Does it beg for parralell resistor in serie on the return signal path and or parralel resistor to the driver to improve (damp somewhat) back EMF from the amplifier whatever it is: damped high order filter inductance, zobel, spl attenuation, if made passive ?
Or does it mean a dedicated ampfor the rweeter so active or dedicated with gain matching if passive would make the tweeter to sound cleaner by avoiding at least mid and bass drivers massive back EMF (is back EMF rings the tweeters with electrical harmonics too or it occurs at alk rhe frequencies involved by the music signal of the playback as a nasty boomerabg ?
Or does it mean a dedicated ampfor the rweeter so active or dedicated with gain matching if passive would make the tweeter to sound cleaner by avoiding at least mid and bass drivers massive back EMF (is back EMF rings the tweeters with electrical harmonics too or it occurs at alk rhe frequencies involved by the music signal of the playback as a nasty boomerabg ?
Is there a problem with back EMF?
The output of the amplifier keeps the woofer crossover and the tweeter crossover separated because it is a short circuit.
The output of the amplifier keeps the woofer crossover and the tweeter crossover separated because it is a short circuit.
Yes, a short for AC. At that point the Voltage will be what the amplifier dictates.
The Purifi blog had a nice example at 5kHz like you say, but AFAICT, this could be also done with a high-pass. The notch should still do its thing with the L||C part going open-circuit at resonance.
I always find it fascinating to read why certain material properties are linked to certain behavior.
One of the funniest ones are like silver sounds bright and copper sounds warm.
Also as if there is only "one best" (dramatic effect!) thing of doing things.
For any driver the goal is pretty simple, no (major) disturbances in the pass band, a workable frequency band and a low (enough) distortion. Nothing more, nothing less.
(depending on context, one could add some additional, like maxSPL etc, although for home hifi that is all very limited)
Unfortunately in physics as well as mechanical properties, there are always compromises to any solution.
Plus there is a way of no (or very little) diminishing return.
So combining this all together, speakers with no substantial issues in the used frequency band, are in fact identical sources.
It's irrelevant what material it's made of, even if it was duct tape or cardboard.
One can debate what the term "usable" means, especially in the sense of distortion (since a good frequency response is given these days), but that is an entire different discussion.
So I think people ask the wrong question.
I think a better question is what a different (more expensive) material will bring us in a acoustical sense?
To already answer that question a little bit; there isn't a lot.
Especially when we add psycho-acoustics, age (sorry) and other things into the mix.
Also keep in mind, you pay more than just for plain audio and acoustic performance.
Making a very fancy face-plate and nicely milled iron looks great, but it doesn't add anything in sense of performance.
Yet it's a huge step in production costs!
The same goes for other fancy materials for like cones and domes.
There is absolutely nothing wrong with that btw!
With a good sturdy dining table I also want good quality chairs.
It doesn't make my poor cooking any better though. 😉
One of the funniest ones are like silver sounds bright and copper sounds warm.
Also as if there is only "one best" (dramatic effect!) thing of doing things.
For any driver the goal is pretty simple, no (major) disturbances in the pass band, a workable frequency band and a low (enough) distortion. Nothing more, nothing less.
(depending on context, one could add some additional, like maxSPL etc, although for home hifi that is all very limited)
Unfortunately in physics as well as mechanical properties, there are always compromises to any solution.
Plus there is a way of no (or very little) diminishing return.
So combining this all together, speakers with no substantial issues in the used frequency band, are in fact identical sources.
It's irrelevant what material it's made of, even if it was duct tape or cardboard.
One can debate what the term "usable" means, especially in the sense of distortion (since a good frequency response is given these days), but that is an entire different discussion.
So I think people ask the wrong question.
I think a better question is what a different (more expensive) material will bring us in a acoustical sense?
To already answer that question a little bit; there isn't a lot.
Especially when we add psycho-acoustics, age (sorry) and other things into the mix.
Also keep in mind, you pay more than just for plain audio and acoustic performance.
Making a very fancy face-plate and nicely milled iron looks great, but it doesn't add anything in sense of performance.
Yet it's a huge step in production costs!
The same goes for other fancy materials for like cones and domes.
There is absolutely nothing wrong with that btw!
With a good sturdy dining table I also want good quality chairs.
It doesn't make my poor cooking any better though. 😉
Having been an aluminium welder by trade the extra cost of welding it is not so much electricity cost but the shielding gas required.
You can weld steel just fine using CO2 but aluminium needs an argon mix and with titanium you need to shield with pure argon and have to shield the back of the weld as well. That is beyond the cost differentials of the materials themselves.
If you'd like a new steel road frame check out Officina Battaglin who make IMO the most beautiful frames available today
Homepage | Officina Battaglin
@b-force (before?) : well I see most of the flag ships today doesn't use metal dome... But there are still some brands whom using craps basic Solen caps in their flag ships, so I surmise, the discussion is always ending or beginning also about the layout. But in parrallel with the drivers or somz drivers that are too soft sounding i never use Solen basic caps (SCR), not they are bad caps, but a little rough with some drivers. I do not talk of the others silver and tin I like quite well 😉
Sometimes, some brands, just focus on a point and forget the whole.
Sometimes, some brands, just focus on a point and forget the whole.
Yes it could, whether you'd choose to do it that way is another story. It won't stop the cone from breaking up though, and the assumption is that the breakup region has to go away.. otherwise power is thinned in that region.
More importantly though, distortion is the least of the problems at breakup. Besides, if you can't hear it at that frequency then distortion becomes near irrelevant at higher frequencies.
Well, it reflects back at higher order (often 3rd order) with a lower frequency.
Moving into de relevant frequency band, even when a filter is applied.
Depending on the severity of this cone breakup, but for even some high-end drivers this can be pretty nasty.
Just like metal cone woofers, most good metal diaphragms no longer use just pure metal, but also incorporate
extra materials to tame 'Ring & Resonance'. This makes good sense to me.
When I said "extra materials", I didn't mean Metal Aloys. I was referring to 'rings & stripes' of plastic/rubber type materials.
Sometimes used in a 'sandwich' dual layer style. Just imagine the underside of a metal dome 'sprayed' with ultra thin rubber.
I don't believe that the coating of metal cones is purely to stop corrosion, it can play a far more important role.
It is my opinion that both Electrical and Mechanical damping can play a part in good sound.
Also, it would have to be a particularly bad driver to need benefit from a L/C/R notch filter.
extra materials to tame 'Ring & Resonance'. This makes good sense to me.
When I said "extra materials", I didn't mean Metal Aloys. I was referring to 'rings & stripes' of plastic/rubber type materials.
Sometimes used in a 'sandwich' dual layer style. Just imagine the underside of a metal dome 'sprayed' with ultra thin rubber.
I don't believe that the coating of metal cones is purely to stop corrosion, it can play a far more important role.
It is my opinion that both Electrical and Mechanical damping can play a part in good sound.
Also, it would have to be a particularly bad driver to need benefit from a L/C/R notch filter.
Breakup isn't nonlinear and it does not distort. It doesn't create distortion either.
All it typically does is reduce some off-axis sound and increase the on axis sound at selected frequencies, or similar. There's not necessarily more output, unless you sit on axis, which isn't what should be done most of the time.
All it typically does is reduce some off-axis sound and increase the on axis sound at selected frequencies, or similar. There's not necessarily more output, unless you sit on axis, which isn't what should be done most of the time.
I seem to recall we've all had this particular conversation a few times before. Breakup can induce / amplify the baseline HD at sub-multiples of its own frequency. This can clearly be seen on HD plots (HD2 - HD5 most conveniently). Mostly an issue with highly rigid cones / domes & much less obvious with softer types with their more progressive / damped TL modes. Yevgeniy and the guys at Purifi most obviously have discussed the mechanisms for this, as have a few of the rest of us. Whether it can be said to 'create' distortion is an interesting philosophical / linguistic question, since I suppose it could be argued that what it's doing is amplifying what already exists, given that there's no such thing as a distortion free motor. Probably best left to those who like debating the finer points of grammar though. 
Either way -it can (not always, obviously) be an issue. Ideally, you'd cross below the problematic HD spike, if it's an LF or midrange of course, but a high-impedance parallel LC notch in series with some units can stamp on a problematic breakup mode & attenuate the HD amplification lower down in the passband / transition band in the process, which is a useful potential alternative to have up the old sleeve. 
Also can be useful with some hard dome tweeters.
Either way -it can (not always, obviously) be an issue. Ideally, you'd cross below the problematic HD spike, if it's an LF or midrange of course, but a high-impedance parallel LC notch in series with some units can stamp on a problematic breakup mode & attenuate the HD amplification lower down in the passband / transition band in the process, which is a useful potential alternative to have up the old sleeve. Also can be useful with some hard dome tweeters.
Last edited:
yeah the breakup is amplifier in acoustic domain, a mechanical amplifier so to speak. Think, frequency response is measured with constant voltage sweep and impedance of a driver is usually relatively "constant" through the breakup, which means current is relatively constant through the breakup so the huge axial boost in frequency response is not from more current making more force in the motor, but must be from mechanical parts (yeah, the cone breakup obviously). Implication of this is that any current through voice coil at the breakup makes the assembly move like usual but gets boosted by the breakup in acoustic domain. Magnitude is usually boosted on-axis, attenuated off-axis. So, any distortion current from driver motor non-linearities, any amplifier distortion, any amplifier noise, any signal in the electrical circuit basically, gets boosted in acoustic domain by the breakup.
Now, imagine notching down the breakup peak with DSP, which effectively notches signal going into the power amplifier. Nice, as signal was notched in electronic domain and then boosted back equal amount by the cone resonance in acoustic domain we measure flat frequency response on-axis. But amplifier noise, any distortion due to driver motor, basically all distortion after the DSP get amplified by the resonance just like before as they were not preconditioned by DSP, they get extra amplification from the breakup.
To reduce amplification of noise and distortion by the breakup one has to increase series impedance in the circuit to reduce all current through voice coil.
Whats the problem, how audible is it?
I don't know, but we can try to reason about it: distortion is usually explained to be inaudible because of the actual musical content masking it, but how is masking here? The breakup is usually left outside pass band of the particular driver, low pass crossover below breakup due to the breakup 🙂 Now, the breakup boosts any distortion out of band, imagine having brickwall filter with DSP but you'd still hear distortion products from pass band extending out of band multiple octaves with no original content on the same bandwidth masking it. So, not that much masking from the driver itself but sound from other driver (tweeter) could mask it? The other drivers are usually at different physical location, which might reduce effectiveness of the masking.
In addition there is the effect on directivity, not sure how audible this would be.
From the above we can reason that if distortion is ever audible in the first place, then this is probably the most audible occurence as it the peak is static and high Q, and thus not related to musical content but to the driver, and masking doesn't seem to be as effective as it could. Intuitively bad breakup should stick out like red flashing light if listening on-axis. But then again, I'm not sure if I can hear it.. haven't learned to hear it yet I think, if its audible at all with the volume levels I usually listen to. Haven't deviced AB test yet.
Takeaway, for best sound use driver with least breakup = smooth polar response and no distortion sticking out. If you have to work with what you've got then use passive network, parallel notch in series with voice coil, to notch the peak down. Especially with active system.
For tweeters, I'm not sure if anyone can hear the breakups past 20kHz, perhaps they do. If its bad then notch it out, or low pass, by increasing series impedance at breakup frequency just like with a woofer.
Now, imagine notching down the breakup peak with DSP, which effectively notches signal going into the power amplifier. Nice, as signal was notched in electronic domain and then boosted back equal amount by the cone resonance in acoustic domain we measure flat frequency response on-axis. But amplifier noise, any distortion due to driver motor, basically all distortion after the DSP get amplified by the resonance just like before as they were not preconditioned by DSP, they get extra amplification from the breakup.
To reduce amplification of noise and distortion by the breakup one has to increase series impedance in the circuit to reduce all current through voice coil.
Whats the problem, how audible is it?
I don't know, but we can try to reason about it: distortion is usually explained to be inaudible because of the actual musical content masking it, but how is masking here? The breakup is usually left outside pass band of the particular driver, low pass crossover below breakup due to the breakup 🙂 Now, the breakup boosts any distortion out of band, imagine having brickwall filter with DSP but you'd still hear distortion products from pass band extending out of band multiple octaves with no original content on the same bandwidth masking it. So, not that much masking from the driver itself but sound from other driver (tweeter) could mask it? The other drivers are usually at different physical location, which might reduce effectiveness of the masking.
In addition there is the effect on directivity, not sure how audible this would be.
From the above we can reason that if distortion is ever audible in the first place, then this is probably the most audible occurence as it the peak is static and high Q, and thus not related to musical content but to the driver, and masking doesn't seem to be as effective as it could. Intuitively bad breakup should stick out like red flashing light if listening on-axis. But then again, I'm not sure if I can hear it.. haven't learned to hear it yet I think, if its audible at all with the volume levels I usually listen to. Haven't deviced AB test yet.
Takeaway, for best sound use driver with least breakup = smooth polar response and no distortion sticking out. If you have to work with what you've got then use passive network, parallel notch in series with voice coil, to notch the peak down. Especially with active system.
For tweeters, I'm not sure if anyone can hear the breakups past 20kHz, perhaps they do. If its bad then notch it out, or low pass, by increasing series impedance at breakup frequency just like with a woofer.
Last edited:
I thought the hypothesis was that the high frequency breakup would be energized by harmonic distortion, and then create IMD distortion interacting with lower frequency spectra? Not that the breakup is audible in and of itself.
Yeah thats the first intuition I think, and kind of the same thing, but its almost opposite: any signal with the frequency gets treated the same by the breakup, boosted /attenuated to various directions in acoustic domain. By boosting I mean the acoustic output is louder than neighbouring frequency with same current through voice coil. Current through the voice coil makes force in the motor accelerating the cone, which makes acoustic sound.
If you start thinking why there is breakup and what it actually is you'll find out that its physical property of the driver parts, size and shape and stiffness, mass and stuff like that, and the same breakup happens as long as the physical object stays the same. Only thing that would change the breakup itself is to change the physical structure or properties of the driver.
Therefore its irrelevant what audio is put through it, be it actual sound, harmonics from lower frequency fundamental, or hypothetical stick attached to the voice coil former and vibrated by vibrator; any force to the former at the breakup frequency makes the cone breakup the same, boost/attenuate acoustic output to same direction with same amount. Basically cone area splits up to regions changes, and their phases differ, which affects the directivity.
edit. well, I'm not too familiar how linear this effect is. With high excursion suspension would get stiffer at the extremes changing things and so on. Maybe the above is too much simplified, but thats how I see the thing happen, its reasoned based on data like the purifi paper with parallel notch distortion reduction technique.
If you start thinking why there is breakup and what it actually is you'll find out that its physical property of the driver parts, size and shape and stiffness, mass and stuff like that, and the same breakup happens as long as the physical object stays the same. Only thing that would change the breakup itself is to change the physical structure or properties of the driver.
Therefore its irrelevant what audio is put through it, be it actual sound, harmonics from lower frequency fundamental, or hypothetical stick attached to the voice coil former and vibrated by vibrator; any force to the former at the breakup frequency makes the cone breakup the same, boost/attenuate acoustic output to same direction with same amount. Basically cone area splits up to regions changes, and their phases differ, which affects the directivity.
edit. well, I'm not too familiar how linear this effect is. With high excursion suspension would get stiffer at the extremes changing things and so on. Maybe the above is too much simplified, but thats how I see the thing happen, its reasoned based on data like the purifi paper with parallel notch distortion reduction technique.
Last edited: