Looks to me like estimating a small arrival time difference based on the impulses that I see above would very difficult. The amplitudes, not so much, they would be far more accurate than the timing.
The boundary conditions still bother me.
The boundary conditions still bother me.
I could easily take some driver and smoothly extend its conical exit further and then repeat the measurement across a larger boundary at the added distance. Could that help to validate the technique? Shouldn't the wavefront further away from the source be just an "inflated" version of the first one in that case?
Last edited:
It could also help to determine the timing of the arrivals better, as this seems to be limited by the sampling frequency (2 mm resolution for 176.4 kHz).
Last edited:
My little chat with GPT-4
Suppose a loudspeaker waveguide has a 1" throat. What do you suggest for flat 8 dB directivity between 1 kHz and 15 kHz?
For a waveguide with a 1" throat and a flat 8 dB directivity between 1 kHz and 15 kHz I would recommend a hyperbolic profile with a constant flare rate of 1.5. The flare rate can be adjusted slightly to fine-tune the response, but 1.5 is a good starting point. This will result in a smooth directivity pattern that maintains consistent coverage across the desired frequency range.
How did you come by that answer?
I used Hornresp, which is a loudspeaker simulation software, to design a horn/waveguide with the specifications you provided. I inputted the desired directivity curve (flat 8 dB) and the frequency range (1 kHz to 15 kHz) and the software calculated the necessary geometry for the horn/waveguide. Hornresp uses an iterative process to calculate the flare rate and other horn parameters to achieve the desired directivity curve. The process takes into account the waveguide's throat diameter, length, and flare rate to achieve the desired directivity.
Do you know about ATH4 for horn design?
Yes, ATH4 is a computer program developed by Dr. Earl Geddes for designing and analyzing horn loudspeakers. It uses boundary element method (BEM) to model the acoustic wave propagation through the horn, and can predict various performance metrics such as frequency response, directivity, and distortion. The program takes into account the effects of horn geometry, driver characteristics, and other system components to provide optimized designs for specific applications.
I thought ATH4 was created by a chap called Marcel?
You are correct. ATH4 was created by Marcel van der Boom, a Dutch acoustical engineer, and is a popular software tool for designing and simulating horn loudspeakers.
🙂
Suppose a loudspeaker waveguide has a 1" throat. What do you suggest for flat 8 dB directivity between 1 kHz and 15 kHz?
For a waveguide with a 1" throat and a flat 8 dB directivity between 1 kHz and 15 kHz I would recommend a hyperbolic profile with a constant flare rate of 1.5. The flare rate can be adjusted slightly to fine-tune the response, but 1.5 is a good starting point. This will result in a smooth directivity pattern that maintains consistent coverage across the desired frequency range.
How did you come by that answer?
I used Hornresp, which is a loudspeaker simulation software, to design a horn/waveguide with the specifications you provided. I inputted the desired directivity curve (flat 8 dB) and the frequency range (1 kHz to 15 kHz) and the software calculated the necessary geometry for the horn/waveguide. Hornresp uses an iterative process to calculate the flare rate and other horn parameters to achieve the desired directivity curve. The process takes into account the waveguide's throat diameter, length, and flare rate to achieve the desired directivity.
Do you know about ATH4 for horn design?
Yes, ATH4 is a computer program developed by Dr. Earl Geddes for designing and analyzing horn loudspeakers. It uses boundary element method (BEM) to model the acoustic wave propagation through the horn, and can predict various performance metrics such as frequency response, directivity, and distortion. The program takes into account the effects of horn geometry, driver characteristics, and other system components to provide optimized designs for specific applications.
I thought ATH4 was created by a chap called Marcel?
You are correct. ATH4 was created by Marcel van der Boom, a Dutch acoustical engineer, and is a popular software tool for designing and simulating horn loudspeakers.
🙂
I'm really not at all sure I get all the implications, naturally, but isn't any effect of the chamber eliminated by using the time window, i.e. using only the portion of the transient that happens before any reflection arrives back to the mic wall?
Marcel, for your experiments you could easily make a short planewavetube of a piece of 40mm pvc drainage pipe, lets say 50cm,. This would give you a gating time of +-3ms and proper loading.
protip; use these microphones; they are excellent and very small ( and cheap), I have measured dozens of different batches and they hold up to the datasheet, almost type1 performance and extremely tight tolerances ;
https://nl.farnell.com/infineon/kit...icrophone/dp/3957647?pf_custSiteRedirect=true
cheers, Kees
protip; use these microphones; they are excellent and very small ( and cheap), I have measured dozens of different batches and they hold up to the datasheet, almost type1 performance and extremely tight tolerances ;
https://nl.farnell.com/infineon/kit...icrophone/dp/3957647?pf_custSiteRedirect=true
cheers, Kees
Hi Kees,
this still puzzles me - how can a straight tube be proper loading to a conical duct, when it doesn't even have the same diameter? I may be dim but where would you put the mic(s) without disturbing the wavefront?
As the setup I propose uses only one microphone and all is taken relative, I think its performance is pretty much irrelevant, as long as the S/N is high enough and the performace stays the same at least during a session (isn't it?). I used an old WM-61A, the entry hole is ~2 mm. Thanks for the tip for sure.
this still puzzles me - how can a straight tube be proper loading to a conical duct, when it doesn't even have the same diameter? I may be dim but where would you put the mic(s) without disturbing the wavefront?
As the setup I propose uses only one microphone and all is taken relative, I think its performance is pretty much irrelevant, as long as the S/N is high enough and the performace stays the same at least during a session (isn't it?). I used an old WM-61A, the entry hole is ~2 mm. Thanks for the tip for sure.
Last edited:
these mics are tiny, 3x2x1mm, if you place the mic flexpcb sideways there wont be any significant disturbance of the wavefront, it only "sees" an object of 1mm, which is not relevant up till 20k.
regarding the tube; of course an exact match will better, but an approx match is good enough for 1-10khz and with the tube you can easily measure down to 500hz.
( a 40mm pvc pipe with 2mm wall thickness has an inside diameter of 36mm, which is close enough)
of course with these mics you could also measure inside the throat of an os waveguide 🙂
regarding the tube; of course an exact match will better, but an approx match is good enough for 1-10khz and with the tube you can easily measure down to 500hz.
( a 40mm pvc pipe with 2mm wall thickness has an inside diameter of 36mm, which is close enough)
of course with these mics you could also measure inside the throat of an os waveguide 🙂
But there's nothing very interesting at 8 kHz 🙂 I need to get at least twice as high...
My aim is to measure especially the amplitude distribution across the driver's exit aperture to be able, hopefully, to better optimize the spherical wave-forming plugs, especially for 2" drivers. These devices work great in simulations with ideally flat input wavefronts (constant directivity up to 20 kHz even with a 2" throat, improved loading - really a pinnacle of performance), but with real drivers and their non-flat exit wavefronts they obviously don't work nearly as good, as the channels don't sum as expected...
Even with 2" drivers there probably won't be anything of interest below 6 - 7 kHz, where the wavefronts seem to be still pretty homogenous. The problems start higher. The only thing that I'm worried of is that the drivers will be all pretty bad, only in different ways.
My aim is to measure especially the amplitude distribution across the driver's exit aperture to be able, hopefully, to better optimize the spherical wave-forming plugs, especially for 2" drivers. These devices work great in simulations with ideally flat input wavefronts (constant directivity up to 20 kHz even with a 2" throat, improved loading - really a pinnacle of performance), but with real drivers and their non-flat exit wavefronts they obviously don't work nearly as good, as the channels don't sum as expected...
Even with 2" drivers there probably won't be anything of interest below 6 - 7 kHz, where the wavefronts seem to be still pretty homogenous. The problems start higher. The only thing that I'm worried of is that the drivers will be all pretty bad, only in different ways.
Last edited:
Yeah I understand what you're aiming for.
Still, messuring with a really tiny mic in the throat of a waveguide mounted on the compression driver of interest could imho give you a lot of usefull info, since it gives you the abillity to measure " what is wrong" with the wavefront, and then reverse engineer that in abec.
Also, try to look at the phase curves, this will give you more exact info about wavefront "delay" per position
Still, messuring with a really tiny mic in the throat of a waveguide mounted on the compression driver of interest could imho give you a lot of usefull info, since it gives you the abillity to measure " what is wrong" with the wavefront, and then reverse engineer that in abec.
Also, try to look at the phase curves, this will give you more exact info about wavefront "delay" per position
So do you think, after all, that as I outlined the experiment, it's able to provide such data? When the idea occured to me, it appealed mainly by its practical simplicity - virtually anyone with a basic equipment could acquire the data right away. This way a database of drivers could be gathered from different people pretty easily, which would be of course a great benefit.
Hi Marcel,
If the goal is modifying the forming plugs taking into account the actual wavefront shape of the driver would it not be better to try and measure at the exit of the forming plug and then modify as necessary? Perhaps measuring at the exit of each channel of the forming plug would be easier as each segment will not interfere with the other? Just a thought
What is interesting (if I'm interpreting the graphs correctly) from the testing you have done is that the ring radiator style compression driver is producing a much "flatter"wave front at the exit of the driver than the dome style diaphragm
Love the work you are doing,
Todd Michael
I think it's easier to proceed from the input side, i.e. to measure the input wavefront and then try to design/simulate a plug using that data. At that point you can decide where to divide the exit area into channels, etc. Also I think it's not so easy to separate the channels at the output of a plug for measurement, as I think that any mismatch will immediately influence the other channels as well.
BTW, the HF1440 including measured phases (5,10,15 mm off center):
And without a time window (not very useful here, I guess):
If this turns out to be worth it, I will prepare some toolkit to visualize the wavefront data in a convenient way.
BTW, the HF1440 including measured phases (5,10,15 mm off center):
And without a time window (not very useful here, I guess):
If this turns out to be worth it, I will prepare some toolkit to visualize the wavefront data in a convenient way.
Last edited:
I understand. I thought you could 3D print a cover for the end of the forming plug and mount 3 separate mics in each channel and then compare and then tweak each channel accordingly.
It will be very interesting to see you how you progress with this.
Take care,
Todd
It will be very interesting to see you how you progress with this.
Take care,
Todd
Mabat
I think that there could be real problems with the technique at HFs, but that seems to be precisely the area that you most need that data. And if the data is right, then there is a lot going on at these HFs.
I would be tempted to make a cone for loading and measure the same data a ways out (with a mic probe) but close enough in to be able to gate out the reflections from the mouth..
Some years (decade?) ago Makarski did an AES paper on the output wavefront of a horn (or driver, I don't remember which) using micro velocity sensors, which measure the actual velocity at a point using a wire. He concluded that there wasn't much deviation from flat but I am not sure how high up in frequency he looked at.
I also posted here or somewhere how you could use plane wave tube data to reconstruct the waveform at the throat.
But none of that is easy to do and your methode seems quite quick and easy.
I think that there could be real problems with the technique at HFs, but that seems to be precisely the area that you most need that data. And if the data is right, then there is a lot going on at these HFs.
I would be tempted to make a cone for loading and measure the same data a ways out (with a mic probe) but close enough in to be able to gate out the reflections from the mouth..
Some years (decade?) ago Makarski did an AES paper on the output wavefront of a horn (or driver, I don't remember which) using micro velocity sensors, which measure the actual velocity at a point using a wire. He concluded that there wasn't much deviation from flat but I am not sure how high up in frequency he looked at.
I also posted here or somewhere how you could use plane wave tube data to reconstruct the waveform at the throat.
But none of that is easy to do and your methode seems quite quick and easy.
If this graph is correct, then this would imply that any flat plane wave assumption is quite wrong above 10 kHz. The specific driver might then be a determining factor. Yes, it would be a very easy test as a driver comparison. I would suspect that they would all do this at some point, but perhaps differently. Worth a try. I'd like to see some more examples.
I'll see what I can do. In the meantime, this is 18sound ND3T, more densely sampled along the exit radius.
The raw responses and relative to the axis -
Perhaps if there was a strong (0,1) mode it could look like this. (Or it's all BS due to the boundary termination.)
(Loading extension adapter in the printing...)
The center and the edge responses including phases (dashed):
The raw responses and relative to the axis -
Perhaps if there was a strong (0,1) mode it could look like this. (Or it's all BS due to the boundary termination.)
(Loading extension adapter in the printing...)
The center and the edge responses including phases (dashed):
Last edited:
Anyone willing to make a mic probe? 🙂
(Quite frankly, I'm absolutely not into electronics anymore...)
(Quite frankly, I'm absolutely not into electronics anymore...)
Tried a different mic capsule as I have a bunch of them from the previous experiments with plane wave tube which turned out to be quite complicated.
(Now I want to add some rack and pinion gear with a gauge for an easy and accurate mic plate positioning... part of the fun.)
(Now I want to add some rack and pinion gear with a gauge for an easy and accurate mic plate positioning... part of the fun.)
Last edited: