The only conceivable reason why low Qms / high damping drivers could sound worse than low loss drivers if the damping is nonlinear. If the damping is linear then the system designer can achieve the desired response based on the given Qts whether dominated by electrical or mechanical damping. This does not restrict the types of amps used. If the Qts is too low then Qe and thereby Qt can be increased by adding some series resistance in the xover, eg from the woofer inductor. This is all done by the system designer to achieve the desired response.
A very low Qts driver has so much back EMF that the motor acts like a velocity servo (coil velocity is proportional to the voltage in). This servo effect can reduce the nonlinear impact of the suspension of a frequency range around fs. this is provided of course that Bl(x) is linear.
At PURIFI we are currently studying the nonlinearity of suspension damping together with the TU of Denmark (DTU). The lab data so far show clear signs of hysteresis in typical speaker suspensions when tracing loops of force versus position - very similar to magnetic BH hysteresis loops. More findings may be published later - the project is in its initial phase still.
The driver designer may chose a high damping rubber type for the surround. This can help flatten the frequency response by damping some of the cone breakup resonances. This adds to the damping at fs also ie leads to a low Qms. SBR rubber has low damping followed by NBR and IIR having the most damping (except for some fluor type rubbers that are pretty extreme). Some cone materials have high damping (eg poly propylene) so they are less dependent on the damping from the surround.
The best strategy is to optimise the cone geometry to get a flat frequency response without needing much damping. The pesky cone edge breakup blip can mostly be avoided by careful choice of the geometry. This is best done using automated shape optimisation on a Finite Element Analaysis model to reduce the trial and error element of building prototypes. This what I have spent most of my time on the last 3 years at Purifi. Designing speaker drivers is both challenging and very fascinating.
A very low Qts driver has so much back EMF that the motor acts like a velocity servo (coil velocity is proportional to the voltage in). This servo effect can reduce the nonlinear impact of the suspension of a frequency range around fs. this is provided of course that Bl(x) is linear.
At PURIFI we are currently studying the nonlinearity of suspension damping together with the TU of Denmark (DTU). The lab data so far show clear signs of hysteresis in typical speaker suspensions when tracing loops of force versus position - very similar to magnetic BH hysteresis loops. More findings may be published later - the project is in its initial phase still.
The driver designer may chose a high damping rubber type for the surround. This can help flatten the frequency response by damping some of the cone breakup resonances. This adds to the damping at fs also ie leads to a low Qms. SBR rubber has low damping followed by NBR and IIR having the most damping (except for some fluor type rubbers that are pretty extreme). Some cone materials have high damping (eg poly propylene) so they are less dependent on the damping from the surround.
The best strategy is to optimise the cone geometry to get a flat frequency response without needing much damping. The pesky cone edge breakup blip can mostly be avoided by careful choice of the geometry. This is best done using automated shape optimisation on a Finite Element Analaysis model to reduce the trial and error element of building prototypes. This what I have spent most of my time on the last 3 years at Purifi. Designing speaker drivers is both challenging and very fascinating.
I think we discussed hysteresis before here. Along with slip-stick problems in mechanical systems. Good to read that you are researching this. As far as I know hardly any publications on this are available, but I'm not keeping track on a daily basis.
As for Daphnary's question: now we can assume you're looking for a bass-mid speaker, skip the Qms quest and go search for drivers that remain linear while making larger excursions. You won't find that info on mfg data sheets normally, but a few well known testers here publish on that.
As for Daphnary's question: now we can assume you're looking for a bass-mid speaker, skip the Qms quest and go search for drivers that remain linear while making larger excursions. You won't find that info on mfg data sheets normally, but a few well known testers here publish on that.
yes, the slip/stick hysteresis has been mentioned long time ago here. One of the models studied by DTU is exactly a friction model. They published an AES paper - will dig it out and post.
hysteresis shows up as odd harmonics that scale only slowly with amplitude - ie they don’t disappear at low levels like for a memory less nonlinearity (a cubic memory less nonlinearity gives a 3rd harmonic ratio that grows with amplitude squared whilst the odd harmonics from hysteresis grows at most by 1st order). So one should test harmonic distortion vs level down below fs where the suspension distortion dominates. higher up the system gets mass controlled and the suspension has little influence.
An interesting 🧐 observation is that we see such slow odd harmonic amplitude scaling at higher frequencies when the cone/former glue has not cured fully.
Listening indicates that the magnetic hysteresis of the motor is not doing anything good for the sound. These nonlinear effects with memory are prime suspects in general.
I imagine that we at some point can include suspension F-X loops at various amplitudes in our data sheets. We have the measurement setup.
hysteresis shows up as odd harmonics that scale only slowly with amplitude - ie they don’t disappear at low levels like for a memory less nonlinearity (a cubic memory less nonlinearity gives a 3rd harmonic ratio that grows with amplitude squared whilst the odd harmonics from hysteresis grows at most by 1st order). So one should test harmonic distortion vs level down below fs where the suspension distortion dominates. higher up the system gets mass controlled and the suspension has little influence.
An interesting 🧐 observation is that we see such slow odd harmonic amplitude scaling at higher frequencies when the cone/former glue has not cured fully.
Listening indicates that the magnetic hysteresis of the motor is not doing anything good for the sound. These nonlinear effects with memory are prime suspects in general.
I imagine that we at some point can include suspension F-X loops at various amplitudes in our data sheets. We have the measurement setup.
here is the AES paper on the friction model: https://www.aes.org/e-lib/browse.cfm?elib=21494
Hi Lars,
There's some interesting claims of surround-less drivers with X-I curve. Love to hear your idea/comments on it!
https://6moons.com/audioreview_articles/ilumnia2/5/
Best,
hi, thank you for the link. the magnetic suspension is of course very interesting in the light of the hysteresis issues of rubber materials. However, linearising X(I) is only making the driver linear at DC but not dynamically. Without a constant Bl(x) we get AM modulation of the midrange (IMD). The linear X(I) is only working well below fs of the driver and this is where distortion typically is the least audible.
Making the inductance L(x) constant is good and also a cornerstone of PURIFI drivers.
cheers, Lars
Yeah when reading the page I think they mixed up some concepts(such as symmetrical == linear) and the idea of (linear X-I curve == no IMD) seems wrong above fs when BL(x) is not flat.
But removing mechanical compliance devices seems to be a really interesting idea. Don't know how much is the trade off. And I think their driver won't work when installed horizontally.
But removing mechanical compliance devices seems to be a really interesting idea. Don't know how much is the trade off. And I think their driver won't work when installed horizontally.
I agree, the symmetry=linear is a very widespred misconception. In this case, the I(X) is drawn as a symmetric V shaped graph and explained as linear. That is indeed taking the misconception a step further.
The point about Doppler is interesting but the question remains what reality is when taking the reflected sound from the room into account.
But the magnetic suspension is indeed interesting - getting rid of mechanical hysteresis and Sd modulation at the same time. Not sure how the cone edge breakup is handled then though. Would be nice with some measurements.
The point about Doppler is interesting but the question remains what reality is when taking the reflected sound from the room into account.
But the magnetic suspension is indeed interesting - getting rid of mechanical hysteresis and Sd modulation at the same time. Not sure how the cone edge breakup is handled then though. Would be nice with some measurements.
Possibly relevant
I have been playing with a pair of SB Audience ROSSO 12MW300 and was noticing some strange and unpleasant harshness/grain/distortion.
It is a familiar sound, I experienced the same with Peerless P830668 to a lesser degree, I was sure this was was cone breakup but started become doubtful due to fact it seem to present over the entire drivers bandwidth, worse at lower end if anything.
These are 12" vs 10" peerless and lack a shorting ring but still have significantly lower inductance than the Peerless and lower Mms. They are designed as Pro sound mid-bass too. I dont expect miracles but it does not seem right for this driver to begin to shows signs of struggling at e.g 200Hz but not be much worse at 1kHz or even higher.
After reading some of the thread my attentiont turned to the efficiency and Qms, both significantly higher for ROSSO, Peerless still significantly higher Qms than the other full range drivers I can compare to.
I tried adding output resistors to an LM3886 based amp, practically the only amp I've used for my DIY speakers since the beginning (with some small changes and optimisations made to it over time).
Interestingly adding around 4 ohms output resistance really helps to reduce this unpleasant quality, while the sound does become softer and rounder.
It will be interesting to see if zero/low negative feedback design can mitigate this problem while maintaining lowish z-out, or it's specifically low damping factor/current drive.
I have been playing with a pair of SB Audience ROSSO 12MW300 and was noticing some strange and unpleasant harshness/grain/distortion.
It is a familiar sound, I experienced the same with Peerless P830668 to a lesser degree, I was sure this was was cone breakup but started become doubtful due to fact it seem to present over the entire drivers bandwidth, worse at lower end if anything.
These are 12" vs 10" peerless and lack a shorting ring but still have significantly lower inductance than the Peerless and lower Mms. They are designed as Pro sound mid-bass too. I dont expect miracles but it does not seem right for this driver to begin to shows signs of struggling at e.g 200Hz but not be much worse at 1kHz or even higher.
After reading some of the thread my attentiont turned to the efficiency and Qms, both significantly higher for ROSSO, Peerless still significantly higher Qms than the other full range drivers I can compare to.
I tried adding output resistors to an LM3886 based amp, practically the only amp I've used for my DIY speakers since the beginning (with some small changes and optimisations made to it over time).
Interestingly adding around 4 ohms output resistance really helps to reduce this unpleasant quality, while the sound does become softer and rounder.
It will be interesting to see if zero/low negative feedback design can mitigate this problem while maintaining lowish z-out, or it's specifically low damping factor/current drive.
adding a series resistance Rs to the driver has two main effects:
1) Qes is increased proportionally with Re+Rs, ie is multiplied with (Re+Rs)/Re. this changes the bass alignment substantially.
2) The magnetic hysteresis distortion of the coil and surrounding iron is reduced by the same factor as in 1) since it is a step towards current drive (hysteresis distortion is effectively a voltage source in series with the driver and its effect on the drive current vanishes in when the current is controlled)
1) Qes is increased proportionally with Re+Rs, ie is multiplied with (Re+Rs)/Re. this changes the bass alignment substantially.
2) The magnetic hysteresis distortion of the coil and surrounding iron is reduced by the same factor as in 1) since it is a step towards current drive (hysteresis distortion is effectively a voltage source in series with the driver and its effect on the drive current vanishes in when the current is controlled)
Wise words.
As an aside, I always have to chuckle when I read "low loss rubber surround" in the specsheet of a typical hifi midwoofer and then notice that the sensitivity is specified around 85 dB.
Last edited:
Rms determines the Qms (Q value of the main fs resonance in free air with temrinals open) and this affects Qts which is the Q value when the terminals are shortted (eg by an amplifier) - agin in free air. Qts is the ratio of sensitivity at fs to the sensitivity at the midband. In conclusion, Qts is normally completely dominated by the electrical damping (Qms>>Qes) so Rms influence on Qts and thus sensitivity at fs is small. Moreover, Rms does not affect the midband sensitivity at all, only around fs.
Why? Those hifi drivers need that "low loss rubber surround", because
Otherwise yeah, low loss rubber surround for low mechanical losses combined with a conductive voice coil former that creates eddy currents that manifests as high mechanical resistance (relative low Qms and high Rms) in the T/S parameters, SEAS drivers with alu VC formers as example. Brilliant!
Last edited:
alu vc former is the best?Why? Those hifi drivers need that "low loss rubber surround", because(of marketing)they are usually low sensitivity due the relative weak motor compared to cone size and moving mass, every 0.1dB counts.
Otherwise yeah, low loss rubber surround for low mechanical losses combined with a conductive voice coil former that creates eddy currents that manifests as high mechanical resistance (relative low Qms and high Rms) in the T/S parameters, SEAS drivers with alu VC formers as example. Brilliant!