Importance of driver slew rate and how to define it
An amplifier's speed is defined by the parameter "slew rate". It is measured in volts per microsecond. The figure varies from amp to amp, but for home audio amplifiers it usually ranges from about 10-100 volts per microsecond. Older amps are on the lower end of the range and are considered slower.
The higher the frequency that needs to be amplified, the higher the minimum slew rate must be.
The higher the amplitude of the highest frequency, the higher the minimum slew rate must be.
It's not enough to be only fast enough... For example, CDs sample rate is 2x the maximum reproduced frequency (disregarding filter). This is done to cover the 2 directions of movement in one cycle. It is not good enough for an amplifier to be able to raise and lower its voltage fast enough to match maximum frequency at the maximum amplitude. If a sine wave were applied to the input, the waveform on the output would be a triangle - not good. Amps slew rate should be at least 10x this amount to be considered acceptable. But the faster the better. Reaching the point when the waveform is not distorted is the goal. Amps don't have the advantage of CD players, which, when outputting their highest frequency, take 5 points and interpolate a sine wave (one of the reasons CDs sound unnatural).
If the slew rate is too low, lower frequency transients are too quiet and high frequencies are clipped to lower amplitude triangle-shaped versions of the input - it doesn't sound good.
I think a woofer's slew rate can be measured by applying tone bursts of increasing amplitude until compression unrelated to voice coil thermals results. Put another way, this is when the cone is lagging behind the signal because EM coupling is too weak. BI is a factor, but not the only one.
Things like BI, Mms, radiating area, frequency, and probably others would need to be used to calculate it. I don't know enough about the physics of it all to come up with exactly how, but if we could have a discussion about it, collaborate, we could come up with a new driver parameter which defines transient response. The higher its value, the more true to the input its output. Thoughts?
I think the frequency that is used should depend on the driver's size, cone diameter specifically. Since the measurement needs to stay away from xmax and beaming, I suggest wavelength 1.5x cone diameter. So 600hz for a 5 inch, 300hz for 10 inch, 200hz for 15 inch. This way up to 130db can be tested with 4mm xmax (basically every driver is compatible)
An amplifier's speed is defined by the parameter "slew rate". It is measured in volts per microsecond. The figure varies from amp to amp, but for home audio amplifiers it usually ranges from about 10-100 volts per microsecond. Older amps are on the lower end of the range and are considered slower.
The higher the frequency that needs to be amplified, the higher the minimum slew rate must be.
The higher the amplitude of the highest frequency, the higher the minimum slew rate must be.
It's not enough to be only fast enough... For example, CDs sample rate is 2x the maximum reproduced frequency (disregarding filter). This is done to cover the 2 directions of movement in one cycle. It is not good enough for an amplifier to be able to raise and lower its voltage fast enough to match maximum frequency at the maximum amplitude. If a sine wave were applied to the input, the waveform on the output would be a triangle - not good. Amps slew rate should be at least 10x this amount to be considered acceptable. But the faster the better. Reaching the point when the waveform is not distorted is the goal. Amps don't have the advantage of CD players, which, when outputting their highest frequency, take 5 points and interpolate a sine wave (one of the reasons CDs sound unnatural).
If the slew rate is too low, lower frequency transients are too quiet and high frequencies are clipped to lower amplitude triangle-shaped versions of the input - it doesn't sound good.
I think a woofer's slew rate can be measured by applying tone bursts of increasing amplitude until compression unrelated to voice coil thermals results. Put another way, this is when the cone is lagging behind the signal because EM coupling is too weak. BI is a factor, but not the only one.
Things like BI, Mms, radiating area, frequency, and probably others would need to be used to calculate it. I don't know enough about the physics of it all to come up with exactly how, but if we could have a discussion about it, collaborate, we could come up with a new driver parameter which defines transient response. The higher its value, the more true to the input its output. Thoughts?
I think the frequency that is used should depend on the driver's size, cone diameter specifically. Since the measurement needs to stay away from xmax and beaming, I suggest wavelength 1.5x cone diameter. So 600hz for a 5 inch, 300hz for 10 inch, 200hz for 15 inch. This way up to 130db can be tested with 4mm xmax (basically every driver is compatible)
Last edited:
Slew rate is defined by upper frequency limit, so its basically a measure of frequency response.
A lowly NE5532 opamp has the slew rate capability to surpass the upper cutoff of just about any electromagnetic driver. Amplifiers are usually limited in HF power response by the output zobel. If you've ever tried to bench test a power amp into a dummy load with a sine wave higher than 20 KHz, the resistor in the zobel will overheat and smoke very quickly, so the higher voltage swings of ultrasonic signals won't even make it to the speakers - typical music doesn't contain that much concentration of power in those higher frequencies, so its not necessary to be able to have full power output past 20 KHz. Yes, there are trnaisents up in those higher frequencies, nothing near that of lower frequencies.
So basically speaking, amplifier slew rate is not as important as some people make it to be. It can affect IMD, but thats a whole different conversation.
A lowly NE5532 opamp has the slew rate capability to surpass the upper cutoff of just about any electromagnetic driver. Amplifiers are usually limited in HF power response by the output zobel. If you've ever tried to bench test a power amp into a dummy load with a sine wave higher than 20 KHz, the resistor in the zobel will overheat and smoke very quickly, so the higher voltage swings of ultrasonic signals won't even make it to the speakers - typical music doesn't contain that much concentration of power in those higher frequencies, so its not necessary to be able to have full power output past 20 KHz. Yes, there are trnaisents up in those higher frequencies, nothing near that of lower frequencies.
So basically speaking, amplifier slew rate is not as important as some people make it to be. It can affect IMD, but thats a whole different conversation.
You didn't read the whole post lol
Edit: it's about measuring or calculating the transient response of woofers. Creating a new parameter to quantify it, using things like BI, Mms, and measurements. Finding its slew rate. Specifically by finding at which amplitude compression not related to voice coil thermals begins
Edit 2: your first sentence should read the maximum frequency at a given amplitude is described by the slew rate
Edit: it's about measuring or calculating the transient response of woofers. Creating a new parameter to quantify it, using things like BI, Mms, and measurements. Finding its slew rate. Specifically by finding at which amplitude compression not related to voice coil thermals begins
Edit 2: your first sentence should read the maximum frequency at a given amplitude is described by the slew rate
Last edited:
Yeah sorry, I thought it was mostly about amp slew rate. I was comparing amps to speakers, so I thought it was relevant.
One would figure it to be a direct relationship of BL to Mms, as well as the enclosure (type) itself.
One would figure it to be a direct relationship of BL to Mms, as well as the enclosure (type) itself.
It's intended to be away from resonance so the enclosure doesn't have much effect on it. Q could be used to calculate or estimate transient response down there
Don’t worry about slew rate. It is an irrelevant derivative. Even your thoughts about the importance of slew rate at 20kHz are off the track.
Care to elaborate on why it's not relevant? (my description of slew rate is there for ease of reference, to be applied to the transient response of a direct radiator (which is not as fast as an amp)
Last edited:
Taking the enclosure out of the equation only affecting system Q, you're only left with BL/Mms and Le, so in theory those designate speaker "slew rate".
The big misconception is that larger drivers can't do "fast" bass. Many larger diameter, low moving mass drivers have flimsy cones that deform at higher output levels and behave in a very non linear fashion, severely distorting while doing so. Having a low mass, stiff cone is important but a design compromise always has to be made. So basically, if the cone doesn't behave pistonic, it doesnt matter how high the BL/Mms ratio is. The compliance of the enclosure air volume has an impact on this too and the air load can further stress various part of the cone, dust cap and surround, further affecting transient response through distortion.
VC inductance (Le) has an affect on distortion and consequently transient response. Obviously the higher Le is, the faster the FR rolls off, but also the greater Le will vary as the VC excursion increases, even if its still within Xmax. An underhung VC generally provides lower Le and better Le linearity, but the tradeoff is much greater fluctuation of Le getting past Xmax.
So in theory, BL, Mms and Le are the main things to consider, but all of the distortion mechanisms such as suspension compliance linearity, enclosure panel resonances, VC inductance linearity, as well as cone deformation and breakup all affect transient response.
The big misconception is that larger drivers can't do "fast" bass. Many larger diameter, low moving mass drivers have flimsy cones that deform at higher output levels and behave in a very non linear fashion, severely distorting while doing so. Having a low mass, stiff cone is important but a design compromise always has to be made. So basically, if the cone doesn't behave pistonic, it doesnt matter how high the BL/Mms ratio is. The compliance of the enclosure air volume has an impact on this too and the air load can further stress various part of the cone, dust cap and surround, further affecting transient response through distortion.
VC inductance (Le) has an affect on distortion and consequently transient response. Obviously the higher Le is, the faster the FR rolls off, but also the greater Le will vary as the VC excursion increases, even if its still within Xmax. An underhung VC generally provides lower Le and better Le linearity, but the tradeoff is much greater fluctuation of Le getting past Xmax.
So in theory, BL, Mms and Le are the main things to consider, but all of the distortion mechanisms such as suspension compliance linearity, enclosure panel resonances, VC inductance linearity, as well as cone deformation and breakup all affect transient response.
Last edited:
For a mechanic-acoustic device like a loudspeaker, the bandpass function of the acoustic output is all that matters in this respect. And if it doesn’t do 20k at the desired level, use a tweeter along with it. There is your slew rate. The simple fact there is the desired output at the desired frequency is proof of the pudding.
I think I need to buy a couple different drivers with the same diameter, Mms, Le, and maybe Cms, with different BI. Feed them tone bursts of increasing amplitude and see how they do
We don't get power response in datasheets, or measurements at 1w, 10w, and 100w (with LF limited to xmax), so we don't really know if the drivers make the desired outputs at the desired amplitudes
For typical domestic uses power compression isnt really much of an issue. It becomes a problem and greater influencing factor with very low efficiency subwoofer drivers in small sealed boxes and very high power PA drivers that are pushed close to their electrical limits for longer durations.
The average high 80 to low 90 dB hifi drivers don't show much power compression unless you require continuous 100+ dB levels from them. Sure, the VC Re increases with heat, which can change alot of the electrical parameters, but for the most part the driver will cope with short transients and you'll likely exceed Xmax well before getting the coil hot enough to make an audible change. Even drivers with smaller VCs (around 1") have no audible problems with louder transients, especially if the VC is wound on an aluminum former, well ventilated and playing in a ported enclosure. Drivers in small sealed enclosures (midbass and midranges crossed over very low) can suffer VC ventilation and heating problems when pushed hard for longer amounts of time.
The average high 80 to low 90 dB hifi drivers don't show much power compression unless you require continuous 100+ dB levels from them. Sure, the VC Re increases with heat, which can change alot of the electrical parameters, but for the most part the driver will cope with short transients and you'll likely exceed Xmax well before getting the coil hot enough to make an audible change. Even drivers with smaller VCs (around 1") have no audible problems with louder transients, especially if the VC is wound on an aluminum former, well ventilated and playing in a ported enclosure. Drivers in small sealed enclosures (midbass and midranges crossed over very low) can suffer VC ventilation and heating problems when pushed hard for longer amounts of time.
I get what you mean about the thermal compression, but that's not what I'm referring to.
The more detailed a speaker is, the more closely the movement of its cone is following the input. To do this, a minimum motor strength is required. The higher the output, the more strength is required to maintain the same level of detail (control)
Detail is lost when a driver's slew rate is too low, when the cone isn't following the input closely enough. Peaks are rounded- the higher the peak, the more severely it's rounded. Because different drivers with almost identical frequency responses are capable of huge differences in clarity, I'm willing to bet driver slew rate is an important and overlooked specification.
I'm quite sure determining driver slew rates and comparing them to the requirements for what's needed to accomplish accurate reproduction at various levels is what needs to be done. The ability of a voice coil former to resist stretching and compressing is probably important too, as it could absorb these micro movents.
As I outlined in my first post, I think short tone bursts, increasing in amplitude until measurable compression, is the best way to determine the slew rate. Slew rate doesn't even need to be reduced to a speed - if it's too difficult or an unnecessary amount of work, it could just be a specification like 1db attenuation at 600hz occurs at 102db (101 measured from 102 input), 3db at 108 (105). I've never measured a driver like this before, but I think these numbers could be realistic for a surprising number of drivers
Edit: even if a driver only compresses a transient by an extra 0.5db compared to an almost ideal driver, the effect on perceived sound quality by a listener could be significant. Myself, I couldn't tell you if someone lowered 500hz on my eq by 0.5db immediately, but after 15 seconds, if I was familiar with the song, I'd know. So the sound of a snare, repeating through a song, seems likely to be noticable
The more detailed a speaker is, the more closely the movement of its cone is following the input. To do this, a minimum motor strength is required. The higher the output, the more strength is required to maintain the same level of detail (control)
Detail is lost when a driver's slew rate is too low, when the cone isn't following the input closely enough. Peaks are rounded- the higher the peak, the more severely it's rounded. Because different drivers with almost identical frequency responses are capable of huge differences in clarity, I'm willing to bet driver slew rate is an important and overlooked specification.
I'm quite sure determining driver slew rates and comparing them to the requirements for what's needed to accomplish accurate reproduction at various levels is what needs to be done. The ability of a voice coil former to resist stretching and compressing is probably important too, as it could absorb these micro movents.
As I outlined in my first post, I think short tone bursts, increasing in amplitude until measurable compression, is the best way to determine the slew rate. Slew rate doesn't even need to be reduced to a speed - if it's too difficult or an unnecessary amount of work, it could just be a specification like 1db attenuation at 600hz occurs at 102db (101 measured from 102 input), 3db at 108 (105). I've never measured a driver like this before, but I think these numbers could be realistic for a surprising number of drivers
Edit: even if a driver only compresses a transient by an extra 0.5db compared to an almost ideal driver, the effect on perceived sound quality by a listener could be significant. Myself, I couldn't tell you if someone lowered 500hz on my eq by 0.5db immediately, but after 15 seconds, if I was familiar with the song, I'd know. So the sound of a snare, repeating through a song, seems likely to be noticable
Last edited:
It is a common misconception that 'transient response' means how fast a driver is so that it can keep up with what it's asked to do.. and that some drivers are fast and some are slow.
Yes, that's the descriptor I chose for slew rate, as slew rate isn't a recognized term to describe loudspeakers.
Whenever a driver is moving, it's moving at the speed of sound. Just not necessarily with the signal. Maybe that's not the best way to put it, but you know what I mean
Whenever a driver is moving, it's moving at the speed of sound. Just not necessarily with the signal. Maybe that's not the best way to put it, but you know what I mean
Last edited:
An analogy of sorts:
A sealed box enclosure works up until the cone's displacement makes up more than a few percent of the box volume. Then the air suspension becomes non-linear, creating distortion - amplitude modulation. At the peaks there is more resistance, so spl falls.
The applied power to a coil surrounded by a permanent magnet results in a (mostly) linear reaction up to a point. The point that it doesn't and non-linear distortion creeps in: amplitude modulation at the peaks
A sealed box enclosure works up until the cone's displacement makes up more than a few percent of the box volume. Then the air suspension becomes non-linear, creating distortion - amplitude modulation. At the peaks there is more resistance, so spl falls.
The applied power to a coil surrounded by a permanent magnet results in a (mostly) linear reaction up to a point. The point that it doesn't and non-linear distortion creeps in: amplitude modulation at the peaks
Last edited:
The concept is flawed, it's not a matter of anyones choice of words. Transient response is simply what happens when you perturb the driver. Transient response is not a graded scale showing which driver is faster or better.
A speaker cone can't be allowed to reach the speed of sound.
Every driver has some amount of non-linearity and that in itself is very hard to quantify through simple parameters alone. It has to be measured by applying varying amplitudes of tone bursts at various frequencies within its bandwidth and power rating. Dynaudio used to publish this type of measurement data but only at specific frequencies, so the tests are sort of inconclusive and likely done for marketing hype.
Being that all electromagnetic drivers have some conductive resistance and temp coefficient, there will always be some sort of loss and non linearity involved. A driver which has an efficiency of 1% will produce 92 dB per 1 watt of input power, so the remaining 99% gets turned into heat, which just makes things worse as the coil continues to heat up if the heat can't be dissipated quick enough to avoid further heating.
Voice coil formers have quite an influence on motion transfer to the cone, as do the semi flexible adhesives used. VCs can also have resonant modes internally. Also, aluminum formers dampen the their own movements as opposed to fiberglass, titanium, kapton and nomex formers. I generally prefer non damping VC formers and think they tend to promote more audio detail retrieval. Titanium is a very good material for a VC former and I don't think it's presence can be measured specifically in any of the electrical driver parameters.
I would think that higher power distortion measurements would be sufficient in showing a driver's capability of accurate dynamic reproduction. When looking at a driver's distortion measurements, H2 will increase somewhat continuously as power levels go up, whereas H3 will stay at a more constant level and won't significantly increase up to a certain power level. When H3 does start increasing, that will show the linearity limits of the driver, specifically at lower frequencies where there is more cone excursion.
Being that all electromagnetic drivers have some conductive resistance and temp coefficient, there will always be some sort of loss and non linearity involved. A driver which has an efficiency of 1% will produce 92 dB per 1 watt of input power, so the remaining 99% gets turned into heat, which just makes things worse as the coil continues to heat up if the heat can't be dissipated quick enough to avoid further heating.
Voice coil formers have quite an influence on motion transfer to the cone, as do the semi flexible adhesives used. VCs can also have resonant modes internally. Also, aluminum formers dampen the their own movements as opposed to fiberglass, titanium, kapton and nomex formers. I generally prefer non damping VC formers and think they tend to promote more audio detail retrieval. Titanium is a very good material for a VC former and I don't think it's presence can be measured specifically in any of the electrical driver parameters.
I would think that higher power distortion measurements would be sufficient in showing a driver's capability of accurate dynamic reproduction. When looking at a driver's distortion measurements, H2 will increase somewhat continuously as power levels go up, whereas H3 will stay at a more constant level and won't significantly increase up to a certain power level. When H3 does start increasing, that will show the linearity limits of the driver, specifically at lower frequencies where there is more cone excursion.
One last shot at TS, then I’m off. Are you familiar with the Dirac pulse and how it mathematically can be described by an infinite number of discrete sinoidal functions? In other words: any normal to fairly high level swept sine or wideband noise reproduced will tell you ‘the speed’ of the DUT? And that is what we do all the time when measuring a speaker.
The use of tone bursts is moot. A tone burst is in no way a sinoidal signal. Run tonebursts through a spectral analyser and you will see.
The use of tone bursts is moot. A tone burst is in no way a sinoidal signal. Run tonebursts through a spectral analyser and you will see.
