Not for the faint of Heart!
But most of what is involved is well explained:
http://www.adamjhill.com/Hill - MSc Dissertation.pdf
But most of what is involved is well explained:
http://www.adamjhill.com/Hill - MSc Dissertation.pdf
Hi David,
I think I found a bug, I'll try to describe it, and, please, see attached Hornresp Export file. When I go into the Loudspeaker Wizard, and look at the response everything looks fine. With S2 in Manual, if I go 1 cm^2 up or down I get a grayed out field, and the Notice "Fr1 is too large". The behaviour goes back to normal when I get down to (in this case) 158 or up to 172. I can get the same behaviour at any S1=S2=S3 condition I have tried, and also, when in S2 Auto clicking S1 or S3 up or down results in a similar behaviour.
Regards,
I think I found a bug, I'll try to describe it, and, please, see attached Hornresp Export file. When I go into the Loudspeaker Wizard, and look at the response everything looks fine. With S2 in Manual, if I go 1 cm^2 up or down I get a grayed out field, and the Notice "Fr1 is too large". The behaviour goes back to normal when I get down to (in this case) 158 or up to 172. I can get the same behaviour at any S1=S2=S3 condition I have tried, and also, when in S2 Auto clicking S1 or S3 up or down results in a similar behaviour.
Regards,
Hi Oliver,
I wondered how long it would take for someone to notice this one
.The nature of the absorbent filling model is such that depending upon enclosure dimensions, flare profile, flow resistivity and frequency, the magnitudes of internally-generated values used in intermediate calculations can become extremely large in some circumstances - to the extent that the floating-point precision of the computer is exceeded, resulting in overflow or underflow errors. This is not really a flaw in the model but rather an intrinsic limitation in ability of the computer's processor to cope with the exceedingly large or small values involved in the calculations. I have attempted to reduce occurrences of the problem by restricting the frequency range to 2000 hertz rather than the standard 20000 hertz, but the problem can still occur from time to time, as you have discovered.
Making a subtle change to the dimensions is usually enough to overcome the problem, unless the flow resistivity value itself is too large, in which case it should be reduced.
Rest assured, the Hornresp absorbent material simulation model has a strong theoretical basis, and you can have confidence in the validity of the results it produces
.Kind regards,
David
Hi Mark,
Thanks for the additional links.
The fundamental problem is how to determine the voice coil temperature given only Eg and Re. I can't see how it can be done
.As you know, manufacturers use all sorts of techniques to try to minimise voice coil temperature, and without a detailed specification of the driver's magnet and voice coil assembly, together with a thorough understanding of the associated heat conduction and dissipation properties, the task of predicting temperature rise really becomes an exercise in futility.
Just the width of the air-gap clearance alone will have a significant effect on the extent of the temperature rise...
Kind regards,
David
Hi Mark,
Thanks for the additional links.
The fundamental problem is how to determine the voice coil temperature given only Eg and Re. I can't see how it can be done
As you know, manufacturers use all sorts of techniques to try to minimise voice coil temperature, and without a detailed specification of the driver's magnet and voice coil assembly, together with a thorough understanding of the associated heat conduction and dissipation properties, the task of predicting temperature rise really becomes an exercise in futility.
Just the width of the air-gap clearance alone will have a significant effect on the extent of the temperature rise...
May I suggest a wizard section that will allow some of the other paramters?
Wire type is important.
Coil height and diameter is important. But that can be reverse calculated from some of the parameters. I have a spreadsheet that does that if you want a peak. ( Not made by me, I wish I had thought of that!)
As for gap spacing there is pretty much a general rule of thumb for clearances.
And I would put it in the simplest of terms. If a person wants the capability to do the calculation, should not the person be able to input the values? Figure out the gap width, the coil height the wire diameter. Number of layers. Then punch it into a calculator.
I am exceptionally busy at the moment but in about a month I could provide some empirical verification if it helps. I have just a few types of drivers sitting in my shed out back of my shop. Many flavors and sizes. And I have a Infrared thermometer to.
I'm game if your up to it.
Regarding thermal compression, in the past for simple sims I've adjusted Re. For a more complex analysis I've driven the driver hard in free air (hot enough to begin to smell the glue) and then quickly measure the t/s parameters while hot. Then rerun the sim with the new hot t/s parameters.
I don't want to discourage this power compression simulation idea but I think this quote is more true than people would like to admit. In addition to the considerations listed in post 3926 you would need to know quite a bit about the venting and cooling incorporated into the driver and be able to analyze the effectiveness of these features.
Also, the model would have to be able to account for frequency, impedance and efficiency of the enclosure if you want a certain voltage level to correspond to a certain amount of heat. There will be a lot more heating at certain frequencies due to impedance (so certain songs will heat the driver more than others at the same volume knob setting), and efficiency of the design changes everything. Seaton and Danley are much more eloquent than I am, so here's a few quotes from the Labhorn design notes, this issue is talked about extensively in that discussion. First quote is from Seaton, the rest are from Danley.
I don't want to discourage this power compression simulation idea but I think this quote is more true than people would like to admit. In addition to the considerations listed in post 3926 you would need to know quite a bit about the venting and cooling incorporated into the driver and be able to analyze the effectiveness of these features.
Also, the model would have to be able to account for frequency, impedance and efficiency of the enclosure if you want a certain voltage level to correspond to a certain amount of heat. There will be a lot more heating at certain frequencies due to impedance (so certain songs will heat the driver more than others at the same volume knob setting), and efficiency of the design changes everything. Seaton and Danley are much more eloquent than I am, so here's a few quotes from the Labhorn design notes, this issue is talked about extensively in that discussion. First quote is from Seaton, the rest are from Danley.
Let me clarify my sugestion.
David could provide a wizard in which those who know and understand could input the required parameters.
If you work around drivers long enough some of what is missing in the spec sheets can be given an educated guess.
Specifically wire diameter can be gauged quite accurately by eye. Gap widths to can be seen with a bright light on many drivers.
Coil height to can be estimated with a good degree of accuracy. If you are determined you can count the number of turns and divide them from the coil stack height and deduce the wire gauge this way.
Tricks of the trade.
There are other means and methods available to get what is missing as well.
I don't expect dead accurate simulations. None of them are. But they are well with in reasonable assumptions.
Bottom line is very much in keeping with the programs philosophy to begin with.
The mathematical tools can be there to use by those who want to understand or do understand how to use them.
With no data in the wizard there could be no input from the wizard.
And like the Thiele Small parameters, derivations of what is missing are possible if you have some of the data.
Now I step off of my soap box.
These are all good points but I am saying that's only half the story.
For example if you compare something like this, from a company with a huge budget specifically targeted to cooling research and development, massive vents in all the places that vents are normally found on both sides of the motor, and smaller vents in places that vents are not normally found, a large split wound inside/outside voice coil, forced air cooling by design, and even a basket design that pulls heat away from the motor and probably several other cooling features that I'm not even aware of - B&C 21SW152-4 21" Neodymium Subwoofer 4 Ohm | 294-689
... to something like this with a sealed back, no vents anywhere, a tiny outside wound voice coil and no design budget for cooling considerations - GRS 5SBM-8 5" Sealed Back Midrange | 292-432
... there's going to be differences.
I don't know how to account for those differences. Maybe you do.
And like I mentioned, everything depends on frequency, impedance and efficiency of the enclosure. I'm not sure how the software would deal with that. A small sealed box playing car audio bass cds vs a stack of Labhorns playing girls and guitars music will have vastly different heat buildup even if both subs had the same DCR (combined dcr in the case of the Labhorn stack) and were given the same voltage.
Again, I don't know how to account for those differences. Maybe you do.
I'm no expert in the field of driver design and testing. My main interest is in enclosure acoustic design and performance so I'm out of my depth here. I would like to see some quantification of power compression but I don't see any accurate way to do that. This would have to be a very complex model taking into consideration several aspects of the driver design (including cooling features and their effectiveness, which you have not addressed yet), the enclosure design, voltage, frequency, impedance and efficiency.
Last edited:
The answer is rather simple.
You take a stab at it.
Any simulation done is Hornresp is just that. An best guess.
A thermal simulation can take only a few actual parameters into account. It cannot simulate real world conditions.
It can give an estimate. And there are fudge factors involved in all audio design. Nothing is cast in stone.
I'll throw in this example.
David allows us to model a driver as a rigid plane. Lossless rigid plane.
But interestingly I have done quite a few designs that measure amazingly close to the simulation. The math works. Period.
There are derivations of thermal modeling. I have many more sources, some are really brilliant and doctoral thesis. However the article by Claus Futtrup is one of the most lucid examples of how it really works.
And as you have shown other people have taken a stab at modeling it.
It is only as useful as the information put in.
Very much like the Horns and other enclosures we design in Hornresp to begin with.
The math behind Hornresp is solid, and I have seen a whole bunch of strange designs modeled successfully in it. I have even done a few strange but true types of enclosures.
We are asking for other aspects of enclosure design and they are indeed proving to make this program world class.
The GRS driver is simple.
It runs on smoke.
When the smoke comes out..... it stops running.
The B&C on the other hand is a whole other animal.
But in the specs you have a few things to begin with. You have former diameter, coil stack height. Re and from there you would be surprised what you can calculate.
There are ways of finding out these tings yah! ( I'm Half German ) ( Which half depends on whether I'm using my head or my muscles )
It runs on smoke.
When the smoke comes out..... it stops running.
The B&C on the other hand is a whole other animal.
But in the specs you have a few things to begin with. You have former diameter, coil stack height. Re and from there you would be surprised what you can calculate.
There are ways of finding out these tings yah! ( I'm Half German ) ( Which half depends on whether I'm using my head or my muscles )
Well, you know more about drivers than I do. If you and Mr. McBean think you can do something useful with this I'm all for it.
I'll still probably measure my drivers hot in free air and run the hot specs through the sim though for cases where an accurate answer is really important. This won't give any correlation to voltage vs heat in the actual design but it will tell what the parameters actually look like at high heat. (For expensive very high power subs that cost a lot of time and money to build, if I can ever afford to do such a thing - my tastes and ambitions these days no longer have any consideration for my wallet or for practicality, which is a very different situation than when I got into this hobby.)
I'll still probably measure my drivers hot in free air and run the hot specs through the sim though for cases where an accurate answer is really important. This won't give any correlation to voltage vs heat in the actual design but it will tell what the parameters actually look like at high heat. (For expensive very high power subs that cost a lot of time and money to build, if I can ever afford to do such a thing - my tastes and ambitions these days no longer have any consideration for my wallet or for practicality, which is a very different situation than when I got into this hobby.)
Last edited:
LOL, David is going to start dreading visiting his own thread. Every time he puts out a major update, more feature requests start pouring in. And they are getting increasingly more complex. (I'm as guilty as anyone for this.)
I've never heard of controlled leakage pattern. Are you talking about intentional diffraction or something like that? Hornresp doesn't even do unintentional (natural) diffraction yet...
EDIT - this doesn't have anything to do with Enabl patterns, does it?
Last edited:
I am referring to the conclosure. I believe they can be modeled like having a series of small ports at different locations depending upon the distance a pressure wave travels down a path. Originally I had a design patented but later chose not to maintain the patent due to the size of enclosure necessary and the complication of designing it.
The basic concept is to vary the reactive force on the driver from the back side, but not create the bass reflex port effects.
Now, if this can be done, effectivly you control both the loading from the front side using a horn and have better control of damping from the back side without effecting transient onset.
Now the backside wave path length needs to be 1/4 wavelength of F0. Of course picking the right driver for this application is also important.
The basic concept is to vary the reactive force on the driver from the back side, but not create the bass reflex port effects.
Now, if this can be done, effectivly you control both the loading from the front side using a horn and have better control of damping from the back side without effecting transient onset.
Now the backside wave path length needs to be 1/4 wavelength of F0. Of course picking the right driver for this application is also important.
I'm still not sure I get it. Are you talking about tuned 1/4 wave stubs? Kind of like the holes down the length of a flute?
TL.app can do that. I recently simulated a Hegeman enclosure for someone, it has 4 tuned 1/4 wave stubs but they are all closed ended and they all originate at the throat chamber, but you could easily model something more like a flute with open ended stubs located at various lengths along the line with TL.app. This software allows for as many branches off the main line as you like, wherever you want them, open or closed, any shape or size you like, on either or both sides of the driver.
Just as a visual example, here's the Hegeman sim I did, but remember you can put branches anywhere at any point in the enclosure, they don't all have to be bunched up in the same spot like this sim shows. Schematic showing driver, throat chamber, and 4 closed ended 1/4 wave stubs of different lengths. Notice how the Hegeman box squashes the impedance peak compared to the sealed box (the stuffing is mostly responsible for this though) but maintains the same response curve for the most part.
FR and Z of the Hegeman with heavy stuffing in first few inches of each stub. (A slightly different stuffing scheme would probably work a lot better for a less wiggly FR but I simulated it exactly as per the plans.)
FR and Z of an (IIRC unstuffed) sealed box with the same total volume.
I was mostly unimpressed with the Hegeman box but this sim demonstrates that what you want can be accomplished with TL.app (if I understand correctly what you want to do).
TL.app can do that. I recently simulated a Hegeman enclosure for someone, it has 4 tuned 1/4 wave stubs but they are all closed ended and they all originate at the throat chamber, but you could easily model something more like a flute with open ended stubs located at various lengths along the line with TL.app. This software allows for as many branches off the main line as you like, wherever you want them, open or closed, any shape or size you like, on either or both sides of the driver.
Just as a visual example, here's the Hegeman sim I did, but remember you can put branches anywhere at any point in the enclosure, they don't all have to be bunched up in the same spot like this sim shows. Schematic showing driver, throat chamber, and 4 closed ended 1/4 wave stubs of different lengths. Notice how the Hegeman box squashes the impedance peak compared to the sealed box (the stuffing is mostly responsible for this though) but maintains the same response curve for the most part.
An externally hosted image should be here but it was not working when we last tested it.
FR and Z of the Hegeman with heavy stuffing in first few inches of each stub. (A slightly different stuffing scheme would probably work a lot better for a less wiggly FR but I simulated it exactly as per the plans.)
An externally hosted image should be here but it was not working when we last tested it.
FR and Z of an (IIRC unstuffed) sealed box with the same total volume.
An externally hosted image should be here but it was not working when we last tested it.
I was mostly unimpressed with the Hegeman box but this sim demonstrates that what you want can be accomplished with TL.app (if I understand correctly what you want to do).
Last edited:
Adding heavily concentrated stuffing is not my idea of a good implementation, and generally one stub is more appropriate, but yes, for simulation purposes holes down the flute is a correct analogy. Each hole length and diameter is treated like a port of a base reflex. Generally the totall volume will probably not be more than a bass reflex port.
That TL seems to have each stub sealed. Seems like ti does not allow flute type ports or long slotted port?
That TL seems to have each stub sealed. Seems like ti does not allow flute type ports or long slotted port?