The Basilar Membrane and Crossover Distortion

For the past 15 years or so I have been researching and developing my interest in the part that the basilar membrane plays in how we interpret the sound of the loudspeaker.
In particular the effects that are produced by trying to integrate two drivers over the mid-band crossover area.

Considering that the basilar membrane can set up all manor of compressions, masking effects and also distortions within the hearing system I have not yet seen any relative articles or discussions on the subject of phase beating and it's connection with speaker crossover distortions in the critical mid range area.

Beat (acoustics) - Wikipedia, the free encyclopedia

With this in mind I have looked into how the critical bandwidth and two tones in very close proximity in frequency can cause roughness, smoothness and a phase locking beat when the two tones line up exactly in frequency. As the two tones go outside the critical bandwidth region they can react in a way that they are de-tuning each other. For example an effect that happens when the multiple strings for the same note of a piano is de-tuned so it becomes a “honky-tonk” piano.

My own crossover topology and design methods reflect my findings so far in the bid to try and decrease crossover distortions, phase anomalies, and colourations that are sometimes caused by conventional crossover methods.

I would be interested if anyone on this site has ever experienced anything to do with this subject or similar.

Thanks Ian Knight.
 
This link might help explain balisar membrane a bit better..
Basilar Membrane Motion 1: Two Tones - Animation | Auditory Neuroscience

Are your xo designs including acoustic offset compensation? That alone helps a great deal with phase alignment between offset drivers on a flat BB. There is much discussion over at PE TT about minimum phase. Worth a visit. Ooops, looks like you have!
 
Hi Allen,

That's an interesting point about the fixed phase relationship.
The beating and reinforcement of certain frequencies will occur according to the crossover frequency points are imposed on each driver and what slopes are used. The closeness of the crossover points will affect the phase relationship of the two drivers. The gap between the two knee frequencies of the crossovers can be quite wide before the drivers start to reinforce each other in the crossover region.
Usually the 1st order crossovers will give the most audible lock on but a combination of low order slopes can work well if the correct implementation of the drivers is used.

Cheers Ian
 
Interesting links re the Basilar membrane.

I'm not seeing the connection with crossover design? Even a poor crossover can't shift frequency so I would think the occurance of beats would be the same nomatter what?

I'm an amateur piano tuner (Dad was a pro) so I am familiar with beats and find them interesting. In the absence of nonlinearity, they are more perceptual than real. That is, when two tones get quite close together we perceive more a single pitch with a varying amplitude, yet a spectrum analyzer would show two tones of constant amplitude.

An interesting phenomonon with pianos is that the overtones of a string are somewhat inharmonic (not exact multiples) This can lead to to notes beating even when the fundamentals are at an exact ratio.

Regards,
David S.
 
Hi Dave,
Thanks for your message.
It is playing the piano that started me off on this journey. I have never been satisfied with the sound of speakers right from an early age!.
Its good to find someone like your self who works with the beating phenomenon.

If the two drivers are used in a certain way and they are crossed over at certain frequency points then this introduces phase shifts that will alter the group delay between the drivers at these frequencies.
I first did experiments over 12 years ago of using a conventional 1st order Butterworth crossover at a middle frequency of around 3khz in a 5.25" 2 way design. I was not happy with the crusty/fizzy midband despite the drivers being smooth in response in that area and capable of reproducing the frequencies well.
I decided to start pulling the crossover frequencies apart, the fizzyness subsided and so did the midrange detail and presence. But what I tried next was to pull apart just a bit farther in frequency and the midband started to come alive and had solidity and focus. I started to pull away slightly and it all locked on. I went too far apart and it did what would be expected.....there was a gaping hole in the middle.
It led me to investigate further how this could possibly be working.

Cheers Ian
( I have more info on my website )
 
I first did experiments over 12 years ago of using a conventional 1st order Butterworth crossover at a middle frequency of around 3khz in a 5.25" 2 way design. I was not happy with the crusty/fizzy midband despite the drivers being smooth in response in that area and capable of reproducing the frequencies well.
I decided to start pulling the crossover frequencies apart, the fizzyness subsided and so did the midrange detail and presence. But what I tried next was to pull apart just a bit farther in frequency and the midband started to come alive and had solidity and focus. I started to pull away slightly and it all locked on. I went too far apart and it did what would be expected.....there was a gaping hole in the middle.
It led me to investigate further how this could possibly be working.

The reason the spread is needed is it produces a peak in the summation if 1st orders are used at the same frequency. BW1 do not sum flat without spread. Peak if not far enough, hole if too far, there is a sweet spot.

I don't know that the experiment listed has anything to do with beat frequencies at all. Still- great discussion thread!

Later,
Wolf
 
Hi Dave,
Thanks for your message.
It is playing the piano that started me off on this journey. I have never been satisfied with the sound of speakers right from an early age!.
Its ..... vestigate further how this could possibly be working.

Cheers Ian
( I have more info on my website )

Are you sure you are not just realizing that no one I am aware of has ever recorded a piano realistically, let alone reproduced it in any speaker? I have only heard one reproduced instrument that almost, again almost, could convince me it was live. It was of an upright bass, half-track master tape, etc. Recorded music can be fantastic, but it sounds nothing like live. I am not sure it ever can.
 
Interesting links re the Basilar membrane.

I'm not seeing the connection with crossover design? Even a poor crossover can't shift frequency so I would think the occurance of beats would be the same nomatter what?

I'm an amateur piano tuner (Dad was a pro) so I am familiar with beats and find them interesting. In the absence of nonlinearity, they are more perceptual than real. That is, when two tones get quite close together we perceive more a single pitch with a varying amplitude, yet a spectrum analyzer would show two tones of constant amplitude.

An interesting phenomonon with pianos is that the overtones of a string are somewhat inharmonic (not exact multiples) This can lead to to notes beating even when the fundamentals are at an exact ratio.

Regards,
David S.

PS..Dave,

I forgot to mention that I find it fascinating that measurement doesn't always correlate with weat we perceive or hear!.
This I think has a bearing on why the manufacturers( B&W) that I have taken my designs to have a real job understanding what I am trying to explain to them. They always seem to rely on measurements first then tweak until the measurement is correct.
Cheers Ian
 
The reason the spread is needed is it produces a peak in the summation if 1st orders are used at the same frequency. BW1 do not sum flat without spread. Peak if not far enough, hole if too far, there is a sweet spot.

I don't know that the experiment listed has anything to do with beat frequencies at all. Still- great discussion thread!

Later,
Wolf
Hi Wolf _ teeth , thanks for your reply.
That is true what you say.
How far apart would you normally you suggest this might happen?.
If the drivers are operating within their natural piston range, and not being used into areas that might make the frequency responses irregular and peaky this gap or spread can be quite wide .
Cheers Ian
 
Are you sure you are not just realizing that no one I am aware of has ever recorded a piano realistically, let alone reproduced it in any speaker? I have only heard one reproduced instrument that almost, again almost, could convince me it was live. It was of an upright bass, half-track master tape, etc. Recorded music can be fantastic, but it sounds nothing like live. I am not sure it ever can.

That's a very valid point that you have made.
We all know that there isn't perfection in this pursuit, but whilst the recording engineers and studios use very poor quality monitors to mix upon I think that we may well have had to be listening to the by-product of that.
I have designed nearfield monitors that are in use in some professional studios in Florida some years ago. There is also one of my 7.1 wall soffit mounted systems used in the executive listening suite in Warner Music New York. Both use my design methods.

Live music is my benchmark. It is also very rare that an orchestras violin section can be reproduced realistically. It usually is balanced so forward in the mix, that just sends amplifiers and speakers in to a real oscillation sometimes.

Cheers Ian
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
I was not happy with the crusty/fizzy midband despite the drivers being smooth in response in that area and capable of reproducing the frequencies well.
I decided to start pulling the crossover frequencies apart,
I suspect we've all battled with this one. For quite some time I was perplexed, with the thought that woofers and tweeters were just not meant to be together.

It is interesting to hear about another's journey through this.

to rely on measurements first then tweak until the measurement is correct.
I feel that the relative dispersion of the drivers and related effects, as it is not always considered, probably accounts for the larger share of this discrepancy. On the other hand the absolute dispersion of a system also has an effect which could be inadvertently attributed to a crossover.

The beating and reinforcement of certain frequencies will occur according to the crossover frequency points are imposed on each driver and what slopes are used.
I could be misunderstanding something here. Do you mean cancellation and reinforcement? Ordinarily the phase from a tone from each of two drivers will remain apart, ie they will track together with a certain offset and won't cross each other over time.

I agree that crossover changes can affect the phase relationship of two drivers. Regardless of the slope you chose for either driver, the apparent gap or slope or whatever, the phase relationship can be set arbitrarily.

An interesting phenomonon with pianos is that the overtones of a string are somewhat inharmonic (not exact multiples)
There is a lot of wonderment inside a piano :) This is surely a field where art is valid and even encouraged.

May I ask about the strings that are made with nested coil-overs, whether their mechanical complexity is responsible for producing the non-harmonically related overtones?
 
Hi Wolf _ teeth , thanks for your reply.
That is true what you say.
How far apart would you normally you suggest this might happen?.
If the drivers are operating within their natural piston range, and not being used into areas that might make the frequency responses irregular and peaky this gap or spread can be quite wide .
Cheers Ian

Drivers are never flat and ultimately won't be best fit for a translatable argument. The loudspeaker Design Cookbook covers BW1 xover "spread factor" with a xover part-value multiplier. Top it off with lobing,
dispersion, Baffle-step, and harmonic distortion signatures, and you are really going back to zero.
Later,
Wolf
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
The Loudspeaker Design Cookbook had me thinking that crossing where the drivers are at 4.5 to 5dB down would be ideal, but I wouldn't say that in general now. The fact is that although the ideal first order crossover is 3dB hot at the most, when you figure in the nulls the power averages to flat in the vertical plane. Furthermore, if you maintain the 90 degree phase difference between the drivers on axis, they'll be flat there.

This would lead me, if all else is as stated, to look to the horizontal planes among other factors.
 
I suspect we've all battled with this one. For quite some time I was perplexed, with the thought that woofers and tweeters were just not meant to be together.

It is interesting to hear about another's journey through this.


I feel that the relative dispersion of the drivers and related effects, as it is not always considered, probably accounts for the larger share of this discrepancy. On the other hand the absolute dispersion of a system also has an effect which could be inadvertently attributed to a crossover.


I could be misunderstanding something here. Do you mean cancellation and reinforcement? Ordinarily the phase from a tone from each of two drivers will remain apart, ie they will track together with a certain offset and won't cross each other over time.

I agree that crossover changes can affect the phase relationship of two drivers. Regardless of the slope you chose for either driver, the apparent gap or slope or whatever, the phase relationship can be set arbitrarily.

There is a lot of wonderment inside a piano :) This is surely a field where art is valid and even encouraged.

May I ask about the strings that are made with nested coil-overs, whether their mechanical complexity is responsible for producing the non-harmonically related overtones?

Hi Allen,

I think the point that you raise about the relative dispersion could be a major factor in the integration process of the the 2 drivers.
I have always felt that a big woofer and a (small) tweeter unit couldn't live together properly either!.
I have been doing all sorts of things to try ascertain what is actually happening when trying to tie-in the two drivers at certain frequencies. This is why I am working on the basis that to give anything in the crossover region a chance of reproducing correctly then for starters we should not be pushing the drivers out of their natural piston range. EG. getting too close to the mechanical breakup in woofers, or conversely getting to close to FS in tweeters.
Although the response is never flat from a driver, at least it is smooth, and working morenaturally in the mechanical and electrical sense.
The dispersion characteristics also become more predictable for each driver.

Yes I do mean the Cancellation and Reinforcement !.
Yes, as you say, the phase relationship can be set arbitrarily. according to the crossover that is used, but just as tilting the speaker back, or altering the plane of the tweeter to so called time align the units , I thinks that we perceive is quite different once the correct spacing of frequencies has been achieved even if both the woofer and the tweeter have identical phase characteristics such as 1st order.

Cheers Ian
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
Keeping away from the woofer breakup region will also be keeping away from narrowing dispersion. A dome tweeter has wide dispersion and so side reflections would otherwise be thin at the crossover region.

Do you like anything about the sound of the mid/treble produced by a good smaller full range driver, apart from them being light on the upper treble?
 
Hi Allen,

I think the point that you raise about the relative dispersion could be a major factor in the integration process of the the 2 drivers.
I have always felt that a big woofer and a (small) tweeter unit couldn't live together properly either!.

Who says they really can? It's just a compromise.. Ideally one would want a two way crossed low, at the schroeder room's frequency and one driver going up all the way from there. But, how do you keep distorsion and dispersion under control, and still manage decent SPL? Has anybody made such a driver yet? Constant directiviy has been THE big thing these last few years.
 
Keeping away from the woofer breakup region will also be keeping away from narrowing dispersion. A dome tweeter has wide dispersion and so side reflections would otherwise be thin at the crossover region.

Do you like anything about the sound of the mid/treble produced by a good smaller full range driver, apart from them being light on the upper treble?

I haven't tried any full range drivers to be honest.
But I agree with what you say about keeping away from the woofer breakup zone. It promotes a lot of things that can be used to advantage when dealing with integrating the tweeter. In fact I have an article regarding this on my website. The articles are intended as an introduction to the way I design speakers.

http://http://knightloudspeakers.com/wp-content/uploads/2013/01/Dealing-with-the-breakup-zone.pdf

If you are interested just sign up to the website it's free!.

Cheers Ian
 

Elias

Member
2003-05-28 5:27 pm
Suomi
For the past 15 years or so I have been researching and developing my interest in the part that the basilar membrane plays in how we interpret the sound of the loudspeaker.


Vau, that's quite a long time to focus on basilar membrane only :D

You may have found out that it serves as a prefilter, and the most interesting stuff on sound perception happends in the modulation domain in time-frequency-amplitude space at the deeper parts of auditory chain.

If you find interest, check my 'basilar membrane' wavelets aka Bark wavelets on my home page.


- Elias