I wonder what would happen if the input signal was band-limited by a very steep low-pass filter so that it completely avoided exciting the higher-frequency resonances. Would the "metal or hard cone character" still shine through?
Oh boy, a few things to react to, but that's good.
Well, that is the usual solution, say 2-3KHz and make it 4th order and problem solved. Accept to my ears, and many others, that is not the best solution wise. I have demonstrated this to some friends of mine and the typical LR-4 solution definitely works in a technical sense, but the sound is smooth and boring. But it just shows what we always knew, that non LR-4 solutions require good driver selections.
Hear, hear! Sometimes 'resonances' are very useful and even good. Like you hinted at, sealed and vented box alignments have controlled resonances that functions as mechanical high-pass filters. No resonance, not bass. 🤔
Hi Boden. Let me try.
As I said, I had seen the 'effect' when discussing with Lynn Olson and I didn't think it in terms of it being a horn. So when Esa's book on "Current-Drive" I was amongst the first to buy it and get a discussion going on the vendor thread he started.
Part of his book is reproduced here with his permission:
chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://asiakas.kotisivukone.com/files/currentdrive.ota.fi/tiedostot/sample2.pdf
On Page 48 of the above link, see below. Read it slowly and carefully), there is a lot going on.
Note the reference to the flare above the reduced area of the cone ideally should cause the centre of the cone nearest the voice coil and then the outer cone will become stationary. That means that the out cone becomes a horn for the inner and closer part to the voice coil part. What he says makes sense. In our discussions we came up with the fact that we had differences. One is that I am into finding ways to achieve current drive based on a ratio that he agreed with and it is 5:1 - this ratio is not a magic number, but if you can isolate Ei (inductive) part of the voice coil by 5:1, then you have achieved the minimum requirement for current-drive (and its benefits). I can do that without resorting to current-drive. He says the only way it can be achieved is by using current-drive. But there is agreement on what current-drive is suppose to achieve. And I credit him with alerting me to the fact that what I had been studying was in fact a 'horn effect' and I give him credit for that.
A cone that has very much less horn effect is the ULD driver we use in Elsinore ULD:
Much shallower cone and it has an additional feature, behind the cone and around its perimeter, it has a mass loading ring. Applying Newton's Laws we now means a significant reduction of velocity around the cone's edge as we approach the horn effective frequencies. This reduces the radiation surface area above 1KHz. This further reduces the horn effect but you can still some of it and there is then the drop as you would expect. This mass loading by Purifi must have been something very deliberate. This is of course a 'soft' fibre cone.
Well, that is the usual solution, say 2-3KHz and make it 4th order and problem solved. Accept to my ears, and many others, that is not the best solution wise. I have demonstrated this to some friends of mine and the typical LR-4 solution definitely works in a technical sense, but the sound is smooth and boring. But it just shows what we always knew, that non LR-4 solutions require good driver selections.
Hear, hear! Sometimes 'resonances' are very useful and even good. Like you hinted at, sealed and vented box alignments have controlled resonances that functions as mechanical high-pass filters. No resonance, not bass. 🤔
Hi Boden. Let me try.
As I said, I had seen the 'effect' when discussing with Lynn Olson and I didn't think it in terms of it being a horn. So when Esa's book on "Current-Drive" I was amongst the first to buy it and get a discussion going on the vendor thread he started.
Part of his book is reproduced here with his permission:
chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://asiakas.kotisivukone.com/files/currentdrive.ota.fi/tiedostot/sample2.pdf
On Page 48 of the above link, see below. Read it slowly and carefully), there is a lot going on.
Note the reference to the flare above the reduced area of the cone ideally should cause the centre of the cone nearest the voice coil and then the outer cone will become stationary. That means that the out cone becomes a horn for the inner and closer part to the voice coil part. What he says makes sense. In our discussions we came up with the fact that we had differences. One is that I am into finding ways to achieve current drive based on a ratio that he agreed with and it is 5:1 - this ratio is not a magic number, but if you can isolate Ei (inductive) part of the voice coil by 5:1, then you have achieved the minimum requirement for current-drive (and its benefits). I can do that without resorting to current-drive. He says the only way it can be achieved is by using current-drive. But there is agreement on what current-drive is suppose to achieve. And I credit him with alerting me to the fact that what I had been studying was in fact a 'horn effect' and I give him credit for that.
A cone that has very much less horn effect is the ULD driver we use in Elsinore ULD:
Much shallower cone and it has an additional feature, behind the cone and around its perimeter, it has a mass loading ring. Applying Newton's Laws we now means a significant reduction of velocity around the cone's edge as we approach the horn effective frequencies. This reduces the radiation surface area above 1KHz. This further reduces the horn effect but you can still some of it and there is then the drop as you would expect. This mass loading by Purifi must have been something very deliberate. This is of course a 'soft' fibre cone.
Last edited:
Exactly. And they don't achieve it the cheap way which is high inductance to suppress it. I have seen many decide on the wrong driver because they were not looking at the right thing.
I was thinking the same so I thought I'd normalise the polar plots Joe put the "X" on (there's little interpolation in there too).
It could be a number of things. I would want to see data with a higher resolution.
Joe, after reading the link you provided, do you believe Esa is advocating for using a driver in this aforementioned upper region?
No, he is not. Definitely not. Rather, since his intentions are to use current-drive, he is talking about how it introduces difficulties. He is making general observations about the 'operation' of drivers using current-drive.
You have to understand that some of it is very awkward English. He has improved a lot since he wrote the book. I gather you don't have a copy? In his favour I think it is an 'important' book, but he knows that I don't believe in current-drive as such (and there are other conclusions I don't agree with), but with maybe one exception, I think current-drive may suit active speakers design and there are several models on the market.
Below is the continuation after the clip I posted earlier. It should give you better context of the previous clip.
So he is actually complaining about that horn effect "accentuation" making it a problem when current-driving.
I see the horn effect as somewhat benign, if that is all that it is. If only an acoustic effect, a kind of internal diffraction of the driver, not assoaciated with any nasty (note qualifier) resonances.
I have developed a way of dealing with it. If the problem is only simple one, then I usually will usually design for the off axis (usually 15°) which takes off a big chunk of the 'accentuation' and certainly by 20°. Since I am looking for a driver where I can use a single inductor, this horn effect may even make it able to use a bigger value inductor. May is the operative word here. That kink I referred earlier may also come into play and prevent the inductor value getting higher and the kink may become to accentuated. But it may lead to a considered BBC dip. So it is a balance trick.
Above 15° off axis.
Above, family of On, 15° and 30° off axis. Use toe-in of the speaker to get the FR balance you need.
My reason for doing the above, and it takes a leap of faith to do it that way, is that I want a large series inductor do do current-drive like things.
Then I finish it off by EQ'ing the current seen by the amplifier.
Allen, please, trust me on this, the benefits is in the sound.
Doing it with LR4 crossovers (so easy to do in SoundEasy) and you get nowhere as good sound. I believe that I am getting the sound of current-drive, but without the downsides and a speaker that is totally amplifier friendly. Could it be that this is about making the amplifier sound better because you are isolating the amplifier from the speaker, and particularly those Ei current modulations that pollute/corrupt the force-factor that is related to the current of the amplifier. That is bad for the design and hence designing away from that results in better sound.
If I may quote Esa again, this time by typing from the book here on page 73, on why the benefits of current-drive is quite real - but also why my approach also works:
"The BL variation due to large displacements causes always in the motional EMF and consequently in the total impedance aberration that shows up, in the mid-frequency region, mostly as modulation of the impedance angle in co0ngruence with displacement. On voltage drive, this modulation is transferred directly to the reproduced signal, giving rise to phase noise, or jitter, and thus adding significantly to nonlinearity during loud bass tones."
Lars Risbo and Bruno Putzeys says this large displacement causes AM distortion (not FM).
The source is Ei impedance modulations. But Esa here has overlooked something: the driver's Ei impedance modulations can also be caused by other things not related to "large displacement" and in 2016 he took note of that. More on that below. The problem is not just related to excursions/displacement at LF.
Back to page 73, Esa continues:
"Instead, on current-drive, this interference is conveyed only to the driver's terminal voltage where it is of no harm [he is saying "no harm" to the voltage, but not as a good thing, a deceptively bad thing - award English makes this unclear]; and hence, nonlinearities are limited to the distortion that stems directly from BL variation itself."
Ironically, they key to the same benefits with either current-drive or voltage-drive are the same:
Driver isolation from the amplifier at critical mid-frequencies (and to a degree also in the tweeter). This can also be achieved in voltage-drive by using a large series inductor and then EQ the current of the amplifier. Ei modulations at mid-frequencies are further isolated that way. Similarly with the tweeter, by using a waveguide and not lowering the crossover, and we get a much smaller value series capacitor. See the pattern? Also some series resistance with the tweeter's coil is useful. Add an LC trap at tweeter Fs to lower distortion, finally EQ the sum current of all of it.
And I do believe that in one post I have covered the fundamental design approach of Elsinore speakers. I didn't set out that way, but it worked out that way.
Later, after Esa wrote his book, he wrote another article in 2016 and also an important one, as it included actual distortion measurements with a microphone, one that is well documented and repeatable by others. In this PDF article he points to multi-tone measurements where there are no large excursions and there are still benefits in current-drive and my inductor drive as used in the Elsinores, also has those benefits. Even when there are no large excursions, the driver's own deficiencies are fed back into the amplifier as Ei modulations.
These are electro-acoustic-mechanical deficiencies of the driver. The amplifier now has to deal with that. In all cases, what improves sound is isolation from those Ei modulations.
"Comparative Measurements on Loudspeaker Distortion"
I have attached it below. This is first time I have tried to attach a PDF file.
I note that here on diyaudio.com some have promoted resistive drive. Of course this approach drops the power of the amplifier to less than 25%.
But it seems that we have three:
1. Voltage-drive.
2. Current-drive.
3. Inductor-drive
4. Resistive-drive
I am for number 3, Inductor-drive, as the best solution. But there is a version of resistor-drive that has not occurred to them.
Whew, a lot of dense stuff above.
PS: If Esa's sees this post, it would be interesting...?
Last edited:
Just found this picture I must have taken as I found it on my computer. It is interesting as it clearly shows that mass loading 'skirt' around the perimeter of the cone. BTW, the cone is a lot heavier than other drivers and yet achieve good sensitivity (and efficiency) because of a very power motor (magnetic circuit) than other drivers.
Do we know this? Such a device could conceivably be used to brace against a non-axisymmetric compliance of the surround?
Possibly. I do think that it did more than one thing, so I would not argue about that. I know that they relied on computer programming because of the number of variables. Change one thing and you change others, some even unpredictably.
You are excluding the contribution of the rubber surround. The surround makes up 20% + of the area of the driver. The surround is a large soft floppy piece of rubber.
[Above is our MFC polycone and SB's measurement.]
See that "X" and I see one behaviour below and above that frequency. Note that dip? Very common, especially as this driver has very low inductance. The cone may be pistonic up to 1KHz or slightly above, but that kink is not 6dB down means that the cone has already started to flex and bend modes are happening (in theory the radiating area should reduce). I think this is what Allen says that it does not mean distortion even if the cone is not longer pistonic. Or else by that kink the response would 6dB down (Allen knows about Roy Allison who said a totally stiff cone would roll off 1st order). So in the MFC cone I see that it is predictable up to that 3KHz kink and ideally it should roll of controlled above 3KHz.
There is much interaction between the rubber surround and the stiffer cone.
If you are claiming adhortative insight into the behavior of the driver include the contribution of the surround.
Thanks DT
True. In the past I have made this point myself.
But the Purifi driver is the subject and I am sure you know that is quite different. It does not have an ordinary surround. It goes close to sum zero, hence 0% and not 20%. So that mass skirt undoubtably work together with what Purifi calls their "funky surround."
Adhortative? That's not a common word, but no, I don't think I was being that.
Uncovering more about breakup from Esa Meriläinen's text..
He refers to 'phase', where the whole cone is breaking up and taking part in the resonance, perhaps with the edge bending back and forth out of phase with the middle. The primary issue there being inconsistent polars..
He is talking about two things though, the second being 'some disconnect' of the outer cone.
Esa's text leading up to that states Figure 3.12a shows results using a current source, so the top end is electrically inflated relative to the common case.
..and implicates the wider radiation determined by cone behaviour is one factor he uses to determine the upper usable frequency.
He refers to 'phase', where the whole cone is breaking up and taking part in the resonance, perhaps with the edge bending back and forth out of phase with the middle. The primary issue there being inconsistent polars..
He is talking about two things though, the second being 'some disconnect' of the outer cone.
Esa's text leading up to that states Figure 3.12a shows results using a current source, so the top end is electrically inflated relative to the common case.
..and implicates the wider radiation determined by cone behaviour is one factor he uses to determine the upper usable frequency.
Last edited:
Hello,
Above you post a photo of a Purify driver and a frequency response plot for a SB Acoustics SB17MFC35-8 6" Poly Cone Woofer.
You went on at length about the "[Above is our MFC polycone and SB's measurement.]".
So no the subject was not about the Purify "funky surround."
I agree that there is more going on than cone resonance. At lower frequencies Cone Resonance is likely minimal. At higher frequencies there is likely a lot going on with the resonances of the soft floppy rubber surround. I can see the rubber surround showing breakup long before the much stiffer cone.
Thanks DT
Perhaps, but sometimes two things can both be right? As we get from the inner cone we have low mass that, as we get to the outer side of the cone, the mass also increases. That means that the outer part of the cone will deaccelerate according to Newton's law. This reduces the radiating area above a certain frequency and also helps of axis response. But I know that explanation is simplistic as there are certain to be cone bending modes, radial or circumferential bell modes.
But Purifi's mass ring around the perimeter does work and you can see it in measurements where even 30° off axis has less roll-off that I have seen in any other driver. But it also makes the Purifi cone heavier at almost 20gr, quite a bit, and it's sensitivity (the "W" version, not "X" version), but the motor is very powerful and keeps the sensitivity high enough (the "X" is much lower). It just shows they did it their way.
> So he is actually complaining about that horn effect "accentuation" making it a problem when current-driving.
Esa is saying that it is only a problem when current-driving?
I have a good story to tell, so why not tell it. This is a subject that I have been interested in for maybe twenty years.
Below, this is a relatively recent measurement of a Satori 7" driver in a vented box. The Red is normal voltage (real) and the Green is current (simulated) to see the difference.
Yes, there is an accentuation. But it is still very there with both voltage, which makes my point. The accentuation is electrical, the horn effect is not.
BTW, note that 3KHz kink. It is very common.
Looking at the on and off axis (the above is on axis), I have drawn certain rational conclusions. Around 1KHz there can be a discontinuity is some drivers.
(Many years ago I measured a very expensive 6.5" Audio Technology driver worth $500 and it was definitely an outer surround problem at 1KHz.)
But that Satori, between 1KHz and 3KHz the response is not dying as if its cone was completely stiff (a la Allison). So there is definitely some bending here, and yet the response of the cone is dropping at 3KHz and would keep dropping, except above 3KHz something else is happening: The horn effect takes over that the response rises and peaks 5-6KHz. This change from cone roll-off to horn, that is what causes the kink. It is especially noticeable on axis - the above is indeed on axis.
But being an horn acoustic effect, when you look off axis and the CSD (waterfall plot to other readers here) shows that there is no serious underlying issue. To me, and my interpretation, that the peak is not a cone resonance, rather an acoustic effect due to cone shape, and it certainly has an obvious horn shape as only the inner, closer to the voice coil, is radiating the primary sound and the horn is secondary.
Summary: Between 1-3KHz we are looking at the cone area reducing and above that, the horn effect takes over, whether voltage or current.
That's my observation. Many years ago I made this quick measurement of two drivers:
The Black was a fibre cone and the Red was a poly cone. The fibre cone was slightly more sensitive. The driver structure, magnet and suspension, all the same. The only difference was the cone. The centre cap was slightly smaller. Other than that, cone shape identical, but two different materials. There was no underlying problem or resonance related to that common peak. Both drivers were suitable for 1st order crossovers, if the response was optimised for 15° off axis.
The common peak is a 'horn' effect and this a voltage, not current.
I showed the above to Lynn Olson and he agreed it was down to cone 'shape' and when later in 2010 I saw Esa's reference to 'horn effect' and I said "of course it is." The effect is acoustic like a rough waveguide, it's not electrical. Yes, using current you will accentuate it, but it was already there!
Last edited:
OK, my mistake.
And thanks to. I too believe something more is going on. Please read the post I made above, I would welcome. With 6-7" drivers I see outer surround problems usually showing up near 1KHz, not above 3KHz, although not excluding anything.
As I was saying.
Together in this region with lobing due to the diaphragm size, these effects are primary concerns with fullrange drivers as there is no protection.
Yes, EQ can fix some things and not others.
Even If the "horn effect" is not electrical it would be reflected in the impedance plot, which it is not.
There would also be discontinuities showing up in Directivity Plot if the cone were flexing as you say.
I am going to use the word 'reflected' in a slightly different way, please bear with me.
I do believe it is 'reflected' into the electrical impedance. But that does not mean that it shows up in an impedance plot as you say.
In my opinion (and I am not the only one) all the electro-acoustic-mechanical deficiencies of the driver gets reflected back into voice coil and modulates the Ei part of the impedance. Since all parts of the impedance are in series then the overall impedance changes.
I know that Esa Merilainen in his own words had said something similar. I would welcome his comments here, if that is possible. These Ei modulations is what current-drive is supposed to fix. If the driver was perfect and no deficiencies, then current-drive would not fix anything. But reflections back into the voice coil is a large issue.
But if doing a measurement plot, sweep frequency or whatever, it won't show up there. It is too static?
Of course these don't show up much if using LR4 crossovers. And the fact that 2KHz and even lower is common these days and using 4th order you are already down 24dB @ 4KHz and the horn effect is well suppressed and will help the Directivity Plot immensely.
I dare not to spend too much time fixing this problem. Yes, you will fix it, but to me this would be like throwing the baby out with the bathwater. If we know that 4th order crossovers sound correct and smooth, then we have a technical fix, particularly suiting the drivers that you have chosen. But the sound is not as good as it can be if we select drivers and do crossovers that are electrically 1st order. As one commentator with vast experience said, this kind of speaker sound "more accessible" and admitted he did not know why. (I would say, look at those Ei impedance modulations.)
Your thoughts?
In my case, I believe fixing problems on the electrical side are more important than fixing issues on the acoustic side. They are both important, but I consider one more important than the other.