Not totally sure if I am following you.
We don't have to imagine rooms, sound pressure levels or kickdrums to picture what kind of effects of modulation are at play.
As soon as one variable moves with another variable, there is modulation going on.
But the fact that Sd doesn't change the same way in one cycle like BL does, makes things a lot different.
We don't have to imagine rooms, sound pressure levels or kickdrums to picture what kind of effects of modulation are at play.
As soon as one variable moves with another variable, there is modulation going on.
But the fact that Sd doesn't change the same way in one cycle like BL does, makes things a lot different.
A change of Sd of -say- 20% would lead to quite some amplitude modulation (1,6dB peak SPL) of the mid/high frequency signal. Sound power wise it's only half and probably not that noticeable.
b_force, yeah I'm trying to imagine how would that be and what you are thinking, hence asked for more how do you imagine it? I see the modulation happens differently with Sd and Bl, minima and maxima of the variable are at different parts during the excursion travel, but it still varies, the variation happens at different frequency.
Take the imagined slowed down video low frequency making huge excrusion, use Bl(x) curve instead of Sd(x), there is less force in the motor at both maxima. Now the modulation of the higher frequency is different, parameters for high frequency are ~normal when the cone is about at center of the travel and drops towards the ends of both positive and negative excursion maxima. Modulation of the high frequency is now twice as fast with the Bl(x) compared to Sd(x) (graph in your post). It takes quarter cycle of low frequency to get from Bl "max" to "min", while with Sd(x) it takes half cycle. Still, the both variables modulate between some values. More or less symmetrically if you recall Klippel Bl(x) graph for example.
Take the imagined slowed down video low frequency making huge excrusion, use Bl(x) curve instead of Sd(x), there is less force in the motor at both maxima. Now the modulation of the higher frequency is different, parameters for high frequency are ~normal when the cone is about at center of the travel and drops towards the ends of both positive and negative excursion maxima. Modulation of the high frequency is now twice as fast with the Bl(x) compared to Sd(x) (graph in your post). It takes quarter cycle of low frequency to get from Bl "max" to "min", while with Sd(x) it takes half cycle. Still, the both variables modulate between some values. More or less symmetrically if you recall Klippel Bl(x) graph for example.
yes that is correct, but again, that is only one half cycle, the other half cycle it's 1.6dB less relative to the steady fixed Sd.
Which is pretty different compared to a symmetrical curve like BL modulation.
In that case both cycles have the same issue = on average actually a real shift.
But you're bringing an interesting point to the table, which is calculating the same effects for BL as well.
At this point I think "probably not that noticeable" is not all the important.
It's just trying to find the physics behind it and if it has any change to begin with.
Which is a little hard if there is no actual data available ... .. . 🙁
this is correct point of view from the low frequency perspective, which makes the high excursion. But in addition there is high frequency stuff (in music), which has much lower excursion by itself, and it happens at the same time and makes on top of the low frequency excursion. What is half cycle for low frequency is multiple full cycles for some high frequency, so there on the highs the modulation is real, the parameter shift spans many full cycles.
Just in case with other words: since high frequency makes much less excursion (for same SPL) it modulates the parameters for itself much less than low frequency does to itself. But when there is both at the same time, the lows make high excursion which shifts the operating point for higher frequencies at speed of the low frequency (and what the parameter curve is in relation to that).
Last edited:
Maybe I am misreading, but I think you're mixing a couple of things up?
The modulation is created by the cone excursion.
Which is created by low-frequencies. Although that part isn't that relevant I think.
On top of this, we get the higher frequencies, which have no (significant) cone excursion.
If we again take the example from the BL, this is quite easy to picture.
Because the BL drops in the same direction when it's on both maximums of the cycle.
Meaning that not even at that particular moment the entire SPL drops (since it's proportional to BL), but also the average value drops (since BL drops in the same direction at both ends).
Now we have to imagine a similar scenario, but instead of dropping at both ends, one end is go higher, while the other end is going lower.
The average now is not the same as the previous, because they cancel each other out.
Because of this effect, I just can't see why this would be more significant as the previous?
Well that is just exactly the point I wanna discuss 😀 😀 😀
I am also totally fine with a mathematical explanation.
Because we don't have to think in Sd modulation, but just modulation with two sine waves basically.
Perhaps misunderstood, although I ment exactly what you write on the quote here.
On the rest of the post I think you are thinking from the low frequency perspective only, and do not consider the high frequencies in relation to the lows (the excursion). This is the part which we write (and likely imagine) different I think. Hence I wrote how to imagine the high frequency part on top of the slow motion low frequency part, trying to express how I imagine this stuff 🙂 Sorry I'm not sure how should I describe this, but I try once more.
Imagine there is the kick drum and singer playing through the same woofer. Now, when ever the kick drum hits, the parameters shift for the singer sound attenuating or boosting it some! When the kick drum doesn't play the singer gets constant output. Imagining this it could be as if some syllable was louder or quieter than intended, when ever the kick drum hits, which ever syllable happens to output while the cone is a particular position due to kick drum 🙂
Reflecting back to the first example of 50Hz and 1000Hz sine tones, the 1000Hz tone does 20 cycles in same time span while the 50Hz tone does one. In case of Sd(x), like in the graph you posted, roughly half of the 20 cycles of 1000Hz have increased amplitude while the other half have reduced amplitude if there is the 50Hz playing at the same time. But if you stop the 50Hz tone the amplitude of all 1000Hz cycles remain constant. Since we rarely listen sine tones, a more practical example could be that if you had high frequency transient that lasts say 1 cycle of 1000Hz it means it could be either compressed or expanded, depending on whether there is simultaneous low frequency output which has moved the voice coil to some position.
Perhaps it's not significantly more significant than any distortion mechanism is different from the other, it's just one of various distortion mechanisms. Sound is fun in a way that the biggest effect dominates, say, there is three different sources for 3rd harmonic, what ever those are. If you eliminate two of those measured 3rd harmonic might drop some, but there is still the third source which keeps it "high". To get really low distortion, all distortion mechanisms need to be addressed.
Since Sd varies between min and max during one cycle of sine tone, and Bl varies two times (both ends), these effects happen at different frequency (from each other), which I think makes them different harmonics (distortion).
Sorry cannot write maths, hopefully someone shines and provides that 😀
Imagination is so much fun, writing about it is also, very hard, but kind of helps imagination so I'll do it anyway, even though it wasn't too helpful for others. I think imagination can work in many ways between people, and it is hard to deliver it through text, or in any way 🙂
Last edited:
No, just the fact that we are talking about modulation means both.
I am most certainly not thinking low-frequency only, but I think we're mis-communicating here.
And I know you're trying to help with your kickdrum and other sounds examples, but they don't work for me at all 😀 😀
No, the most significant needs to be addressed, later the other ones if they even exist.
Fact is, we have only very little clue about its significance.
An even bigger fact is, that it's just a full hypothesis still.
Sure, there is some theoretical foundation to work from, but that doesn't say much.
If people make a claim, they have to back it up with actual numbers.
Otherwise we can also claim that the position of the moon is important as well.
Hehee, miscommunication it is then 🙂
I added this following edit after you commeted, perhaps this is related to what you are thinking?
Both Sd and Bl vary with excursion, it's just that the frequency of the variance is different. Since Bl varies faster (multiple times during one cycle of excursion), having two minima and one maxima between, while Sd seems to have only one of both, the Bl must make higher harmonics than the Sd.
Yeah for sure, loudest dominates in graphs. Not necessarily audibility.
Is there any other drivers than purifi that have -80db distortion figures? I sincerely don't know, but hunch is there aren't too many. I think is very likely the low number was not achieved if there was Sd modulation, so in this sense it must be one of the most significant distortion sources for the -80db number.
But as you say it would be nice to have some more on it. I can imagine it is a distortion mechanism so can take it without as it seems to have measured effect at least since total distortion number is less than many others have. It is another matter whether it's audible or not and personally I'm not too concerned about it in isolation, because all distortion mechanisms depend on excursion, so it's very effective try to have minimal excursion on a system if possible.
Last edited:
Sorry, I said this the other way around.
For the BL the frequency is higher, for Sd frequency is lower as well as amplitude, since maximums don't add up
Well they add up, just a lot less.
This is really hard to answer.
I don't know many other drivers that have a similar set of parameters and motor design.
There are just way to many other variables.
So you automatically aren't comparing apples with apples anymore.
Where do you base this on if we don't have any numbers to begin with?
What I do know is that if we don't have any (or very little) excursion, this is not even a problem to begin with!
So as long as we can isolate other issues like BL(x) and Le(i) we could maybe find an answer.
Fact is that Purifi has more tricks on board, one big significant addition is the inner magnet system as well as the huge demodulation rings.
(as well as a bunch of others)
Also nothing new, but the implementation is a bit different.
Besides, maybe the Sd(x) only contribute to those last few digits.
Which is a very different significance compared to "very obvious and clear modulation".
Yes, that is the correct conclusion for any of these modulation problems.
The exception here is Le(i) (and I think another one I always seem to forget...)
It's another way of saying that at a certain point it's looking for solutions by creating a problem to do everything with one driver.
But anyway, that is not the point here.
The point is to find out that certain claims hold any truth in it, and if so how much.
Hi, this is a thing I don't follow, what do you mean by adding up? what do you refer as amplitude here, amplitude of harmonics generated?
I'm thinking if both Bl or Sd varied 10%, aren't both having same magnitude effect? Per my rough understanding SPL output of driver is directly proportional to both.
Yeah just logic, since all kinds of motors have existed for long time, and if none has reached such low total distortion number then there is likely some other distortion mechanism that dominates the number, basically the suspension.
So motor would make most distortion, but if that is fixed then suspension is next to dominate numbers. Isn't non linear compliance problem at lows only, where suspension is significant part of driver operation, but on midrange it has veru little or no effect? Non linear Sd would be about opposite suspension related problem, affects more mids than on lows. Logically it has somewhat important role, quite much right after motor issues?
Last edited:
That's only true if all other components are the same.
Most other motors don't have a internal magnet for example.
The signal that is super imposed (modulated) by the change of this variable.
No they do not, since Sd(x) doesn't has the symmetry like BL(x) has.
Therefore the sum of the imposed (modulated) signal(s) can never be the same.
Again, see Sd vs excursion graph earlier.
That's the part I am trying to explain.
Yeah I think my mental model is not complete to understand relationship of changes in cone travel or in cone area to waveform, and to dissect that into components, and I don't remember seeing such stuff in sense that would help imagine this stuff. Modal decomposition or what is it officially/mathematically? Perhaps I can't respond further on your thoughts.
Why not take what lrisbo said that it's just difference between even or odd harmonics, that get generated, not any more complicated than that? I mean, the difference of Sd(x) graph and Bl(x) graph general shape just means different harmonic series?
Some obervations on the significance:
In your graph Sd varies between ~110 and ~140, which is quite much. If it made the 1.6db amplitude modulation as markbakk calculated, it is roughly similar wiggle what a -20db sound gives interfering with 0db sound, and so roughly 10% distortion and as such quite significant. I don't know what the max excursion on the graph represents, mechanical failure limits or was it just some reference SPL that made such excursion. Perhaps the other distortion mechanisms gave 10% distortion long before the max excursion values on the graph, but 10% is the value xmax is usually determined and I think the graph shows Sd making xmax, so kind of significant.
I added paragraph to previous post after your response, before I noticed you had already responded. What is your thought on this? Sorry about bad language, can't edit the original and I don't want to edit the quote either 😀
Why not take what lrisbo said that it's just difference between even or odd harmonics, that get generated, not any more complicated than that? I mean, the difference of Sd(x) graph and Bl(x) graph general shape just means different harmonic series?
Some obervations on the significance:
In your graph Sd varies between ~110 and ~140, which is quite much. If it made the 1.6db amplitude modulation as markbakk calculated, it is roughly similar wiggle what a -20db sound gives interfering with 0db sound, and so roughly 10% distortion and as such quite significant. I don't know what the max excursion on the graph represents, mechanical failure limits or was it just some reference SPL that made such excursion. Perhaps the other distortion mechanisms gave 10% distortion long before the max excursion values on the graph, but 10% is the value xmax is usually determined and I think the graph shows Sd making xmax, so kind of significant.
I added paragraph to previous post after your response, before I noticed you had already responded. What is your thought on this? Sorry about bad language, can't edit the original and I don't want to edit the quote either 😀
Last edited:
Yes, as can be seen by previous post from the Klippel graph, as well as what @markbakk said.
So non-linear compliance has no effect at all on mid-range performance, since it's physically can't
Taking someone's word for it still doesn't provide us data?
That it has the potential to generate harmonics is not the question here.
Do you think/mean Sd(x), in case of straight sloped line it's mostly linear process, no harmonics generated? or what's the question again?
lrisbo let's believe in post #193 it's not linear as long as Sd varies, I guess never minding the shape of Sd(x) curve, as long as the Sd is not constant.
Trying to scope what is missing in the posts or my understanding, I can't quite get your angle on this what you are after? what's the argument again?😀 is it that there is no data to support claims that Sd(x) is a thing, and also question about how meaningful it is? these are already speculated and thats all we can do without data. Was there something else?
Last edited:
Exactly this.
With no data it is and always will be just a hypothesis, where we only can guess how significant it is or not.
Like I mentioned before, I can also claim that the position of the moon is important, since it provides a theoretical force on the system as well.
Obviously we all understand that idea is a little far fetched, but at this point we have absolutely zero.zero evidence how significant Sd(x) is.
Again, it's not that I don't want to believe it, but without any data it's nothing more than just a believe, no proof.
Believing is something we don't do in science.
I see your point and it's true, evidence would be good. Although, I trust my reasoning based on the graph you posted and find it very plausible that it makes quite significant difference in distortion plots, so it's not just belief assuming the graph can be trusted. Where the graph is from?
Anyone with purifi drivers wants to replace the surround with normal aftermarket one and do some measurements and sound samples before and after?🙂
Anyone with purifi drivers wants to replace the surround with normal aftermarket one and do some measurements and sound samples before and after?🙂
That's interesting, because with reasoning I think it's very plausible that it makes very little difference in distortion plots 😉