When you have a driver at hand, then all you can do is increase series impedance with it to reduce some distortion, but the Bl(x) or Sd(x), no can do other than use better driver or change system design.
Basic system design stuff, design the system so that it's non issue, and in addition use best driver(s) for the budget as well as consider the circuit impedance if at all possible. Select driver that has all the varying parameters under control to meet target performance, and/or design the system so that excursion is kept low. So, for cheap small speaker IMD could be audible issue, but for a big speaker or high budget speaker it's likely not an issue.
Last edited:
You've completely grasped it again. 🙂
In the very early 2000 years >B&C Speakers< "invented" the "double silicone spider".
These low frequency transducers were the first to survive modern music in the pro audio / sound reinforcement business. But these transducers were not well received because the double silicone spider prevented sound people from exceeding x-max.
Last edited:
...that voltage drive is the remedy for coil jump out?
This otherwise marvelous coil jump out paper was specifically set up to deter from current drive.
Cite of the last sentence of the Abstract:
"The importance of this phenomenon increases as the use of current amplifiers becomes more
widespread, since the resulting low system damping makes jump resonance more likely."
They have not investigated this claim.
This is anti science!
However, they have successfully managed to anchor this gross nonsense in the public eye.
What do you mean they have not investigated it? the paper introduces what is jump resonance and shows relatively simple math how it plays out with or without damping factored in, and using a test procedure (simulation) to demonstrate it happens on low damping situation. It's logical to use low damping in a test, if that is what allows the phenomenon to happen in the first place. The math they lay out shows low damping make the phenomenon more pronounced, which means current drive amplifier would make any driver more likely for the phenomenon, like they state in the abstract. Not sure what would be wrong with the statement? You are right it would be more thorough if high damping system was also simulated.
Here is how to imagine how the system is assumed to work, at least how I understand what they present in the paper:
The effect is assumed to happen with non-linear suspension, in other words when excursion gets high the suspension stiffens and makes more returning force than if it was linear. Think simplified an ideal non-damped system where mass and spring cancel each other so if you poke the system with a finger just a little and it makes great oscillation, and keeps on oscillating. Now, factor in non-linear suspension and poke it hard enough to make the suspension stiffen at peak excursion, which returns the mass back faster than it went there, and it's kind of a runaway situation. When the suspension stiffens relationship of the mass and spring forces change, which changes the resonating frequency of the system momentarily and thus name jump resonance: As signal sweep is approaching the system resonant frequency, quite close to the resonant frequency the excursion goes up enough to hit the stiffening suspension, which makes the system resonant frequency change and throws the system into oscillation even when it's not yet at the normal low signal resonating frequency, now the system jumped into resonating at another frequency giving suddenly great excursion.
Now, imagine there was damping in the system, which opposes movement and no "runaway" happens. The resonant frequency of the system would not jump so that it would sustain a resonance at another frequency, only momentarily at the peak excursion but the damping would quickly slow it down. Even though the non-linearity is still there at the peak excursion and could happen either way, the system gravitates away from it, slowing down before hitting the opposite end of excursion.
It's very very logical to me thought this way. If one thinks its bogus life can go on, no need to stress about it, if there is no more elaborative papers on it then it isn't likely very common or problematic phenomenon.
If you have a good explanation how to imagine the system then please do post as it's fun trying to understand stuff in this kind of way, where it feels intuitive. Math is not intuitive to me, while imagining stuff like this very much is.
ps. here very scientific illustration about extra momentum from stiffening suspension, the other has some damping, while the other one has less and runs away 😀😀
Here is how to imagine how the system is assumed to work, at least how I understand what they present in the paper:
The effect is assumed to happen with non-linear suspension, in other words when excursion gets high the suspension stiffens and makes more returning force than if it was linear. Think simplified an ideal non-damped system where mass and spring cancel each other so if you poke the system with a finger just a little and it makes great oscillation, and keeps on oscillating. Now, factor in non-linear suspension and poke it hard enough to make the suspension stiffen at peak excursion, which returns the mass back faster than it went there, and it's kind of a runaway situation. When the suspension stiffens relationship of the mass and spring forces change, which changes the resonating frequency of the system momentarily and thus name jump resonance: As signal sweep is approaching the system resonant frequency, quite close to the resonant frequency the excursion goes up enough to hit the stiffening suspension, which makes the system resonant frequency change and throws the system into oscillation even when it's not yet at the normal low signal resonating frequency, now the system jumped into resonating at another frequency giving suddenly great excursion.
Now, imagine there was damping in the system, which opposes movement and no "runaway" happens. The resonant frequency of the system would not jump so that it would sustain a resonance at another frequency, only momentarily at the peak excursion but the damping would quickly slow it down. Even though the non-linearity is still there at the peak excursion and could happen either way, the system gravitates away from it, slowing down before hitting the opposite end of excursion.
It's very very logical to me thought this way. If one thinks its bogus life can go on, no need to stress about it, if there is no more elaborative papers on it then it isn't likely very common or problematic phenomenon.
If you have a good explanation how to imagine the system then please do post as it's fun trying to understand stuff in this kind of way, where it feels intuitive. Math is not intuitive to me, while imagining stuff like this very much is.
ps. here very scientific illustration about extra momentum from stiffening suspension, the other has some damping, while the other one has less and runs away 😀😀
Last edited:
And then the coil of the poor vifa transducer hit the backplate with voltage control. With its last loud bang. So no proof of the math.
Look my posting #63.
Now please listen:
This jump resonance thingy is a "red herring" landed in my thread: Drive Current Distortion Measurement.
Build a Loudspeaker!
Measure it!
I Did it!
That was very enriching for me!
Yeah just reading it from the beginning, jump resonance in the paper is different thing than what hörnli posts on the paforum thread (first page).
On the thread is same stuff we'd already had here. You wrote there that when Bl drops, impedance drops so much that current increases so much that force on the cone actually increases, making the former jump out / hit backplate. This is completely inline what KSTR also wrote, which I didn't understand properly until now. This phenomenon indeed happens with voltage drive, while the jump resonance on the paper does not, which is different phenomenon.
So nice, understanding has increased, thanks!
On the thread is same stuff we'd already had here. You wrote there that when Bl drops, impedance drops so much that current increases so much that force on the cone actually increases, making the former jump out / hit backplate. This is completely inline what KSTR also wrote, which I didn't understand properly until now. This phenomenon indeed happens with voltage drive, while the jump resonance on the paper does not, which is different phenomenon.
So nice, understanding has increased, thanks!
Last edited:
I don't know, some of the claims seem to be a mess. For one thing, there seems to be an (untested?) assumption that with current drive a big resonance exists due to multiplication of drive current x blocking voltage. People look at the published impedance curves and assume that that is the real impedance all the way out to Xmax and beyond.
However, Bl is already known to sag at high displacement, so the blocking voltage and motional impedance will also sag.
If real-world output levels are ±5 mm on bass peaks, or 0.5mm, the optimum damping level may be completely different for the same speaker.
However, Bl is already known to sag at high displacement, so the blocking voltage and motional impedance will also sag.
If real-world output levels are ±5 mm on bass peaks, or 0.5mm, the optimum damping level may be completely different for the same speaker.
Hi, where is this? We've got some momentum now, so could try and solve any remaining confusion 😀
yeah this is what has been discussed, except I'm not sure what you mean by blocking voltage?
For sure, isn't some PA boxes tuned with this in mind, that with high output level (and heat) the system aligns to target. Damping would be part of system Q.
Last edited:
So - what measurement microphone do you use?
Seriously - you are quoting SM57 use for measurements? And comparisons with a Dayton UMM6? Of course do these mini electret mics have plenty of noise AND THD cause they can't cope with even medium SPL in measurements.
You need to use the right tool. And a 100 bucks microphone is "surprsingly" not good enough to do more serious work in acoustics.
The simulated bass peak for the 3" Vifa driver is just not plausible. Consider this: if the voice coil is clamped then there will be no impedance peak. And the same thing happens twice during each cycle of a sine wave: at the peak acceleration where velocity goes to zero. Then, as the cone starts moving again, a voltage is generated, but at the extremes (> ±3 or 4 mm) it could be less than half of what would be expected with small signals where Bl strength is at max.
So the motional impedance will be modulated up and down as well.
To get the supposed 8mm of displacement, it looks like they forgot to make the Bl actually drop off to a lower level at the peaks (along with lower back EMF and therefore lower motional impedance).
Hi,
I'm not sure if I'm catching your line of thought so in order not to increase confusion I'll first summarize what has been going on in the thread:
The thread has had at least three major topics and very confusing discussion between them: starting from measuring current because microphones have distortion, then about coil jump out issue mixed with jump resonance paper.
Now, your post touches the latter two:
For the 8mm displacement: The system in the simulation has low damping, so drive current in simulation was likely chosen to get max excursion at resonance (with the low damping). It is constant current sweep, so thats why output drops outside the resonance, spring and mass comes into play above resonance. Point of the paper is not amount of excursion, but to demonstrate the excursion jumps suddenly from low to high, and not at low signal Fs.
Anyway, of course coil could pop out with enough current with current drive as well, so max excursion can be reached with current or voltage drive, even though the Bl sags.
Does it make sense?
I'm not sure if I'm catching your line of thought so in order not to increase confusion I'll first summarize what has been going on in the thread:
The thread has had at least three major topics and very confusing discussion between them: starting from measuring current because microphones have distortion, then about coil jump out issue mixed with jump resonance paper.
Now, your post touches the latter two:
Jeah this seems reasonable, I have same logic, also KSTR explained it, and this is also what hörnli has been trying to communicate over. It makes coil jump out from the gap with voltage drive: as impedance drops, current goes up so much that force increases, instead of going down with Bl like intuition says, and could push the coil out.
the test in the jump resonace paper is done with current drive, so the motional impedance has no effect on current, and no effect on acoustic output. Also, this prevents the coil jumping out like hörnli has been trying to communicate, current doesn't increase as Bl drops, so force drops with it.
For the 8mm displacement: The system in the simulation has low damping, so drive current in simulation was likely chosen to get max excursion at resonance (with the low damping). It is constant current sweep, so thats why output drops outside the resonance, spring and mass comes into play above resonance. Point of the paper is not amount of excursion, but to demonstrate the excursion jumps suddenly from low to high, and not at low signal Fs.
Anyway, of course coil could pop out with enough current with current drive as well, so max excursion can be reached with current or voltage drive, even though the Bl sags.
Does it make sense?
Last edited:
Hmm, trying to think about the system and what would make coil pop out from the gap. Force at peak excursion is accelerating the cone to the other direction, right, and if the force increases at peak excursion in voltage drive situation it should prevent the coil popping out from the cap? I mean, as cone goes past the rest position it has great speed and electrical damping would try to slow it down, and also F=Bli is already slowing it down, and if it increases it should not help coil pop out but prevent it?
So, in order to coil hit backplate or jump out either F should reduce at peak excursion to let the coil slip through so to speak, or electrical damping to reduce so that the speed was more than should? perhaps it's more complicated, perhaps Bl drops, current increases in average, coil heats up which would reduce both the F and electrical damping and at some point in time the coil pops out, especially if the operator tries to push the system to compensate reduced output. Or something like that.
Hörnli, could you explain more what you knoe about the subject? I'll read the thread you linked, but its tough with translator.
So, in order to coil hit backplate or jump out either F should reduce at peak excursion to let the coil slip through so to speak, or electrical damping to reduce so that the speed was more than should? perhaps it's more complicated, perhaps Bl drops, current increases in average, coil heats up which would reduce both the F and electrical damping and at some point in time the coil pops out, especially if the operator tries to push the system to compensate reduced output. Or something like that.
Hörnli, could you explain more what you knoe about the subject? I'll read the thread you linked, but its tough with translator.
I mentioned the case of a higher frequency signal in the presence of significant low-frequency excursion, from the point of view of intermodulation distortion. If you take a look at the equivalent circuit developed here, you'll see that current drive would result in the highest susceptibility of acoustic output to variation in Bl:
https://audiojudgement.com/speaker-equivalent-circuit/
In fact, an amplifier with negative output impedance would be the most appropriate in reducing distortion...
https://audiojudgement.com/speaker-equivalent-circuit/
In fact, an amplifier with negative output impedance would be the most appropriate in reducing distortion...
I'm forgetting here that at resonance mass is cancelled by spring so the force goes to velocity directly, thus at peak excursion if force increases it would push the cone further, and could ppp it out from the gap.
Quote from Esa's article in EDN
"
...
When a driving force F = BlI is applied to the system, the force divides into three components:
F = BlI = mA + kX + bV
where mA is the force accelerating the mass, kX is the force stretching the spring (X = displacement), and bV is the force moving the (virtual) damper. mA forms the dominant force in most of the drivers operating band (i.e. above the resonance), kX forms the dominant force below the resonance, and bV is dominant at the resonance, where the two other components are equal but opposite phase and hence cancel each other out. On voltage drive, the current I is not constant but exhibits a notch in the resonance region.
...
"
ps. Note the last sentence, included that in the quote as it's relevant. It doesn't go into the issue here, that there is also a peak in current, what makes coil jump out, current fluctuates at resonance with the Bl with voltage drive. Anyone with the current drive book, is there section about this phenomenon?
Hi, could you help, how you derive this information from the equivalent circuit? I'm stuck on how to determine acoustic output from it? by analyzing current through the laat transformation, through the cap and resistor far right?
Currently I'm not sure there is difference in Bl related IMD between current or voltage drive, because Bl is (over)compensated on voltage drive only at the resonance and there is nothing we can do about it above as any current would be perpendicular to it and not have an effect. Whats your reasoning behind it, can you read it from the equivalent circuit somehow?
The air load on the right has the capacitive element to reflect the degree of coupling between diaphragm and air, which appears as mass at low frequencies (determined by diaphragm size).
To achieve undistorted signal at this point would require constant voltage - otherwise the non-linearities presented by the shunt elements would take effect. This would clearly need a means of counteracting the effects of the series elements - hence the utility of negative output impedance.
This isn't the whole story, of course, as the model does not show the fact that, in practice, the series elements (Re and Le) are themselves dependent on factors such as excursion and current (Le), and temperature (Re). Which means that, if it weren't for the shunt elements, current-drive would be desirable...
To achieve undistorted signal at this point would require constant voltage - otherwise the non-linearities presented by the shunt elements would take effect. This would clearly need a means of counteracting the effects of the series elements - hence the utility of negative output impedance.
This isn't the whole story, of course, as the model does not show the fact that, in practice, the series elements (Re and Le) are themselves dependent on factors such as excursion and current (Le), and temperature (Re). Which means that, if it weren't for the shunt elements, current-drive would be desirable...