DIY bass horn build...Up next - Page 9

Aww, BFM caught me ghosting there.

I'm not sure how you judged the sensitivity, but Ilkka doesn't measure it.

"Frequency response was measured crossover at its max and min settings and also bypassed if available. 10 second sine sweep from 200 Hz to 10 Hz was used. Level was matched at 90 dB at 50 Hz at 2 m distance."

He also used 2m distances for everything, so the conversion to standard 1m apply.

I do agree that it wasn't the best horn, but at least it looks like they used the mobile audio version of the XLS, IIRC.

Man, I really wish he had done impedance graphs.

hint... hint... for those doing similarly...

I have that driver in my horn database, so I know what it's capable of. It's totally unsuited to a 45Hz horn, Fs being half what it should be. You can fudge with it to get a useful result, but with not much sensitivity. Even in a horn with appropriate Fc it doesn't work well. I'm not surprised that it took them a year to come up with what they got.
But it does run clean, like all horns do.

Ok, I just thought you might have a way to tell from the info from the site. I'm always looking for a new way to do the same 'ole things.

"I will be switching to a TH build instead. I can get 2-4 db extra sensitivity out of it at least on paper. I may also switch the driver to the big b&c because it has a smoother response and a bit more overall sensitivity."

i kind of like that thinking. the b&c 380 liter tapped horn with ~30hz corner and a little over 2:1 compression looks really nice. [no need to post it, i'm sure you are looking at it among several other ideas.] the exposed driver of the tapped horn may provide for better cooling for a p.a. rig. same for the b&c driver. from your testing, it should be much harder to destroy the b&c with over-excursion in the th than the lms, no? i know you can put a high pass on there, but just the peace of mind factor that if somehow you put a massive un-passed bass signal through the th, the driver will be fine.

So the peaks and notches are due to reflections at the fold points.

Leaving aside construction logistics, if instead of folding an expanding spiral were used, would these go away?

Along that line it would seem that 90 deg folds would be better than 180 deg ones; is that the case?

"So the peaks and notches are due to reflections at the fold points."

hi noah, who suggested this was the case? there are several horns of different fold geometries that more or less conform to the predictions made by programs such as hornresp. even my models that back-engineer commercial designs seem to produce consistent results regardless of the folding geometry that is ultimately employed.

hornresp allows for modeling conical, exponential, hyperbolic, etc. expansions. the peaks and notches are there even in the ideal unfolded horn models.

i know little to nothing about the underlying math of horns. the peaks and notches come from the driver placement along the horn path, the size of the rear chamber for front loaded horns, and how the various driver parameters work against the air load in the horn. interestingly, too weak of a motor creates big peaks and dips, but so does too strong of a motor. the driver parameters that seem to have the greatest effect are bl, sd, mmd and to a lesser extent cms (so you can see optimally smooth horn response is going to be related somehow, i'm not sure how, to the efficiency equation). le effects upper end rolloff.

the learning curve on hornresp isn't too bad. it's a little steeper than with winisd, but given how much you know already, i'm sure you would pick it up very quickly.

ricci, another reason for using the b&c would be that you would be one of the few guys, the only that i can think of, who has built a ported and a comparable tapped horn that use the same driver. this would allow a good comparison of the enclosure vs. almost all other design comparisons that change *both* the driver and the enclosure. that muddies up the water a lot.

there is never a perfect comparo (i realize that all systems must be viewed as systems not cobblings of components), but given that you are experienced enough and credible enough to provide a good perspective on such a comparo (and the models appear to be close enough for a comparo), well...let's just say that i would be super interested to hear your opinions on how the two designs compare.

Only if the distance between the folds is at least 1/4 wavelength, and then only if the folds don't have reflectors. By the same token if the fold to fold distance is less than 1/4 wavelength no reflectors are necessary. With the wavelengths of the subwoofer passband fold sourced peaks and dips are quite rare unless the design is seriously flawed.


Not with subs, for the same reason. The wavelengths are too long for it to matter.

"So the peaks and notches are due to reflections at the fold points."

hi noah, who suggested this was the case?"

I thought in a recent post, I guess I misunderstood.

"Only if the distance between the folds is at least 1/4 wavelength..."

A horn is at least 1/4 WL long, right, so if you start at the mouth and subtract a leg at a time, mightn't this be the source of the peaks/dips?

If not, what is causing them?

Are they there in unfolded horns?

"A horn is at least 1/4 WL long, right, so if you start at the mouth and subtract a leg at a time, mightn't this be the source of the peaks/dips?"

it's not. #248 for some of the things that influence peaks and dips even in an unfolded ideal horn.

"Are they there in unfolded horns?"

yes. again #248. "the peaks and notches are there even in the ideal unfolded horn models."

Yes, they are present in un-folded horns.

I'll quote myself from another post for the rest.

"This isn't the best horn to use as an example, but lets run with it. Horns are quarter wave resonators just like pipes. Unfortunately, the flare makes the math a pain. To start off lets take the Volvotreter TH, and convert it into a TPipe.



Now the resonances are determined by the quarter wavelength of the length of the pipe. In this case 2m. Now as this is a pipe with one end open, and one end closed it is known as a closed-end air column. This means it will only have odd order harmonics. The peaks you see are these harmonics. (Most any physics book will explain a lot better than I can.) This is the math to figure out where the harmonics will be:

1st H = (4/1)*Column L
3rd H = (4/3)*CL
5th H = (4/5)*CL, and so on..

Now go Speed of Sound/H_= Hertz.

In this case we get:
(4/1)*2m= 8 ... 343/8= 42.875Hz H1
(4/3)*2m= 2.67 ... 343/2.67= 128.625Hz H3
(4/5)*2m= 1.6 ... 343/1.6= 214.375Hz H5

You can verify the math in HR:





Black is the TH, and gray is an OD.
This also shows how you trade bandwidth up top for some at the bottom in a TH.

Now if you use Volt's actual TH you get H1-36.5, H3-109.4, and H5-182.3
You will find that is is a little off, but close. This is because the simplified math doesn't take the flare rate and mouth end correction into account.

There is also something else to learn here. The dips seen are just areas were the harmonics do not yet overlap. Here is the same setup, only converted into a sealed enclosure of 100l.



The black is the OD Pipe from above, the gray is a sealed enclosure.


All horns work this way. The trick is to get the flare and size right to fill in the areas between the peaks. "

I forgot to add that the post above was using Volvotreter's Tangband 38Hz Tapped Horn as an example from a few post before that one. Only converted into a true Tapped Pipe in my post, for ease of explanation.

You can find it here about halfway down the page:http://volvotreter.de/th.htm

you are the man soho. just lifted me up one rung on the (extremely tall) horn math ladder. :-)

the same helps explain the suckouts too.

This pic says a lot.


Notice in the the Tapped Pipe goes in-phase then out-o-phase as it moves up the frequency tree passing the various resonances.

The tap when in phase fills in the dip, and when out of phase it increases it.

Later I was asked about the difference in the numbers reached with the simple equations and what HR shows for Volt's TH. The simple resonance numbers for the Volt TH are 36.5, 109.4, and 182.3 and the HR sim will show peaks at 45Hz, 120Hz, and 183Hz on the FR graph. This was my answer:

"If you set the voltage to 0 (in the HR sim) you can clearly see the harmonics of the horn without any other influences in the FR graph. They will show up as H1-43.5hZ, H3-111.5Hz, and H5-183Hz The discrepancy is caused by the flare rate and mouth of the horn. You can account for this and get closer numbers than the simple equations above, but the math is something I can't do off the top of my head. The true acoustic path length of the horn is different from the length of the parts, because of the interaction between the horn and the outside pressures at the mouths transitional boundary.

Now if you throw the voltage back into the equation (change the voltage to something other than zero in HR) the FR peaks will move around again. The horn's low corner will be ~45Hz, the next peak is now at 120Hz, with the third at 183Hz. This is caused by the interaction between the drivers two sides, and the driver/tap positions within the horn. This is also a pain to figure out in any simple way, but it can be done.

I said this was not a good example horn for just these reasons. This is more of a TL with the taps brute forced to give good results. A lot of the driver in mouth THs are like this. It helps to look at them as a Bandpass with the horn as the coupling chamber, and the driver front and rear taps as the two ports into it. The closer to a true horn, the better the peaks will match up. In any I have done the peaks are in the same positions."

soho, that pic is the one i was looking at when i wrote my reply.

all these peaks and dips make me wonder if stuffing the horn with some sort of light batting or semi-randomly spaced foam plugs could help beat them down...or even if there is a more elegant solution, such as creating a baffle with anti-harmonics based on how it is mounted and what material is used--a shock absorber for the harmonics if you will, but only at the resonances. that is probably wondering too far off topic though...

ricci, i called this post "a little inspiration?". maybe it can work here...

http://www.avsforum.com/avs-vb/showt...2#post18485532

just a heads-up, one thing to keep in mind as you model a tapped horn is the size of mouth given your minimum width requirements such that the height of the mouth will be sufficiently large to insert the driver. you are probably all over that, but i just thought i'd kick it in as it isn't directly obvious. (i'd also make the horn in the mockup a little deeper and block out a section of the throat (or use it for more throat) in order to get s1 to about 1/2 s3 than in the linked first-cut mockup). lots of options...can't wait to see what models/ideas/plans you are cooking up down there...

for a p.a. rig, that is insane.

think'n out loud here...

why couldn't you just believe me that you are not the first person to think of building a horn subwoofer ? you thought you could outsmart everybody didn't you ?

happens to me as well ...

but anyway:

Vas - 1
Ricci - 0

better luck next time !

Sorry, I had been skimming and didn't read your post close enough.

I've been a bit fixated on the idea of the length of the folds being in issue because a lot of designs just go back and forth across one of the box dimensions, which would seem to reinforce one frequency.

As far as the peaks/dips in an unfolded horn, I just have to say "duh" to myself -of coursr there are a series of higher order modes with multiples/submultiples of the 1st mode freq.

well at least we can agree on that.

That could occur if the horn had no flare. But it does.

"That could occur if the horn had no flare. But it does."

i'll be the one to ask the stupid question...why?

The pathways are of differing lengths from the wall to the end of each partition. If it was a TL with no taper you'd have issues.

may i ask for the maths?

i can handle most anything up to real analysis. ;-)

in the big picture...

"Ok. The B&C it is."

psyched! if i can provide any help from the sidelines, i'm in. if not, just as good.

I don't get that either.

Flare is handled by increasing spacing of the internal partitions; the outside dimension of the box stays the same.