capacitor or high output impedance amplifier, then varying motional impedance would have less effect on current through voice coil and thus the compensation of Bl would reduce.
If all this shows up in acoustically measured distortion plot, it means that low circuit impedance could have less distortion at driver main resonance than higher circuit impedance. And if ot is, them measuring current it would be opposite, varying impedance would show up modulating current, more distortion.
Hi,
back EMF is in phase, or actually opposite phase, at drivers resonance and lags 90deg above.
Snippet from https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/
"
The assumed control of cone motion
Suppose that in a voltage-driven speaker some unwanted mechanical disturbance force strives to exert the cone just at the resonant frequency. Here, the mass and spring forces (mA and kX ) cancel each other out, and thus all of the force first ends up to move the (mechanical) damper. Thus, the disturbance force translates into the velocity of the cone (F = bV ). This velocity now generates an EMF that is in phase with the velocity and hence also in phase with the disturbance force. This in-phase EMF in turn introduces an in-phase current through R c , which in turn effects a force that counteracts the original disturbance force. Thus, the disturbance-induced velocity becomes greatly reduced by the velocity-induced counterforce.
Suppose now that a similar mechanical disturbance force strives to exert the cone at some frequency well above the resonance region. Now, the force ends up accelerating the mass and thus translates into cone acceleration (F = mA ). The consequent velocity comes, by definition, 90° behind the acceleration and hence also behind the disturbance force.
Ignoring inductance, the EMF generated and hence also the resulting current and force are now perpendicular to the original force and therefore do not in any way counteract it.Thus, throughout the whole mid-frequency region, the motional EMF no longer damps or controls anything but merely acts as an uncontrolled interference source, wreaking havoc on the vulnerable V/I conversion performed by the voltage-driven speaker. (The so-called back-EMF is thus a back-EMF only near the fundamental resonance; elsewhere it is a perpendicular EMF.)
If the voice coil inductance is taken into account (and it should be), the action gets even worse. The inductance namely introduces yet an extra phase lag in the current caused by the EMF, meaning that the phase of the supposed damping force lags considerably more than 90°, and thus the whole mechanism actually turns to enhance the original disturbance instead of damping or suppressing it.
"
I like to imagine the whole thing like so:
at drivers resonance, even small force makes the cone move a lot, so relatively small current through voice coil results in great displacement and thus relatively big backEMF, this is the impedance peak. Now, the back EMF is function of the motor properties, and if any of those change like Bl, it would affect the impedance peak, which would in turn change how much current flows through the voice coil with constant voltage from amplifier.
So, to me it seems logical and plausible that as Bl drops with great displacement increase in current would compensate and keep acoustic distortion low. As the backEMF (impedance) drops with dropping Bl, more current flows and the F in the motor changes less than it could. This happens since force to the moving cone is F=Bli, as B drops the F would drop, but if i rises the F would not drop as much at least. Not sure how's the balance though, if i completely compensates for Bl, likely not, but some yeah.
If not already clear for everyone this works only at drivers resonance, where the impedance is motional impedance and involves the force factor Bl. Also assumption that it's voltage drive system so that driver impedance dominates circuit impedance to have great effect on circuit current. If there was extra impedance in series with the driver at resonance, like a resistor or capacitor or high output impedance amplifier, then varying motional impedance would have less effect on current through voice coil and thus the compensation of Bl would reduce.
If all this shows up in acoustically measured distortion plot, it means that low circuit impedance could have less distortion at driver main resonance than higher circuit impedance. And if ot is, them measuring current it would be opposite, varying impedance would show up modulating current, more distortion.
Last edited:
Attachments
Last edited:
Hi, could you add legend to the lines, what I'm looking at? thanks!
edit. alright it is from picture on the first page, measured current and dark colors are with series resistor and light colors without.
Not sure how to read that, definitely the light green trace varies a lot around resonance, while the other doesn't. About at ~45Hz or something like that difference between light green and dark green trace is huge, while about at ~48Hz they almost meet, and then depart again.
Which driver is it? Also, what kind of Bl(x) plot there is? is there enough excursion in the measurement to get significant deviation in Bl? and how symmetric Bl(x) is which, would balance effects between even and odd harmonics? Also suspension asymmetry would likely contribute so not sure how to actually compare the graph. And we'd likely need both acoustic and current measurement to compare the two. Anyway, I'm not sure if there is enough data in this one experiment to draw conclusion one way or another. Interesting stuff nevertheless
edit. alright it is from picture on the first page, measured current and dark colors are with series resistor and light colors without.
Not sure how to read that, definitely the light green trace varies a lot around resonance, while the other doesn't. About at ~45Hz or something like that difference between light green and dark green trace is huge, while about at ~48Hz they almost meet, and then depart again.
Which driver is it? Also, what kind of Bl(x) plot there is? is there enough excursion in the measurement to get significant deviation in Bl? and how symmetric Bl(x) is which, would balance effects between even and odd harmonics? Also suspension asymmetry would likely contribute so not sure how to actually compare the graph. And we'd likely need both acoustic and current measurement to compare the two. Anyway, I'm not sure if there is enough data in this one experiment to draw conclusion one way or another. Interesting stuff nevertheless
Last edited:
Yes thanks
I bet sound is better with the series resistor because the whole bandwidth is affected:
excursion at low frequency modulates driver parameters which affects the whole bandwidth of the driver, and thus high impedance anywhere on the bandwidth would reduce (some) motor distortion which ought to sound better. Even though acoustic distortion products originating at the resonant frequency were higher than they could, was it with or without resistor (series impedance), the sound should be better with the resistor assuming distortion was audible in the first place. I could speculate that around resonance either could sound just fine as it's room dominated sound anyway, for example, if Fs was 50Hz, 2nd harmonic would be around 100Hz and 3rd at 150Hz.
For this reason I'd say what actually happens is very interesting and fun to try figure out, not sure if it's worth losing night sleep though as sound ought to be better with resistor no matter what happens at the resonance
I bet sound is better with the series resistor because the whole bandwidth is affected:
excursion at low frequency modulates driver parameters which affects the whole bandwidth of the driver, and thus high impedance anywhere on the bandwidth would reduce (some) motor distortion which ought to sound better. Even though acoustic distortion products originating at the resonant frequency were higher than they could, was it with or without resistor (series impedance), the sound should be better with the resistor assuming distortion was audible in the first place. I could speculate that around resonance either could sound just fine as it's room dominated sound anyway, for example, if Fs was 50Hz, 2nd harmonic would be around 100Hz and 3rd at 150Hz.
For this reason I'd say what actually happens is very interesting and fun to try figure out, not sure if it's worth losing night sleep though as sound ought to be better with resistor no matter what happens at the resonance
the other way round
If I do something like that again, I'll put arrows on the graphs
Yes is isInteresting stuff nevertheless
and now I'm taking a nap
As has been said before, the described measurements are meaningless - or at least the 20R measurements are. The more the drive to the speaker is made "constant-current", the more any distortion in current is dominated by the effect of the series resistor.
This thread is a curious mixture of the Witchsmeller Pursuivant's horse witness - and the interesting...
This thread is a curious mixture of the Witchsmeller Pursuivant's horse witness - and the interesting...
It's a sort of feed-forward correction and quite crude at at. BL enters the process twice, one time as force factor for the current and the second time for the read out velocity voltage. With full motional feedback (Zout = -Re) and BL having dropped to 0.5x, the feed-forward corrects to 0.25x, that is, it applies 4x the current not the required 2x, thus overcompensating things (until BL becomes so low that the MFB runs out of gain for feedback). With a bit of "give" (reduced gain) in the motional feedback, a trade-off correction can be made.
Therefore...
... I would certainly agree to this.
Damping is governed by more important things anyway. I'm typically aiming at a resultant Qtc of below 1, and preferably at ~0.6 for the system so that cone motion is certainly not ringing when excited externally -- and the thing here is that the motor and suspension distortion is a term applied external to the cone motion. A good example is recovery from gross excursion overdrive. When the signal goes quickly back to low levels or zero we want the recovery to the rest position to be fast but aperiodic, ideally. If it's ringing with a Q above 2 or so, we get the much-feared "one-note bass" with loud transients. And IMHO this still manifests itself even in a quite live room with lots of modes... whereas a tiny change in large-signal distortion is much more benign.
Looking back on post #3 it fails a sanity check: The 20 ohm resistor linearises the measured distortion as measured through the sense resistor, but this does not prove any connection to the audio. That would still have to be done with a microphone.
As for the EMF part of the discussion above, it needs to be spelled out to me or, ideally, illustrated with colourful animations. Even if back EMF would boost the output at the resonance frequency to compensate for Bl sag because the phases are aligned, this seems like magical thinking. As soon as nonlinearity is introduced, the phase alignment is broken because we're now talking about harmonic frequencies that are far above the resonance.
As for the EMF part of the discussion above, it needs to be spelled out to me or, ideally, illustrated with colourful animations. Even if back EMF would boost the output at the resonance frequency to compensate for Bl sag because the phases are aligned, this seems like magical thinking. As soon as nonlinearity is introduced, the phase alignment is broken because we're now talking about harmonic frequencies that are far above the resonance.