Re: Re: Re: Horn Design - Sanity Check
(if the bottom bass horn is omitted) - here the phase plug is present. Anyway how do I eventually make one? Saw someone at the goodsoundclub mentioning: About the shape of the phase plug. It simply is close to conical. The area through the plug follows the tractrix curve. There is also a 5mm front cavity between the diaphragm and the plug so the plug is quite close to the diaphragm.
So it feels to me like a throat area of say ~300cm2 is the way to go, given that Tractrix benefits are anulled because of dimension compromises, which contour would you suggest?
I guess for the mouth area here logically stands 1,4142x150 right?GM said:
Is beaming really an issue in the intended frequency range? (below 500 Hz) The midtops used above are like these - http://www.occasioneitalia.net/gestione/get_pic.php?id=10418
(if the bottom bass horn is omitted) - here the phase plug is present. Anyway how do I eventually make one? Saw someone at the goodsoundclub mentioning: About the shape of the phase plug. It simply is close to conical. The area through the plug follows the tractrix curve. There is also a 5mm front cavity between the diaphragm and the plug so the plug is quite close to the diaphragm.
That's right, I guess driving one horn with two drivers is also possible - in the end the air pressure in the front chamber just increases - but I expect even worse result predictability.
I guess I risk speaker damage using too high CR right? These are fairly robust but also very expensive - these stem from a Meyer Sound Line Array; I was told three drivers exist in Czech rep, two of them are currently in my workshop. A replacement driver costs (was told) some 1200USD (yes it seems a lot but given the exoticness of the thing it might pretty be true). The speaker model should be MS 415N, I have placed the measurement protocol here: Parameters, Impedance and frequency plot
So it feels to me like a throat area of say ~300cm2 is the way to go, given that Tractrix benefits are anulled because of dimension compromises, which contour would you suggest?
Re: A Suggestion
Thanks for your opinion Tim, The driver is actually 168 mm high so the rear chamber does not necessarily have to be that big - given the drivers low Vas I would think that lower Vrc is okay - anyway the only difference in hornresp seems to be a shallower dip at cca 200 Hz with larger volume
Made a few shots of the speaker
Tim Moorman said:
Thanks for your opinion Tim, The driver is actually 168 mm high so the rear chamber does not necessarily have to be that big - given the drivers low Vas I would think that lower Vrc is okay - anyway the only difference in hornresp seems to be a shallower dip at cca 200 Hz with larger volume
Made a few shots of the speaker
Re: Re: Re: Re: Horn Design - Sanity Check
"I wish I could write as mysterious as a cat." ~ Edgar Allan Poe
The only things that Hornresp really needs to be aware of "about sound" are the solid angle into which the acoustical power radiates, the velocity of sound (assumed to be 344 metres per second) and the density of air (assumed to be 1.205 kilograms per cubic metre).
With regard to the "accuracy" of Hornresp results, the attached screenprint compares predicted versus measured electrical impedance for a given horn loudspeaker. I would have thought that the correlation is close enough for the program to be considered a reasonably useful design tool.
I think the real point that "Romy the Cat" is probably trying to make is that simulations can only take you so far. The ultimate test is how the loudpeaker actually sounds in practice, to the listener. I agree with this view.
Nevertheless, Hornresp will at least get you started in the right direction, if nothing else...
Kind regards,
David
"I wish I could write as mysterious as a cat." ~ Edgar Allan Poe
The only things that Hornresp really needs to be aware of "about sound" are the solid angle into which the acoustical power radiates, the velocity of sound (assumed to be 344 metres per second) and the density of air (assumed to be 1.205 kilograms per cubic metre).
With regard to the "accuracy" of Hornresp results, the attached screenprint compares predicted versus measured electrical impedance for a given horn loudspeaker. I would have thought that the correlation is close enough for the program to be considered a reasonably useful design tool.
I think the real point that "Romy the Cat" is probably trying to make is that simulations can only take you so far. The ultimate test is how the loudpeaker actually sounds in practice, to the listener. I agree with this view.
Nevertheless, Hornresp will at least get you started in the right direction, if nothing else...
Kind regards,
David
Attachments
Brett said:
Couldn't have said it better myself - thanks Brett
.Seriously - the important message here is to read the Help file thoroughly before "diving in at the deep end". All the features available in Hornresp are covered. The Help file is only 16 pages long, so it doesn't take that long to work through from beginning to end.
Despite the so-called "clunky” interface, the program is actually quite easy to use once the user understands how it operates.
Kind regards,
David
David McBean said:
Couldn't have said it better myself - thanks Brett
Seriously - the important message here is to read the Help file thoroughly before "diving in at the deep end". All the features available in Hornresp are covered. The Help file is only 16 pages long, so it doesn't take that long to work through from beginning to end.
Despite the so-called "clunky” interface, the program is actually quite easy to use once the user understands how it operates.
Kind regards,
David
Thanks for adding your points David; I was confused by the program - as soon as I added any dimensions into the program I was not able to change the contour but still saw screenshots of hornresp with tractrix simulations - so I thought oops maybe this was only included in previous versions.
After a while of playing with hornresp I must admit the interface has some logic (status bar field dexriptions are a nice detail). The only thing that could IMHO get better is the contour selection - I need to either start a new design or null all the contour related fields before I can change the contour. But I can live with it.
You can lead a horse to water...
David,
Terrific work, and Horn Response is getting better all the time.
I miss a few features now and then, but I manage to muddle through. And I rediscover what a jewel this program really is.
Anything that allows me to take a $39 woofer of questionable pedigree and create a tapped horn sub of massive 20 Hz output on a few watts, while still in a manageable enclosure, well... that really IS MAGIC!
(Thanks, too, to Tom Danley.)
Romy is right about the ears being the ultimate tool, but HR, $40 driver, and a sheet of plywood might just give me the means to produce that rare vintage of pure, clean sound we all crave.
BTW, a PHL 1120 will also provide one helluva midrange (200Hz -2KHz, and beyond) in a fairly small horn, if anyone cares to model it.
Tim
David,
Terrific work, and Horn Response is getting better all the time.
I miss a few features now and then, but I manage to muddle through. And I rediscover what a jewel this program really is.
Anything that allows me to take a $39 woofer of questionable pedigree and create a tapped horn sub of massive 20 Hz output on a few watts, while still in a manageable enclosure, well... that really IS MAGIC!
(Thanks, too, to Tom Danley.)
Romy is right about the ears being the ultimate tool, but HR, $40 driver, and a sheet of plywood might just give me the means to produce that rare vintage of pure, clean sound we all crave.
BTW, a PHL 1120 will also provide one helluva midrange (200Hz -2KHz, and beyond) in a fairly small horn, if anyone cares to model it.
Tim
Hi deiksac,
Rest assured, I only add new features in - I never take anything out.
The more you use Hornresp the more you will come to realise that there is always "method in my madness". Everything (including the "clunky interface") is done for good reason.
It should not be necessary to start a new design or null all the contour-related fields before you can change the contour. If the existing profile is Con, Exp, Hyp or Sph, and you are in edit mode, simply click the relevant segment length input box to highlight it, and then press the appropriate keyboard key to input the first letter of the profile you want to change to. If the existing profile is Lec, Obl or Tra, and you are in edit mode, simply double-click on the profile label to change back to the Con default, and then input the first letter of the profile you require (if that is not Con).
Remember though, horn segments 2, 3 and 4 can only have Con or Exp flare profiles.
Note also, that all values in a horn segment (and any following segments) can be “bulk erased” or reset to zero by entering a blank field in one of the input boxes associated with that segment.
Hope this helps.
Kind regards,
David
deiksac said:
Rest assured, I only add new features in - I never take anything out
.deiksac said:
The more you use Hornresp the more you will come to realise that there is always "method in my madness". Everything (including the "clunky interface") is done for good reason
.deiksac said:
It should not be necessary to start a new design or null all the contour-related fields before you can change the contour. If the existing profile is Con, Exp, Hyp or Sph, and you are in edit mode, simply click the relevant segment length input box to highlight it, and then press the appropriate keyboard key to input the first letter of the profile you want to change to. If the existing profile is Lec, Obl or Tra, and you are in edit mode, simply double-click on the profile label to change back to the Con default, and then input the first letter of the profile you require (if that is not Con).
Remember though, horn segments 2, 3 and 4 can only have Con or Exp flare profiles.
Note also, that all values in a horn segment (and any following segments) can be “bulk erased” or reset to zero by entering a blank field in one of the input boxes associated with that segment.
Hope this helps.
Kind regards,
David
No offense intended by the interface remark David -we all appreciate the effort you put in, and I know it's deliberately done in that way. Compared to Akabak it's a dream. That was supposed to be an (unhappily expressed) reflection that it takes practice to be able to use it swiftly, as it's not the most intuitive layout for relative newcomers. One could say the same about pretty much every piece of audio-related software of course, so it's in good company, and it's one of the best simulating tools around. That it's free is nothing short of remarkable, given that it can be regarded as a genuine alternative to all the other horn-related software out there.
Regards
Scott
Regards
Scott
Having done a good deal of horn modeling myself, I will say that there are areas where its easy and areas where its difficult to impossible.
Modeling a horns impedance and its load on the driver is the easiest part. Thats because the Horn Equation is basically an impedance equation - it is one dimensional. But the true wave motion is three dimensional and as such any aspect of the horn that requires a good knowledge of the three dimensional aspects will be poorly represented by any model based on the Horn Equations - such as Hornresp.
In practice what does this mean? The simulations will be very good for impedance - as shown above - good for the very low frequencies where the wavelengths are much longer than the horn dimensions, but they will likely fail miserably as the wavelength starts to become comparable in size to the actual device.
That said, it turns out that virtually all contours act the same when the wavelengths are very large - only minor differences occur in the loading and the response. This is true for just about any contour with the same throat and mouth area and length. The connecting shape just doesn't matter that much. And where the shape does make a bigger difference, the Horn Equation is no longer accurate.
Horn design is a tough nut to crack.
Modeling a horns impedance and its load on the driver is the easiest part. Thats because the Horn Equation is basically an impedance equation - it is one dimensional. But the true wave motion is three dimensional and as such any aspect of the horn that requires a good knowledge of the three dimensional aspects will be poorly represented by any model based on the Horn Equations - such as Hornresp.
In practice what does this mean? The simulations will be very good for impedance - as shown above - good for the very low frequencies where the wavelengths are much longer than the horn dimensions, but they will likely fail miserably as the wavelength starts to become comparable in size to the actual device.
That said, it turns out that virtually all contours act the same when the wavelengths are very large - only minor differences occur in the loading and the response. This is true for just about any contour with the same throat and mouth area and length. The connecting shape just doesn't matter that much. And where the shape does make a bigger difference, the Horn Equation is no longer accurate.
Horn design is a tough nut to crack.
True. I was reading some of your papers over Christmas, Earl & what you described in them confirmed why I've been feeling a bit jaundiced about a lot of LF horns. Rayleigh & Webbster etc. have lasted for ~8 - 14 decades which isn't exactly bad going but their limitations are becoming increasingly obvious.
That said, if modelling software like Hornresp, MathCAD et al based on the 1 dimensional wave equation get you in roughly the right neighbourhood, at least that's a start.
That said, if modelling software like Hornresp, MathCAD et al based on the 1 dimensional wave equation get you in roughly the right neighbourhood, at least that's a start.
Scottmoose said:
Of course!! I still use my SPEAK software, which is virtually an antique, because it is about 90% correct. The next 5-9% could be modeled with more comprehensive codes, but I find a compromise solution to work better, something I call "hardware in the loop". Basically, some of the analysis is a computer model, but some is actual measured data. With this approach I can get about 95-99% of the way. The last couple percent has to be done with real hardware as no model can get this close to reality.
I can assure that Hornresp is quite accurate as long as the wavelenghts considered are not too short in comparison with horn dimensions.
On the other hand, one needs some understanding and common sense to do reliable computer modelling of anything. The closest example is electronic circuit modelling. BTW: Where can I buy PSpice ideal transformers and current sources?
On the other hand, one needs some understanding and common sense to do reliable computer modelling of anything. The closest example is electronic circuit modelling. BTW: Where can I buy PSpice ideal transformers and current sources?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.