@tmuikku
Good stuff. Can you do one with Sd variation?
@kimmosto
What are your thoughts about the above simulation by tmuikku ?
Does it have an high enough level of accuracy to merit consideration as an added feature to future versions of VituixCAD?
Other variables could include traced Kms(x), Sd(x). Also instead of a static voltage applied, perhaps a variable voltage?
Or M-noise?
I, for one, think it is quite instructive, although some people prefer written explanations. I get that. Everyone learns differently.
Good stuff. Can you do one with Sd variation?
@kimmosto
What are your thoughts about the above simulation by tmuikku ?
Does it have an high enough level of accuracy to merit consideration as an added feature to future versions of VituixCAD?
Other variables could include traced Kms(x), Sd(x). Also instead of a static voltage applied, perhaps a variable voltage?
Or M-noise?
I, for one, think it is quite instructive, although some people prefer written explanations. I get that. Everyone learns differently.
Last edited:
Here for comparison, 1st order electrical low pass made with DSP or with 5mh series inductor visualizing that effect of varying Le is less with the inductor.
Inductor:
DSP:
Because if the data is available then you can kind of estimate the effects by looking at the data, how much parameters vary with excursion with normal listening level. Visualizing it this way, with vituixCAD for example, is nice way kind of to get some visual representation of possible effects to get some intuition on possible audible effects, but it's really too simple to fully describe what happens. I'm no sure if it's possible to actually visualize it all. Any visuals help to get some intuition on the stuff, and realize varying of the parameters is likely audible, while some drivers perform better and some worse.
Inductor:
DSP:
Did some about a year ago, see here for example: https://www.diyaudio.com/community/...istortion-speaker-drivers.294787/post-7121933 and see previous page for more. Basically Sd modulates amplitude of whole bandwidth, directivity changes around xo with the excursion of lows.
This would be fun, but I'm not sure if it's necessary
Because if the data is available then you can kind of estimate the effects by looking at the data, how much parameters vary with excursion with normal listening level. Visualizing it this way, with vituixCAD for example, is nice way kind of to get some visual representation of possible effects to get some intuition on possible audible effects, but it's really too simple to fully describe what happens. I'm no sure if it's possible to actually visualize it all. Any visuals help to get some intuition on the stuff, and realize varying of the parameters is likely audible, while some drivers perform better and some worse.
Last edited:
I assume hysteresis in this respect is the lag between cause and affect. Would you consider woofers with stiffer suspensions and smaller Vas to have more of that lag than woofers with looser suspensions, larger Vas? Would the increase in power to the voice coil overcome hysteresis in a woofer with smaller Vas value?
Good stuff tmuikku !I bet the GIF would be quite funky if all properties were varied
Adhering to those three points has been working very well for me.
That said, I keep finding the more ways I use, the less important it's been to mandate highest quality drivers.
I guess when drivers only have to produce in the meat of their passband, it's an easier task for them.
Yes, exactly, the bigger the system the better it seems to sound, is also what I've realized going bigger and bigger Even though complexity increases and comes with all kinds of new issues fun factor keeps going up, even a poorly tuned big system seems to sound much better than small speakers. Of course there can be quite nasty big ones like some old pa speakers, or good small speakers with modern drivers like Purifi or others with effort to reduce variance of all the parameters.
Even though complexity increases and comes with all kinds of new issues fun factor keeps going up, even a poorly tuned big system seems to sound much better than small speakers. Of course there can be quite nasty big ones like some old pa speakers, or good small speakers with modern drivers like Purifi or others with effort to reduce variance of all the parameters.
Last edited:
Here another demo for fun. On this post only BL varies. Same driver with extreme BL values of 14 and 20 as on the previous posts.
As BL seems to affect system Q a lot, it indicates there might be audible differences between large closed box and small closed box equalized the same response because how the system parameters change with excursion.
Here is large and small box, click the thumbnails to see animation how BL affects system Q.
Here is with 200Hz lowpass filter with DSP, but not yet equlized the same
And here the small box equalized to ~same response as large box when BL is 14. I've always thought that DSP EQ makes them sound basically the same. Test like this shows that the responses won't be the same with small and big box even though they are equalized same, because system Q depends on BL for example, which varies with voice coil position.
Well, the response varies dramatically on all these so not sure what to make with it, perhaps they all sound different kind of bad, or good. Perhaps difference is not audible with low playback level. Anyway, interesting test
Speculating with the demos hints that a subwoofer ought to sound better using two drivers, other one mounted the opposite way for some of these effects to average out. Also suggests that don't be shy on high pass filter(s).
These simple illustrations also indicate to me that static graphs, like we usually stare at, happen only when excursion is kept minimal What ever that means for perceived sound quality. Main takeaway is that variance of parameters should be kept low, effects seem dramatic illustrated like this.
As BL seems to affect system Q a lot, it indicates there might be audible differences between large closed box and small closed box equalized the same response because how the system parameters change with excursion.
Here is large and small box, click the thumbnails to see animation how BL affects system Q.
Here is with 200Hz lowpass filter with DSP, but not yet equlized the same
And here the small box equalized to ~same response as large box when BL is 14. I've always thought that DSP EQ makes them sound basically the same. Test like this shows that the responses won't be the same with small and big box even though they are equalized same, because system Q depends on BL for example, which varies with voice coil position.
Well, the response varies dramatically on all these so not sure what to make with it, perhaps they all sound different kind of bad, or good. Perhaps difference is not audible with low playback level. Anyway, interesting test
Speculating with the demos hints that a subwoofer ought to sound better using two drivers, other one mounted the opposite way for some of these effects to average out. Also suggests that don't be shy on high pass filter(s).
These simple illustrations also indicate to me that static graphs, like we usually stare at, happen only when excursion is kept minimal
What ever that means for perceived sound quality. Main takeaway is that variance of parameters should be kept low, effects seem dramatic illustrated like this.
Last edited:
kouiky, Interesting project. Maybe some good background reading would be this thread https://www.diyaudio.com/community/threads/a-tale-of-12-subwoofers-distortion-and-15-dollars.360855/
Here I explored harmonic distortion and intermodulation distortion measurement of 12" woofers and subwoofers. I believed I gained some knowledge by listening to 13 different speakers in the same room. I didn't find one driver that I preferred for all aspects of bass reproduction but I found a few favorites:The Dayton RSS315HE-22 is the deep bass champion, clean and deep with impact, however upper bass sounded smeared.
The Morel TiCW 1258Ft excels at attack, reproduction of the Japanese taiko drum puts most woofers to shame but the Morel can do it well. The Morel also handles transitions between notes better than the rest, what I mean is that when excursion for deep bass is required and than there's a quick note change this transition is smooth with no overhang. I've heard other say this is due to the large voice coil having a good grip on the cone.
The Audio Technology 12 B 77 25 10 is the king of decay, It can capture such details as the string ringing out and slightly changing pitch. Even with deep bass the decay is incredible like someone is really putting the brakes on. I believe this woofer is used in the Rockport Antares and Stereophile magazine noticed a similar effect of a quick decay https://www.stereophile.com/content/rockport-technologies-antares-loudspeaker-page-2.
I've attached charts of the different drivers from the above thread in Intermodulation distortion (low %=cleaner sound) in a 85 liter sealed and 31hz vented (85 liter) box. Notice also that I included the results of a 6.5" Purifi in with the 10-12" drivers. The Purifi drivers sound very clean even under alot of power, a pair of 10" Purifi has close to the surface area of a 15" and this should do well for your project.
Attachments
Problem with big woofers is the larger VCs they have, which can have radial breakup resonances of their own. The problem shows itself in the midbass when a less reinforced VC begins to radially deform at certain frequencies. This can also cause VC rubbing in extreme cases when the VC moves further forward out of the gap while midrange frequencies modulate the signal ie. vocals, organ, piano or any source of strong sustained mid tones. Compression drivers can also suffer from radial VC breakup when pushed too low.
I've always advocated a separate midbass driver for larger systems. It has its benefits but also drawbacks. The main drawback is having another xover point in the lower mids, which can be very audible. The solution is a first order filter. That requires a robust midbass that can cope with lower freq shallow HP slopes. A lower compliance driver is better for this application, but running the driver in its own natural LF rolloff with a 1st order HP creates a 2nd - 3rd order HP and changes the total phase angle. The solution is to run the midbass driver one octave above its natural sealed box rolloff, but it makes the midbass enclosure much larger.
I've always noticed IMD distortion on larger 2 and 3 way systems with xover pointsI higher up in the mids. The problem is worse with high BL LF drivers with high VC Le. Its also an issue with high Q mid drivers using very small back chamber air volumes, which causes Fs to change as well as Le.
Low VC Le is a very important feature to look for if you want low distortion lower mids. As others mentioned before, a passive filter can reduce the effect, but so can a zobel. Its never a bad thing to use a properly designed zobel circuit, as it cleans up the mids substantially on drivers which play higher up into the mids. Don't be tempted to cheap out on the zobel capacitor. It also needs a properly sized resistor on higher output midbass drivers. I put zobels on almost anything if practical. Many times a zobel can be worked into the xover filter. You just have to watch that the impedance doesn't dip too low in the midbass.
I've always advocated a separate midbass driver for larger systems. It has its benefits but also drawbacks. The main drawback is having another xover point in the lower mids, which can be very audible. The solution is a first order filter. That requires a robust midbass that can cope with lower freq shallow HP slopes. A lower compliance driver is better for this application, but running the driver in its own natural LF rolloff with a 1st order HP creates a 2nd - 3rd order HP and changes the total phase angle. The solution is to run the midbass driver one octave above its natural sealed box rolloff, but it makes the midbass enclosure much larger.
I've always noticed IMD distortion on larger 2 and 3 way systems with xover pointsI higher up in the mids. The problem is worse with high BL LF drivers with high VC Le. Its also an issue with high Q mid drivers using very small back chamber air volumes, which causes Fs to change as well as Le.
Low VC Le is a very important feature to look for if you want low distortion lower mids. As others mentioned before, a passive filter can reduce the effect, but so can a zobel. Its never a bad thing to use a properly designed zobel circuit, as it cleans up the mids substantially on drivers which play higher up into the mids. Don't be tempted to cheap out on the zobel capacitor. It also needs a properly sized resistor on higher output midbass drivers. I put zobels on almost anything if practical. Many times a zobel can be worked into the xover filter. You just have to watch that the impedance doesn't dip too low in the midbass.
Hi,
have you tried measuring this? A breakup should be linear phenomenon in itself but since this owuld be heart of the motor it could vary electrical parameters with increasing level which would make it non-linear effect to sound then, I guess, and it should show up in any impedance measurement, right? perhaps impedance measurement with various drive voltages?
Isn't small speakers even more troublesome, as cone is smaller it goes greater excursion for same SPL? I mean can you generalize like that, could it be that bad sound is rather due to noisy box or edge diffraction at low frequency or something like that, symptoms of big structure and not electrical problems?
Is absolute value of Le meaningful regards distortion at all, isn't it just how much it changes with x?
I don't see how zobel would affect Le(x) for example, only increasing series impedance would and thus don't see effect on non-linear distortion. Zobel has effect on frequency response though, especially with high Le driver making it important for linear distortion.
Sorry if I sound blunt, I'm interested on your perspective on things as I really don't have too much experience on various systems, especially passive ones. I wrote more polite and elaborate and very long post but had to cut it down as it was toooo long
To make the post relevant to the thread, I'd also say that a low crossover on a three way can be quite audible because it's quite difficult to measure acoustic response at low frequencies properly at home, also room starts to have quite serious effect on things making it really hard to evaluate success by ear, It's hard to pinpoint whats wrong if the system doesn't sound right. Big boxes can be quite noisy and contain resonances pass band and passive xo parts get expensive and so on, making it hard to tweak. Perhaps phase problems on the low mids are very audible, more so than around 1-3kHz where two way xo usually is. Many of the problems exists for two way systems as well, even more so as cone breakup and stuff like that are even more on the passband as xo would usually be higher up. I'm saying it's not easy to make good sounding system with big woofer regardless of effects of BL or Le.
Well, anyway, currently I think active system with FIR capability and use steep phase linear filter for the low xo would be great, this is something I need to try on my system. Currently having ~4nd order acoustic crossover little below 200Hz with minimal 50litre box for 15", compensated with EQ. I think I liked 2nd order filter as well, but this one has less noise included, better sound in general, perhaps for multiple reasons. Bass felt less impactful at first, perhaps hearing effect of group delay, but it got better by adjusting the system with room better so not sure if I return to lower order filters. Case in point that room has very strong effect on things, and its hard to attribute some sound to some specific property of speaker as it could be due to something completely different, and most likely combination of things. FIR would be nice tool to have in a toolbox. Also, try series inductor with active system to listen if there is audible distortion reduction.
Last edited:
I have found that Z(x) is a good proxy for Le(x).
I’ve forked this out in a new post for those interested:
https://www.diyaudio.com/community/...-a-proxy-for-imd-and-amd.402805/#post-7440735
Thanks to @Arez for reminding me to do these tests.
@tmuikku You bring up some valid arguments and perspectives. Of course there are exceptions to just about every rule.
The mid VC radial breakup modes absolutely show up in the impedance curve. The problem is identifying this and separating it from other mechanical issues ie basket, spider or surround resonances. Some manufacturers (ie 18sound) address this on their compression drivers by folding and reinforcing the bottom edge of their VC formers to stiffen them up. A light weight VC has its benefits but is highly susceptible to radial modes. Plain polyimide (Kapton) formers can be the worst culprits. I also see this on a few of the 2.5" VC equipped Eminence cone drivers. I only discovered the phenomenon by accident when glueing a separate layer of Kapton ribbon onto the lower edge of the VC of a sacrificial Peavey scorpion woofer with removable magnet. This driver had a weird buzzing sound at a specific lower mid frequency, which I was determined to isolate. The resonance showed up on the impedance trace as a sharp spike. It actually left a pattern of high and low spots on the lower edge of its VC winding surface and could be calculated as a standing wave corresponding to the circumference of VC. This was some 25 yrs ago before measurement systems were reasonably cheap to diy people. I found the impedance peak with an impedance meter and sine generator.
Yes, in most cases a radial mode in a VC is linear, which predominately shows up as an even order HD peak. When its pushed harder, it becomes non-linear as rubbing on the upper pole plate restricts further motion. I used a black permanent marker to color the VC windings before the test as a method of identification. You could see the high and low spots at drive levels over roughly 20V around 900 to 1khz. A negative BL offset started to pull the VC into the gap at higher drive levels, which was whole different issue, but this smeared the marker colored VC windings which was a further interesting observation.
Smaller diameter drivers are obviously more susceptible to IMD due to needing more cone excursion for a given LF output level. IMD has a distinct type of sound. In extreme cases it basically sounds like AM modulating a tone with an LF waveform on a synthesizer if you're familiar with that sort of thing. Thats the best way I can describe it.
The higher a driver's Le is, the greater the percentage of varying Le you end up with at the same cone excursion. While its convenient to have high Le in a driver to substitute for a large value series xover inductor, it creates a higher percentage of varying inductance, constantly shifting the LP frequency. This is non-linear distortion and is one of the reasons why subwoofer drivers usually sound terrible when playing mid frequencies. The sound can be described as being dirty, muddy and lacking definition. Its often blamed on the extra moving mass of the driver, but that usually just affects cutoff frequency and can be overcome with more BL as long as Le doesn't noticeably increase with it.
Zobels help with reducing rhe effects of Le by flattening the driver's impedance rise higher up and providing the passive xover filter with a stable impedance instead of a continuously varying Le with increased VC excursion. The resulting voltage drop on the driver varies less thanks to a more stable impedance curve which won't vary nearly as much with VC excursion.
Line level active filtering is obviously immune to impedance fluctuations and variations. Thats a significant advantage of DSP over passive filters. They however can create other issues of varying degree. One is latency, which can be a problem with live monitoring, where timing and absolute phase is an issue. The other is pre-ringing. For domestic hifi use, DSP is usually more practical than passive filtering, but it can add noise and some minute artifacts which passive filters don't. I like DSP for LF processing and filtering. I don't like what it does to full range audio when it nesses with certain aspects of time domain. Some DSP is better than others. It depends on too many smaller things whether it approaches the potential neutrality of a well designed passive filter network.
Sure, its hard to isolate these symptoms by ear and differentiate them from other sources ie. enclosure standing wave modes, cabinet wall flex, cone-, basket-, spider- or port resonances, etc. You have to remove the driver from its enclosure environment to blame it on the driver itself, then use other ways of isolating each issue from other related ones. Many of these issues were discovered by accident looking for other problems. I'm a very curious guy and always want to find out why certain issues cause other similar symptoms. This is the only way to improve the sound of an already respected driver or piece of gear when most people would leave well enough alone. There's always room for improvement im my book. Many times you end up with better than average performance from less expensive drivers, which is always a good thing, but you can go backwards really easily if you get too gutsy and experimental ie. speaker cone treatments, etc.
Sorry for the long winded post. Just trying to give reasons why things may be the way they are.
The mid VC radial breakup modes absolutely show up in the impedance curve. The problem is identifying this and separating it from other mechanical issues ie basket, spider or surround resonances. Some manufacturers (ie 18sound) address this on their compression drivers by folding and reinforcing the bottom edge of their VC formers to stiffen them up. A light weight VC has its benefits but is highly susceptible to radial modes. Plain polyimide (Kapton) formers can be the worst culprits. I also see this on a few of the 2.5" VC equipped Eminence cone drivers. I only discovered the phenomenon by accident when glueing a separate layer of Kapton ribbon onto the lower edge of the VC of a sacrificial Peavey scorpion woofer with removable magnet. This driver had a weird buzzing sound at a specific lower mid frequency, which I was determined to isolate. The resonance showed up on the impedance trace as a sharp spike. It actually left a pattern of high and low spots on the lower edge of its VC winding surface and could be calculated as a standing wave corresponding to the circumference of VC. This was some 25 yrs ago before measurement systems were reasonably cheap to diy people. I found the impedance peak with an impedance meter and sine generator.
Yes, in most cases a radial mode in a VC is linear, which predominately shows up as an even order HD peak. When its pushed harder, it becomes non-linear as rubbing on the upper pole plate restricts further motion. I used a black permanent marker to color the VC windings before the test as a method of identification. You could see the high and low spots at drive levels over roughly 20V around 900 to 1khz. A negative BL offset started to pull the VC into the gap at higher drive levels, which was whole different issue, but this smeared the marker colored VC windings which was a further interesting observation.
Smaller diameter drivers are obviously more susceptible to IMD due to needing more cone excursion for a given LF output level. IMD has a distinct type of sound. In extreme cases it basically sounds like AM modulating a tone with an LF waveform on a synthesizer if you're familiar with that sort of thing. Thats the best way I can describe it.
The higher a driver's Le is, the greater the percentage of varying Le you end up with at the same cone excursion. While its convenient to have high Le in a driver to substitute for a large value series xover inductor, it creates a higher percentage of varying inductance, constantly shifting the LP frequency. This is non-linear distortion and is one of the reasons why subwoofer drivers usually sound terrible when playing mid frequencies. The sound can be described as being dirty, muddy and lacking definition. Its often blamed on the extra moving mass of the driver, but that usually just affects cutoff frequency and can be overcome with more BL as long as Le doesn't noticeably increase with it.
Zobels help with reducing rhe effects of Le by flattening the driver's impedance rise higher up and providing the passive xover filter with a stable impedance instead of a continuously varying Le with increased VC excursion. The resulting voltage drop on the driver varies less thanks to a more stable impedance curve which won't vary nearly as much with VC excursion.
Line level active filtering is obviously immune to impedance fluctuations and variations. Thats a significant advantage of DSP over passive filters. They however can create other issues of varying degree. One is latency, which can be a problem with live monitoring, where timing and absolute phase is an issue. The other is pre-ringing. For domestic hifi use, DSP is usually more practical than passive filtering, but it can add noise and some minute artifacts which passive filters don't. I like DSP for LF processing and filtering. I don't like what it does to full range audio when it nesses with certain aspects of time domain. Some DSP is better than others. It depends on too many smaller things whether it approaches the potential neutrality of a well designed passive filter network.
Sure, its hard to isolate these symptoms by ear and differentiate them from other sources ie. enclosure standing wave modes, cabinet wall flex, cone-, basket-, spider- or port resonances, etc. You have to remove the driver from its enclosure environment to blame it on the driver itself, then use other ways of isolating each issue from other related ones. Many of these issues were discovered by accident looking for other problems. I'm a very curious guy and always want to find out why certain issues cause other similar symptoms. This is the only way to improve the sound of an already respected driver or piece of gear when most people would leave well enough alone. There's always room for improvement im my book. Many times you end up with better than average performance from less expensive drivers, which is always a good thing, but you can go backwards really easily if you get too gutsy and experimental ie. speaker cone treatments, etc.
Sorry for the long winded post. Just trying to give reasons why things may be the way they are.
Last edited:
Hi,
long posts are fine for me, thanks! I hope its fine to respond, I try to keep it short
Here is quick VituixCAD test to show. It is simulated woofer response I used earlier, specs from a woofer whose Klippel data shows Le varies about 0.13mH with xmax. There is 2nd order low pass and Zobel illustrated showing response with and without the zobel. Inductor in series with the driver simulates the 0.13mH change that happens with xmax travel:
Here is effect of zobel. It linearizes the load impedance for the filter. Without zobel the filter interacts with driver impedance and there is a peak in the response. But, effect of Le(x) is almost the same with or without zobel, as proportion of driver Le to circuit impedance (for backEMF) stays about the same. It has some effect on the electonic low pass filter, as you say.
But, what if we take the cap out and just use series inductor, make it 1st order filter. This increases impedance in series with the driver as the "bypass" cap is now removed. Now zobel actually makes bypass here, reduces impedance in series with the driver on high frequencies and makes effect of Le(x) worse in this filtering scheme:
To actually reduce effect of Le(x) just analyze circuit from perspective of the driver, what series impedance is with the driver, and try to increase it. Le(x) needs to appear as current in the circuit to make into acoustic sound. Increasing circuit impedance the current is reduced as load for backEMF was increased. Or, basically circuit impedance is now so high that Le(x) makes proportionally less of it, making variance less, which ever feels more intuitive way to think about it
Also, if it was fully passive it matters which kind of topology there is, not just for frequency response but also for distortion.
I'm slowly finding out my way what is important and what is not. DSP has been great so far, using impedance manipulation as well but not sure if it's particularly audible when system is big enough. Haven't done much A/B testing yet, just going forward for now what ever aspect it is that pokes through somehow, improving listening skills before all
long posts are fine for me, thanks!
I hope its fine to respond, I try to keep it short Jeah definitely could be reason if there is audible difference. I have no such drivers so don't know about he sound, but great amount of Le(x), possibly huge surround as well for Sd(x) and so on, it's not hard to believe there is quite some effects to sound.
I think this is not true, as the Le would vary still, it's effect to circuit impedance in series with the driver would be pretty much the same thus effect of current in the circuit would be pretty similar and thus acoustic output, even if there is parallel network like Zobel.
Here is quick VituixCAD test to show. It is simulated woofer response I used earlier, specs from a woofer whose Klippel data shows Le varies about 0.13mH with xmax. There is 2nd order low pass and Zobel illustrated showing response with and without the zobel. Inductor in series with the driver simulates the 0.13mH change that happens with xmax travel:
Here is effect of zobel. It linearizes the load impedance for the filter. Without zobel the filter interacts with driver impedance and there is a peak in the response. But, effect of Le(x) is almost the same with or without zobel, as proportion of driver Le to circuit impedance (for backEMF) stays about the same. It has some effect on the electonic low pass filter, as you say.
But, what if we take the cap out and just use series inductor, make it 1st order filter. This increases impedance in series with the driver as the "bypass" cap is now removed. Now zobel actually makes bypass here, reduces impedance in series with the driver on high frequencies and makes effect of Le(x) worse in this filtering scheme:
To actually reduce effect of Le(x) just analyze circuit from perspective of the driver, what series impedance is with the driver, and try to increase it. Le(x) needs to appear as current in the circuit to make into acoustic sound. Increasing circuit impedance the current is reduced as load for backEMF was increased. Or, basically circuit impedance is now so high that Le(x) makes proportionally less of it, making variance less, which ever feels more intuitive way to think about it
Yeah the filters are not affected, but still the Le(x) makes distortion with active systems with "high damping factor" amplifiers, which basically short the driver and let Le(x) make maximal fluctuation to circuit current and thus into acoustic domain. Add inductor between amp and the driver to make impedance manipulation reducing effect to current by Le(x), and use DSP to make the response you want
Also, if it was fully passive it matters which kind of topology there is, not just for frequency response but also for distortion.Likewise!
I'm slowly finding out my way what is important and what is not. DSP has been great so far, using impedance manipulation as well but not sure if it's particularly audible when system is big enough. Haven't done much A/B testing yet, just going forward for now what ever aspect it is that pokes through somehow, improving listening skills before all
The effects of interaction between a series inductor and VC Le will depend on the ratio of the two and the surrounding passive filter topology, as you demonstrated. How the filter is staggered in which sequence of parts also plays a lage role. If the zobel falls within the worst case scenario of having the impedance minimum of both Le following parallel C1 and reactance of Zobel, it will in theory cause the most chaos in the upper driver rolloff as you pictured.
Simulation is required to get the best compromise between all the scenarios and when implemented correctly per sim results, the effects of (delta) Le on upper FR greatly diminish. Of course, if you use a monster motor driver with 4 layer VC that has a 2+ mH Le along with a smaller series inductor, it will magnify the peaking effect the zobel has higher up on the upper rolloff, so the proportions matter alot. A huge series inductor will push all that out of band, as the driver will operate in its smoother phase region as it should.
That peaking you illustrated above can imposter as the breakup ringing of a stiff cone.
Simulation is required to get the best compromise between all the scenarios and when implemented correctly per sim results, the effects of (delta) Le on upper FR greatly diminish. Of course, if you use a monster motor driver with 4 layer VC that has a 2+ mH Le along with a smaller series inductor, it will magnify the peaking effect the zobel has higher up on the upper rolloff, so the proportions matter alot. A huge series inductor will push all that out of band, as the driver will operate in its smoother phase region as it should.
That peaking you illustrated above can imposter as the breakup ringing of a stiff cone.
Hi, jeah it's rather easy to play with a system in simulator and try to figure out caveats.
It's a complex system though, speaker with only passive network, as the same network manipulates impedance for both the amplifier and the driver so to speak, which both have effects on frequency response and distortion, amplifier load is easy to analyze but effect of backEMF is often not considered.
With active system one can make the frequency response with filters before amplifier and use passive parts just to manipulate impedance of the circuit to reduce distortion. Passive filters it's just more things to juggle around, I could say that for example using first order filter is not good idea for most systems but require quite elaborate planning. Why not use 1st order filter just for impedance and make acoustic response better with steeper filters in DSP
It's a complex system though, speaker with only passive network, as the same network manipulates impedance for both the amplifier and the driver so to speak, which both have effects on frequency response and distortion, amplifier load is easy to analyze but effect of backEMF is often not considered.
With active system one can make the frequency response with filters before amplifier and use passive parts just to manipulate impedance of the circuit to reduce distortion. Passive filters it's just more things to juggle around, I could say that for example using first order filter is not good idea for most systems but require quite elaborate planning. Why not use 1st order filter just for impedance and make acoustic response better with steeper filters in DSP