Hi Marcelo,
There would appear to be a number of existing software applications that can already do what you are seeking.
For instance:
Strassacker: Speaker Building, Components
I am not familiar with the design optimisation techniques or algorithms used by such programs, so it would be a steep learning curve for me to attempt to include something similar in Hornresp.
Kind regards,
David
Hi arcgotic,
Everything is working just fine for me. Could you please export the two (Nd, ME1) or three (Nd, ME1, ME2) records you are using to specify the multiple entry horn, and post the exported files so that I can check to see if they generate the same error for me. If they do, then I should be able to track down the problem for you.
Kind regards,
David
Everything is working just fine for me. Could you please export the two (Nd, ME1) or three (Nd, ME1, ME2) records you are using to specify the multiple entry horn, and post the exported files so that I can check to see if they generate the same error for me. If they do, then I should be able to track down the problem for you.
Kind regards,
David
Thanks for the shared link. crossing finger for you to be able to do it in a future, you know, we have a lazy part of us that wants things easy
David,
I found the website below and comparing the results with the one you shared they are different. I'm mention this because both says that are simulating following W.J.J. Hoge formulas. I didn't found the original publication from this author and at what condition they are valid but crosschecking formulas I confirmed it's using the ones indicated below.
mh-audio.nl - Vented System
Vb = 15 * Vas * Qts^2.87
Fb = 0.42 * Fs / Qts^0.9
F3 = Qts^-1.4 x 0.26 x Fs
The port area could follow the rule of 1/3*Sd area and pot length adjusted accordingly to reach Fb or a more complicate routine could be a loop till air particle at port outlet reach 30m/s at maximum input power.
I will try to contact W.J.J. Hoge to see were the equations were published.
W. J. J. Hoge's Business Pages
For sealed box, I think the optimization remains the same, Qtc=0,707 (driver + loundspeaker), or other values requested by the uses.
I found the website below and comparing the results with the one you shared they are different. I'm mention this because both says that are simulating following W.J.J. Hoge formulas. I didn't found the original publication from this author and at what condition they are valid but crosschecking formulas I confirmed it's using the ones indicated below.
mh-audio.nl - Vented System
Vb = 15 * Vas * Qts^2.87
Fb = 0.42 * Fs / Qts^0.9
F3 = Qts^-1.4 x 0.26 x Fs
The port area could follow the rule of 1/3*Sd area and pot length adjusted accordingly to reach Fb or a more complicate routine could be a loop till air particle at port outlet reach 30m/s at maximum input power.
I will try to contact W.J.J. Hoge to see were the equations were published.
W. J. J. Hoge's Business Pages
For sealed box, I think the optimization remains the same, Qtc=0,707 (driver + loundspeaker), or other values requested by the uses.
Don't know about any AES/whatever paper, but he published 'Confessions of a Loudspeaker Engineer' with all this and much more plus basic XO filter design in the 8/78 issue of AUDIO.
GM
edit: Fortunately it's available: https://www.americanradiohistory.com/Archive-Audio/70s/Audio-1978-08.pdf
GM
edit: Fortunately it's available: https://www.americanradiohistory.com/Archive-Audio/70s/Audio-1978-08.pdf
Last edited:
Hi Marcelo,
Out of interest I tried entering the Hornresp default driver parameters into the two calculators. As you say, they generate completely different results, neither of which appear to be optimum (the first calculator produced some very strange results indeed).
Another approach perhaps, would be to simply adjust the Vrc, Ap and Lpt slider controls in the Hornresp loudspeaker wizard while looking at the combined power response chart. It only took me about 30 seconds to come up with a far better result than that obtained using either set of calculator-generated values
.Kind regards,
David
Ah, that brings back memories - thanks GM
. I could not recall the contents, but I vividly remember the cover, even though it is 40 years since I would have last seen it. I have a thing about wizards - along with hundreds of millions of others, I just love the Harry Potter books .Hi nc535,
Not a problem
.It turned out to be easier to fix than I was anticipating. One line of code had gone missing - I have no idea how, or when, it happened.
Kind regards,
David
Another approach perhaps, would be to simply adjust the Vrc, Ap and Lpt slider controls in the Hornresp loudspeaker wizard while looking at the combined power response chart. It only took me about 30 seconds to come up with a far better result than that obtained using either set of calculator-generated values
Hi David,
It looks great and promissing. If the Hornresp based on driver's T/S can automatic optimized the Vrc, Ap and Lpt for the flatter frequency responce the work is done. Probably the way you found are more precise compared to the others that might be based on experiments.
Thanks.
Regards,
Marcelo
I think it's basically curve-fitting at work. I think those equations also assume a certain measure of box losses (usually Ql=0.7). Hornresp does not take box losses into consideration.
OTOH, the equations typically use simple t/s parameters and usually don't take Le, and definitely not semi-inductance, into consideration. They shouldn't be considered a substitute for a good design program like Hornresp.
Hi Brian
The proposal wasn't to replace hornresp by the equation, but to introduce an automatic optimization routine into the hornresp.
The mentioned equations was to start the discussion based on a published document resulted from a research effort.
The hornresp already have Lossy Le based on experiments and also semi-inductance, both already incorporated, so if the user enable those parameter they could be used in the optimization process.
Taking as example, It's worth to mention that some DSP have some automatic routine to optimize equalization to achieve the flatter response as possible for a specific sound system in a specific environment. Automatic routine will never be perfect and will never cover all possibilities but they do an important job, mainly for the ones that are growing in experience.
Think about design Tapped Horn drawing on paper with a pen and now think about design tapped horn using CAD with parametric sketch. So automation in certain way helps to move forward.
Optimize wizard for bass reflex could be a tool like lossy Le witch the user could decide to use or not.
To do something like that, you need to define what the goal is. Those "design equations", when given a driver's t/s parameters, provide a box size and tuning that's supposed to give the flattest response in the passband - that is the goal of those equations is. IRL though, using a maximally-flat build with a low Qts driver like what you see in pro audio is going to produce sub-optimum results.
So, if you want an optimization routine, you need to define what the goal is. Maximally-flat is probably not the best goal to aim for. So what should goal be?
Yep, the goal there is defined - adjust the EQ until the target response is achieved.
we do mistakes during life to get experience, but better if it can be short cut. So even if the optimization routine could produce sub-optimum results, they will better than a bad design. If the user choose a bad driver to a BR, so there is nothing Hornresp could other then extract the best results for that specific driver.
The compare tool is there to compare different simulations (different drivers)
Mates could propose to David the targets they think are interesting to be considered.
The optimization routine could vary regarding the target that cold be selected by a menu.
For a while i'm asking just the optimization for a flatter response. Do you have other targets in mind?
Just to clarify - the Vrc, Ap and Lpt slider controls were adjusted manually, not by using any automated iterative "curve-fitting" algorithm. In effect, it was simply a 30-second trial-and-error process, while observing the resulting response
.