This is what's going to be repetitively stated as mantra without any further thinking by almost anybody around here - with Bjorn Kolbrek possibly being the only exception I know of
😉
I do not agree.
There simply is no such limit. If you can determine specs involved well enough - and handle sim setup with care - results obtained by BEM Software (AxiDriver in my case) are great up to wherever you like and have horse power and patience for
Seems me and Bjorn being the only ones *fully* appreciating BEM capabilities at the time being (not that I claim to play in the same league)
Michael
Last edited:
Yeah - one might be tempted to think so - but I grantee, its not that way around
🙂
Michael
Last edited:
According to the data posted, the curve is for 1cm in front of the diaphragm, which means the curve is for practically at the throat of the horn. However, if this type of modeling gives consistent results comparible to measurements at various angles, then probably the tool can just be used like this for performance predictions.
Hello Soongsc,
Being as much confident in Michael than in you I'll wait this point to be cleared by him.
For what I understand Michael tried to simulate a situation as in Martin Seddon's measurement of the Yamaha JA6681 on the AH160 horn (measurement near of the mouth, this means at distance of more than 1 meter from the driver)
Best regards
Jean-Michel Le Cléac'h
Being as much confident in Michael than in you I'll wait this point to be cleared by him.
For what I understand Michael tried to simulate a situation as in Martin Seddon's measurement of the Yamaha JA6681 on the AH160 horn (measurement near of the mouth, this means at distance of more than 1 meter from the driver)
Best regards
Jean-Michel Le Cléac'h
Jean-Michel, I am running a second simu now which is closer to the near field / 4Pi situation of the measurements we have.
Might take some time before I come back...
Soongsc - you were exactly right in that the mic distance is referenced to the diaphragm !
Michael
Might take some time before I come back...
Soongsc - you were exactly right in that the mic distance is referenced to the diaphragm !
Michael
It took me some time to figure out how to use AxiDriver in a way that would be useful in diagnosing possible problems. Also found a bug that was fixed in the latest release.
Last edited:
And where *exactly* starts HF - *you* think ?
10kHz, 100kHz, 1MHz 10MHz ?????
Sure there are limits - as said - the basic limit is horse power versus time available...
Anything else you would have to show a prove for first !
😉
Michael
10kHz, 100kHz, 1MHz 10MHz ?????
Sure there are limits - as said - the basic limit is horse power versus time available...
Anything else you would have to show a prove for first !
😉
Michael
At high frequencies, compression and expansion process nonlinearity gradually increases. There seems to be no hard line as to where the problems starts to occur. To avoid this, the math model needs to be such that takes such process into consideration. In the mean time, it's just human intelligence that chooses what makes sense and what not on a case by case basis.
Soongsc thanks for mentioning – its been a while that I looked after an updated version of AxiDriver.
Regarding upper frequency limit of BEM simu – again – you have to deliver prove first that such non-liearities happen at around 10-20kHz in a dominant manner – I haven't seen any so far.
#########
Back to where my simu actually was taken....
Checking twice sometimes isn't enough ! 🙁
I possibly should report Joerg Panzer (AxiDriver programmer) this curiosity as well...
Here is what I found out : Simus are referenced different !
- for "a pressure curve" (what scoongsc referred to) plots are referenced in distance to where the contour joins the driver
- for "a pressure conto" (what my trace were drawn from) plots are referenced in distance to where contour joins infinite baffle
Quite puzzling - I admit..
So - my simu, shown a few postings before, was at the horn mouth (*not* inside the horn ) - being in full accordance to where the measurements where taken !
For comparison how FR develops along distance - I show again the JMLC horn in infinite baffle situation :
horn contour and sound field at ~ 1kHz
on-axis SPL at throat :
on-axis SPL at mouth :
on-axis SPL at 1m distance :
on-axis SPL at 2m distance :
on-axis SPL at 4m distance :
Quite interesting – no?
Michael
Regarding upper frequency limit of BEM simu – again – you have to deliver prove first that such non-liearities happen at around 10-20kHz in a dominant manner – I haven't seen any so far.
#########
Back to where my simu actually was taken....
Checking twice sometimes isn't enough ! 🙁
I possibly should report Joerg Panzer (AxiDriver programmer) this curiosity as well...
Here is what I found out : Simus are referenced different !
- for "a pressure curve" (what scoongsc referred to) plots are referenced in distance to where the contour joins the driver
- for "a pressure conto" (what my trace were drawn from) plots are referenced in distance to where contour joins infinite baffle
Quite puzzling - I admit..
So - my simu, shown a few postings before, was at the horn mouth (*not* inside the horn ) - being in full accordance to where the measurements where taken !
For comparison how FR develops along distance - I show again the JMLC horn in infinite baffle situation :
horn contour and sound field at ~ 1kHz
on-axis SPL at throat :
on-axis SPL at mouth :
on-axis SPL at 1m distance :
on-axis SPL at 2m distance :
on-axis SPL at 4m distance :
Quite interesting – no?
Michael
Last edited by a moderator:
But now it even gets more interesting !
For sanity check, below "just another bunch of pretty simus" of the same JMLC horn on same driver as shown in the posting before – but this time *no* infinite baffle set-up – but much closer to 4Pi free air.
🙂
horn contour and sound field at ~ 1kHz
on-axis SPL at throat :
on-axis SPL at mouth :
on-axis SPL at 1m distance :
on-axis SPL at 2m distance :
on-axis SPL at 4m distance :
Any questions of interest left ? 🙂
Michael
For sanity check, below "just another bunch of pretty simus" of the same JMLC horn on same driver as shown in the posting before – but this time *no* infinite baffle set-up – but much closer to 4Pi free air.
🙂
horn contour and sound field at ~ 1kHz
on-axis SPL at throat :
on-axis SPL at mouth :
on-axis SPL at 1m distance :
on-axis SPL at 2m distance :
on-axis SPL at 4m distance :
Any questions of interest left ? 🙂
Michael
I have no intention of writing a thesis on that subject. If you think your sims are good enough, then it won't even matter either. I think for what I've paid for AxiDriver, it certainly has provided valuable information.
Are we able distinguish driver aberration from horn aberration in a range 8-20kHz? Power response plus impedance needed?
http://www.diyaudio.com/forums/multi-way/140190-jean-michel-lecleach-horns-59.html#post2053981
http://www.diyaudio.com/forums/multi-way/140190-jean-michel-lecleach-horns-59.html#post2053981
I don't care about AxiDriver in particular (except its a great tool available for free to everyone) - but I *do* care about BEM as a analysis tool in general and for "diffraction alignment device" development in particular.
I read so much down talking about BEM that I feel a need to correct for this - it's just like fighting the windmills until - once and a sudden - everybody will jump the train 😀
I'm certainly aware that any simu has limits - it's just that I haven't come across yet - when having a closer look, until now its always been the restrictions of not having precise enough specs or not setting up high enough resolution !
Thats quite a difference to tellin' *BEM* has problems...
🙂
On the other hand I certainly would be interested to more seriously test for the limits - a thesis would be of little help I'm afraid - but - you could suggest a cheap and widely available horn driver combination thats well spec'ed (verified!) and demonstrate what you think the restrictions are.
Others can then check back...
(sort of what I tried to do here with the JMLC measurements and specs available)
😉
Michael
Last edited:
I see lots of people either doing sims or just doing measurements. Up to now, besides the earlier work you posted on bipole horn, I have not seen others that have done both. Lots of people are talking about compression drivers, but no body has even came out and said they simulated the compression driver section starting from diaphragm and includes the phase plug.
I think if anyone is serious in doing what I have posted, any concaved dome driver with a modified face plate would work. It is really not necessary to copy.
If anyone wants to discuss the discrepencies between my measurements and my sims, please feel free to do so where it is posted. I cannot disclose proprietary information, but I can do my best to answer questions that may arise.
BTW, AxiDriver is also for sale. I did purchase a license because I found it useful enough to use. I would also recommend others that find this software worthwhile to actually purchase a license. Everyone is responsible for the economy.
Last edited:
Michael,
I'm curious as to what horn contour data point increments you use. Most of my sims were 2mm increments.
I'm curious as to what horn contour data point increments you use. Most of my sims were 2mm increments.
Depends
- whether I'm in patient mood or not
🙂
Mostly I use a special software tool - I've got coded for that very case - to reduce contour points, that only marginally affects final overall resolution.
This way I can crank up effective resolution and keep number crunching reasonable...
Michael
Last edited:
Hi Michael,
Interesting indeed, particularly when your "on-axis SPL at 4m distance" response is compared to the Hornresp far-field directivity response for the same system 🙂.
AxiDriver:
200Hz = 71 dB
2000Hz = 77 db (+6 dB)
Hornresp:
200Hz = 71 dB
2000Hz = 77 db (+6 dB)
Kind regards,
David
Attachments
Hello David and Michael,
For me, the picture is more clear now. We can see that Hornresp simulate quite accurately the axial response at a large distance from the mouth.
There still a small difference between the response simulated by Hornresp and by Axidriver in the high frequency part of the response (above 2kHz for the example shown) as the reponse predicted by Hornresp fall more quickly.
Then, when we compare to a true measurement, there still some difference with both predictions. I give in attached file a .wav file (saved as .txt) corresponding to an unprocessed pulse response I obtained at 1 meter precisely from the mouth of a Le Cléac'h horn having an Fc of 320Hz mounted on a TAD TD2001 compression driver.
Then we can analyze this measured pulse response. Compared to the simulated response, the measured response is flatter for sure. This indicate that my equivalent model for the TD2001, have to be refined. It was obtained at a time when Hornresp did'nt had the directivity tools and only the power response was available.
Now that we know that the axial response in Hornresp seems (is) accurate, few parameters of the equivalent model should be revised to obtain a better modelisation of the driver to accord to the measurements.
Thank you for your contribution.
Best regards from Paris, France
Jean-Michel Le Cléac'h
For me, the picture is more clear now. We can see that Hornresp simulate quite accurately the axial response at a large distance from the mouth.
There still a small difference between the response simulated by Hornresp and by Axidriver in the high frequency part of the response (above 2kHz for the example shown) as the reponse predicted by Hornresp fall more quickly.
Then, when we compare to a true measurement, there still some difference with both predictions. I give in attached file a .wav file (saved as .txt) corresponding to an unprocessed pulse response I obtained at 1 meter precisely from the mouth of a Le Cléac'h horn having an Fc of 320Hz mounted on a TAD TD2001 compression driver.
Then we can analyze this measured pulse response. Compared to the simulated response, the measured response is flatter for sure. This indicate that my equivalent model for the TD2001, have to be refined. It was obtained at a time when Hornresp did'nt had the directivity tools and only the power response was available.
Now that we know that the axial response in Hornresp seems (is) accurate, few parameters of the equivalent model should be revised to obtain a better modelisation of the driver to accord to the measurements.
Thank you for your contribution.
Best regards from Paris, France
Jean-Michel Le Cléac'h
Last edited: