For the first two bends (that make up the 180 degree bend), yes this is correct. For the third bend you didn't do advanced centerline. See picture below.
I'm going to disagree with tb46 and say no. I've never put a driver right in a bend before so I've never had to draw this out but it only makes sense to me to draw it out like any other bend. L34 is not the distance from the acoustic center of the driver to the mouth per se. Everything has to go on the centerline so I'd say you need to put S3 at the center of the horn flare at the point that the driver taps into the centerline of the flare.
That might sound a bit confusing so here's a picture showing how I'd do the advanced centerline around the last bend approaching the mouth, and where I'd assume S3 is (the red arrow, which upon further reflection I might even push back a couple more inches down the red line away from the mouth), and how to measure L34 (the distance along the red centerline from the arrow to the mouth, + 0.75 inches to account for material thickness passing through last 3/4 inch of the mouth itself.
Once you start placing S2 and S3 at places other than right on the red centerline your length measurements are going to start to get too long and your finished product will be tuned too low when you measure it.
Note that I didn't measure anything, I just drew some lines in so they may not be in exactly the right spot but they are close enough to show how I would approach this.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Hmm, the design didn’t line up with the driver’s specs, so using your path-length, net Vb and a bit of fudging to get the presumed theoretically ‘correct’ throat area, flare factor since the driver won’t physically fit to get its default HF response, all its sims are still statistically identical to yours, so should perform ~as predicted if the published specs are accurate and you build it correctly, but if I’m understanding how you arrived at the alignment, then your construction sim is wrong.
When I view the schematic it doesn’t have the requisite parabolic ‘bow’ shape of a parallel wall cum angled divider board construction; so guessing that you designed each section using the ‘par’ flare rather than designing using the ‘con’ flare and then having the wizard adjust the entire path-length to a parabolic one caused enough increase in net Vb to require adjusting while maintaining the same throat area.
Of course, this will require changing the angle of the divider board.
It probably doesn’t change actual performance any measurable amount, but since this is a learning exercise, figured I should point it out.
GM
Post #101
Hi just a guy,
As I mentioned on P.8 Posts #72-77 I'm just not certain what is correct in this particular case, and I suggested to just use the "advanced centerline method", and be done w/ it. If one could build and measure, that might tell us what is correct here. It's a little bit like the Keystone (where it is much more so) in the fact that there is a small stub in the corner below the driver that has to be acoustically active, but doesn't quite show in the simulation. It would be nice to know.
Regards,
Hi just a guy,
As I mentioned on P.8 Posts #72-77 I'm just not certain what is correct in this particular case, and I suggested to just use the "advanced centerline method", and be done w/ it. If one could build and measure, that might tell us what is correct here. It's a little bit like the Keystone (where it is much more so) in the fact that there is a small stub in the corner below the driver that has to be acoustically active, but doesn't quite show in the simulation. It would be nice to know.
Regards,
I don't know what you mean by "stub in the corner below the driver". Are you talking about the distance between where the acoustic center of the driver would be to the red centerline?
If you want the horn to be tuned right you have to put S2 and S3 on the centerline to calculate the segment lengths. This might not accurately portray the distance from the acoustic center of the driver to the mouth but there's nothing you can do about that (and I'm not sure the driver's acoustic center matters anyway except in relation where it taps into the flare length) but it's the only way to get the right tuning. In fact if any of the segment lengths don't follow the red centerline you aren't actually using the centerline (or advanced centerline) method at all.
I've still never seen the Keystone plans but I'm still not convinced it impossible to simulate with reasonable margins of error.
There's multiple reflections going on between the driver and end wall, terminus and end wall plus the long reflections of course. Then there's the volume lost due to the driver 'pinching' it and terminus end correction + local boundary condition, so the acoustic path-length of S3 is a complex calculation.
FWIW, my simpleminded way of calculating a folded horn's physical path-length puts this one ~57.1 cm too long compared to the sim, lowering its tuning ~6.5 Hz, so hope it gets built/measured to see what it turns out to be.
GM
FWIW, my simpleminded way of calculating a folded horn's physical path-length puts this one ~57.1 cm too long compared to the sim, lowering its tuning ~6.5 Hz, so hope it gets built/measured to see what it turns out to be.
GM
To elaborate a bit, I didn't look at any of the numbers at all. But a quick glance at the 180 degree bend suggests there's a few problems with this fold that I didn't mention in my previous posts. All the numbers are symmetrical on the left side of the box, which suggests the flare does not increase in csa at all between the beginning of the 180 degree bend and the end of the 180 degree bend.
Like I said, I never put a driver in the middle of a bend before and it might be a bit problematic but I don't think it's a really big issue, if the driver wasn't in the middle of a bend this would all be very straightforward. When the driver isn't in a bend the location of the driver's acoustic center isn't important at all, we consider the segment marker defining the driver position to be the spot where the driver location taps into the centerline along the length of the horn.
Anyway, I don't see any need to build this if it's not correct, and it appears there are a few problems.
But assuming the problems were addressed, I still don't think it needs to be built and measured to find where S3 should be. There are probably 100 simulated, built and measured tapped horns with this exact same layout, with the driver right inside of the last bend. Most of the early tapped horns looked exactly like this. There's quite a bit of data to be had for anyone willing to spend the time looking it up and studying it.
Alternatively, all that needs to be done to cut through the confusion is move the driver out of the bend. There are a few different ways to do that, either by altering the sim or the fold.
For the OP, I would suggest putting this issue on hold by moving the mouth to the end of the box instead of firing out the side. That will simplify everything. (This is just for practice anyway, isn't it? You don't need to build this, you just need to practice folding.) Then you need to export the horn data chart and make sure that every segment marker and the distance from the free end of the divider panel to each outside wall of the box has the correct cross sectional area for it's location on the horn path length.
Last edited:
Hi James,
Just to clarify:
Acoustic impedance Za = p / U
Where:
Za = acoustic impedance at a given surface
p = effective sound pressure averaged over the surface
U = effective volume velocity through the surface
p and U are both complex quantities, having real and imaginary components
Acoustic intensity = p ^ 2 / (rho * c )
Where:
p = sound pressure
rho = density of air
c = velocity of sound in air
Kind regards,
David
Just to clarify:
Acoustic impedance Za = p / U
Where:
Za = acoustic impedance at a given surface
p = effective sound pressure averaged over the surface
U = effective volume velocity through the surface
p and U are both complex quantities, having real and imaginary components
Acoustic intensity = p ^ 2 / (rho * c )
Where:
p = sound pressure
rho = density of air
c = velocity of sound in air
Kind regards,
David
Yup, there's a problem at the 180 degree bend. The end of the divider panel needs to be a bit closer to the bottom panel in order for the horn expansion to be carried through the bend.
Help, Experiment #2 is under attack - where's my rifle!!!!!!
Seriously Gentlemen, your criticism is exactly what I am looking for. I am sailing without a map and your comments will save me from self destruction.
All my large power tools are packed away in a shed for the winter, so it will be at least March before I actually get around to building something. At that point I will actually be building at least one of my designs (there are more ideas that will come before that point) and I will attempt to do some actual acoustic measurements on the finished build.
So in the meantime, this is a learning experience for me to try to learn as much as I can about the acoustics/physics/design/folds of tapped horns.
I am going to pause my "Question of the Day" temporarily as I am learning so much, so fast, I am worried I may have an aneurism.
But please don't stop criticizing my designs. Let me know every possible flaw. I will try to respond to some of last night comments individually as the day progress's.
Thanks again, you guys are amazing.
Seriously Gentlemen, your criticism is exactly what I am looking for. I am sailing without a map and your comments will save me from self destruction.
All my large power tools are packed away in a shed for the winter, so it will be at least March before I actually get around to building something. At that point I will actually be building at least one of my designs (there are more ideas that will come before that point) and I will attempt to do some actual acoustic measurements on the finished build.
So in the meantime, this is a learning experience for me to try to learn as much as I can about the acoustics/physics/design/folds of tapped horns.
I am going to pause my "Question of the Day" temporarily as I am learning so much, so fast, I am worried I may have an aneurism.
But please don't stop criticizing my designs. Let me know every possible flaw. I will try to respond to some of last night comments individually as the day progress's.
Thanks again, you guys are amazing.
Last edited:
Mr. McBean,
I believe you are the creator of this Hornresp program, and I would like to take a minute to congratulate you. I am not sure what spurred you on to create this program, or what has kept you motivated to continue the updates, but what an amazing piece of software you have created. The level of knowledge you must be at to understand all the parameters involved in speaker design and then create a software program to model those factors makes me feel very small. Hornresp is like the thinking man's video game.
I see there is a 400+ page forum post dealing specifically with Hornresp, and someday I hope to live long enough to read through all of that. The only possible improvement I can think of is if somehow there could be a way to save all the driver parameters, so you could easily swap out different drivers in a cabinet design, for a quick comparison.
Thank you Sir for all the work you have put into Hornresp. I believe I may be addicted though!
Posts #49/59/62/72/74/76/77/78/106
Hi just a guy,
As I pointed out in Post #40 I did not do a Hornresp simulation of the Keystone sub, and the drawing in Post #94 of the Keystone sub thread reflects that http://www.diyaudio.com/forums/subwoofers/185588-keystone-sub-using-18-15-12-inch-speakers-10.html . I suggested an alternate path to use for a Hornresp approximation of the design in the drawing in the lower right hand corner. Thanks to xrk971's work in AkAbak on the Karlson enclosure I think now, that there is actually a way to represent the actual design in a simulation, but so far I'm only close, and don't want to post the unfinished result.
Thanks for catching that DHAA has placed the top of the divider in the middle of the 180 degree bend. I thought we had covert that, but then, we covert so much ground, that it probably did not register. Let me just point back to Post #59 for a good example on how to do this correctly, and in Post #59 it should say "pdf in Post #737" not 734 (dang typos, and no easy way to correct them). As to the L34 distance littlemike and epa - Posts #77/78 - seem to like the driver acoustic center to mouth (originally it said throat, but that was later corrected) (i.e.: the shortest path) too, and they have both designed, built and measured extensively.
Post #106: "... There are probably 100 simulated, built and measured tapped horns with this exact same layout, with the driver right inside of the last bend. Most of the early tapped horns looked exactly like this..." I'm not aware of that many "build and measured" THs, I have work relatively carefully through cowanaudio's and volvotreter's examples, and even those never quite answered that above question. If you have one particular example that shows simulation, and measurement results I would appreciate a link/reference to take a look at it.
It is clear to me now, that I should not have brought this up.
Hi just a guy,
As I pointed out in Post #40 I did not do a Hornresp simulation of the Keystone sub, and the drawing in Post #94 of the Keystone sub thread reflects that http://www.diyaudio.com/forums/subwoofers/185588-keystone-sub-using-18-15-12-inch-speakers-10.html . I suggested an alternate path to use for a Hornresp approximation of the design in the drawing in the lower right hand corner. Thanks to xrk971's work in AkAbak on the Karlson enclosure I think now, that there is actually a way to represent the actual design in a simulation, but so far I'm only close, and don't want to post the unfinished result.
Thanks for catching that DHAA has placed the top of the divider in the middle of the 180 degree bend. I thought we had covert that, but then, we covert so much ground, that it probably did not register. Let me just point back to Post #59 for a good example on how to do this correctly, and in Post #59 it should say "pdf in Post #737" not 734 (dang typos, and no easy way to correct them). As to the L34 distance littlemike and epa - Posts #77/78 - seem to like the driver acoustic center to mouth (originally it said throat, but that was later corrected) (i.e.: the shortest path) too, and they have both designed, built and measured extensively.
Post #106: "... There are probably 100 simulated, built and measured tapped horns with this exact same layout, with the driver right inside of the last bend. Most of the early tapped horns looked exactly like this..." I'm not aware of that many "build and measured" THs, I have work relatively carefully through cowanaudio's and volvotreter's examples, and even those never quite answered that above question. If you have one particular example that shows simulation, and measurement results I would appreciate a link/reference to take a look at it.
It is clear to me now, that I should not have brought this up.
Last edited:
True and is part n’ parcel of my last critique, which should be addressed when the sim is corrected to parabolic and the sketch to match.
GM
The consensus appears to be that my 180 degree fold is not right. I now believe I see what I did wrong.
What I think I did wrong: I centered the top of the baffle board so that is was evenly spaced between all three cabinet walls. Therefore, when using the Advance Centerline Method, the first 90 degrees of my 180 degree fold were identical to the second 90 degrees, which is wrong. My thinking was that the center would be where it should be, but after putting more though into the consequences, the first 90 degree dimension should be slightly smaller than the second 90 degree dimensions, because the horn path should be gradually expanding.
How I think I can fix it: What I believe I should do is to go back to Hornresp and export the horn path .txt file. Using that, I should re-plot my baffle board, essentially dropping it down slightly from where I had it in my SketchUp drawing.
So, can someone please confirm if this is the proper solution? I will then redraw the SketchUp drawing with the proper dimensions and move on to my next imperfection. Thanks.
What I think I did wrong: I centered the top of the baffle board so that is was evenly spaced between all three cabinet walls. Therefore, when using the Advance Centerline Method, the first 90 degrees of my 180 degree fold were identical to the second 90 degrees, which is wrong. My thinking was that the center would be where it should be, but after putting more though into the consequences, the first 90 degree dimension should be slightly smaller than the second 90 degree dimensions, because the horn path should be gradually expanding.
How I think I can fix it: What I believe I should do is to go back to Hornresp and export the horn path .txt file. Using that, I should re-plot my baffle board, essentially dropping it down slightly from where I had it in my SketchUp drawing.
So, can someone please confirm if this is the proper solution? I will then redraw the SketchUp drawing with the proper dimensions and move on to my next imperfection. Thanks.
"Did not register" is probably an understatement. I'll address this further in my next post.
Maybe 100 built and measured designs was an overstatement, I'm not sure, but all the early tapped horns looked exactly like this so there should be at least a few. The cowanaudio and volvotreter examples alone should give what is needed, assuming there are plans and accurate measurements. If there are not, or if these plans and measurements don't correlate then more work needs to be done. Or the easier plan of attack would be to avoid putting the driver in bends.
I don't agree. There should be some kind of consensus backed up by data for all issues, this one included. If I had time I'd do a study with the two examples I just mentioned and the Keystone, and maybe a couple of other models with plans and reliable measurements.
Unfortunately I don't have time, and for the time being, my position is that this issue is not that complex. Starting with the basics, you can take the Hornresp "Horn Data" chart and use that to fold up a horn flare without even knowing if the flare is going to be used for an end loaded front loaded horn, an offset driver front loaded horn, an end loaded or offset back loaded horn or a tapped horn. No matter what, the folded flare will be exactly the same EVEN IF YOU DON'T KNOW WHERE THE DRIVER IS GOING TO BE. All you need is the horn data chart and you can draw the folded flare and draw the centerline (or advanced centerline) through the middle. Then, depending on what type of alignment the flare is for (end or offset flh or blh or tapped horn) you just drop the driver into the picture of the folded flare at the appropriate point. What you do not do is change the flare to accomodate the driver location, if that is done the sim no longer matches the folded drawing and the length of the flare will be wrong.
The rules of folding using the centerline or advanced centerline are pretty clear, I don't see much if any room for improvisation. I already mentioned that it may be problematic to put the driver right in a bend but if you change anything about the flare to accomodate the driver location you are not using the centerline (or advanced centerline) folding method.
And I'll repeat this one part because it's so important. You can draw out the folded flare without even knowing what type of horn the flare is going to be used for or where the driver will be.
Last edited:
Thanks, Just a Guy. I just wanted to make sure I understood that correctly. I've got a busy day going on, but I will try to get that drawing re-posted later today if I can.
One additional question. With such a simple fold as this one, as long as I got the end of the baffle correct, wouldn't everything else automatically fall into place with the proper dimensions (since the slope of the baffle board stays consistent)? Or are there additional anomalies I should be on the lookout for? Thanks.
The biggest flaw I see is using 1/2" rather than 3/4" material for a sub.
Saving weight and loosing upper output and "punch" is not a compromise I'd be willing to make.
Hornresp assumes inert walls, 1/2" won't be at high power levels.
This brings up something I've been meaning to ask for the horn experts here - what equation should be actually used to calculate a parabolic expansion?
The equation I use in my spreadsheet to estimate the expansion between S1 to S4 is as follows:
S(d)=((S2-S1)*d/L)+S1
Where
S1 = cross-section at the start of the path
S(d) = cross-section at point "d" on path
S2 = cross-section at the end of the path
L = path length
I believe this gives an exact match for what's actually happening a square or rectangular horn with two parallel sides built out of flat panels. However this might not match what HornResp is using.
Hmm... I didn't get that.
I assume this is the critique of which you speak. I guess tb46 understood just fine, but embarrassing as it is, I have to admit I don't understand much of what you said here at all. I understand the part about changing the angle of the divider board but that's about it.
First of all, how did you know how he arrived at the alignment? It's been my understanding that the OP is just playing with the sliders like most people.
What do you mean by "statistically identical"? I used your design routine on this example and identical is not a word I would use to describe the results.
The whole second paragraph is confusing to me. It sounds like you are making some assumptions about his design routine and where it went off the rails, but as I mentioned, I thought his design routine consisted entirely of playing with the sliders.
I didn't see anything in this post (or the next one) that addresses the issue of the non expansion of csa in the 180 degree bend, that's why I posted about it myself. Clearly you say it's mentioned, and tb46 seems to agree, so apparently I'm completely clueless. (Which actually isn't anything new a lot of the time.)
To be more clear about this -
I figured it out. It took awhile but I got it.
It's really too bad you won't discuss it, it would be nice to have confirmation that I'm correct. For example, (and without going into any detail at all) if I break this design routine down into 6 parts, I'm 100 percent sure I'm doing the first 5 steps correctly but the last step (which arguably isn't even terribly important) I'm only 93.8 percent sure due to a small percentage of discrepancy. While studying a bunch of your previous design examples I noticed that in some cases you gave three different designs for a given driver, a small, medium and large version. This adds a bit of complication but I'm still pretty sure I know how you came up with all three sizes.
The next step is to use this design routine to analyze if it's actually doing what it's supposed to. I have some doubts about this part.
You've made it pretty clear you don't want to discuss this, but since I've figured it out that might change things. I doubt it, but you know my email address if you reconsider.
- Status
- Not open for further replies.