Collaborative Tapped horn project

GM:

Absolutely no problem at all. I see how given the tools available, a quick build to an excellent product is normal. Most of us will opt for using these tools as it is by far the most expedient. For some undetermined reason I've never been one to take this practical approach. Nope, I've got to beat on things until my brain throbs uncontrollably.

I do not have the resources to come up with the suggested drivers I keep reading about. Best I can do is work with the cast offs I find while dumpster diving in Ebay. As a result, I find my satisfaction in the theoretical, rather than the substantive. What I have discovered is that reasonable driver, coupled with a good design, makes for an enjoyable result. It's that elusive good design that I'm in pursuit of. To get to that good design, I need to know how to best use the tools.
 
diyAudio Moderator
Joined 2008
Paid Member
I think people put far too much on what driver they use, ie: its reputation. I understand you want certain specs for your tapped horn, that's obviously not what I'm talking about though. I'm currently getting some great sound out of some embarrassingly cheap drivers as subs.
 
Here's a quick sketch of what I am trying to model . the 2 spkrs front load into a common cavity (what I called the horn) The rear of each spkr is kept separate from each other, as I want to investigate the effects of different resonant lengths being summed inside that common cavity.

I hope this makes sense.
OK, this is how I would start.

I have marked the nodes on the picture to make it a bit easier to understand. The thicker lines outline the individual segments, and the thin line is the distance to use. The green segments are ducts, red are waveguides, and the purple two are ducts or waveguides depending on whether there is any expansion to them in the dimensions not shown. The gray line is the radiator.

This "simple" model will not take the drivers physical displacement in the horn path into consideration. As that is a harder nut to crack.

Hope this helps.
 

Attachments

  • example.jpg
    example.jpg
    29.9 KB · Views: 419
Try this example:

http://www.diyaudio.com/forums/subwoofers/145603-tapped-horn-car-6.html#post1936713

(I think there are some errors in the script provided, though.)

The essential point is to understand that AkAbak allows you to split and merge ducts and waveguides. Remember to match the areas - if you are joining two 100 cm2 mouths to a single throat, the throat has to be 200 cm2.

Here's a basic 4 segment tapped horn schematic using Hornresp-like areas (S1 to S5) and segment lengths(L12 to L45). Exporting a design from Hornresp should produce a similar schematic.

Code:
|            -------Driver1-------
|           |                     |
|S1<--L12-->S2<--L23-->S3<--L34-->S4<--L45-->S5-- Radiator

With example AkAbak node numbers added:

Code:
|            -------Driver1-------
|           |                     |
|S1<--L12-->S2<--L23-->S3<--L34-->S4<--L45-->S5
|11         12         13         14         15-- Radiator
|------------------------------------------------

Define all the horn segments with the "waveguide" element.
(A "Horn" element in AkAbak is a "Waveguide" with a "Radiator" added to its mouth.)

Now add a second horn in parallel:

Code:
|            -------Driver1-------
|           |                     |
|S1<--L12-->S2<--L23-->S3<--L34-->S4<--L45-->S5
|11         12         13         14         15-- Radiator1
|------------------------------------------------
|21         22         23         24         24-- Radiator2
|S1<--L12-->S2<--L23-->S3<--L34-->S4<--L45-->S5
|           |                     |
|            -------Driver2-------


This should generate the same SPL curve, but 6 dB higher in output.

Finally, merge the 2 horns:

Code:
|            -------Driver1-------
|           |                     |
|S1<--L12-->S2<--L23-->S3         |
|11         12         13         |
|----------------------S3<--L34-->S4<--L45-->S5
|21         22         13         14         15-- Radiator
|S1<--L12-->S2<--L23-->S3         |
|           |                     |
|            -------Driver2-------

Remember to make S3 (for the L34 segment), S4 and S5 equal to twice the original areas.
The SPL curve should match that for the previous version.
This script simulates the drawing you provided earlier of the two horns oppposite each other with a shared mouth area.

You can now make the lengths of the 2 segments in the "second" horn different from the corresponding sections in the "first" horn.
 
soho54,
Thanks, I think I've got it based on how you defined the nodes. What I think is going on is, as a snippet:

.
.
.
.
Waveguide 'W3' Node=10=150 .........
Waveguide 'W4' Node=100=150........
Duct 'D5' Node=150=151.......

And, yes, I am radiating from the front, not down.

I guess what I was missing is setting up node 150 for WG3 and WG4 may have an area of 1000 each but D5's node 150's area could be at 2000 (example only). The ducts I used originally were intended to keep the areas constant, that was the only use. For whatever reason I assumed the areas needed to be constant

I tried this out in AkAbak with success.

Knowing this is going to go a long ways in my modeling. Thanks!!!!!!!!!

I have yet to look, so I should be quiet until I do, but while here I can't help but ask about modeling microphonics. I'd like to introduce a set of drivers with a different bandpass into the same chamber as the 2 original drivers

BTW, what did you use to mark up the pdf?
 
Don Hills, Thanks for the post. Must have come in as I was writing response to soho54.
You directly stated what occurred to me while looking at soho54's markup : the areas of multiple outputs can be summed into a single input. That is HUGE.

Now I will go back and read balance of your post.

Finished reading.
Yup, I got it. Makes sense. I can mash things together any 'ol way I wanna knowing this fact.


The technique of doing the 180 (or whatever angle) is to create a series of waveguides apparently. Is there a rule of thumb for how fine to divide?
 
Last edited:
You don't have to fold / divide when you're simulating in Hornresp or AkAbak. Design it as a straight horn, then fold where required when you build it. Folds in a tapped horn have no significant effect on the response within the intended (bass) passband. They only affect the response at high frequencies.

I've tried it both ways in AkAbak - a simple straight horn with 3 segments, then a 15 to 20 segment model of the actual horn as folded and built. I carefully modeled the bends in a simliar manner to soho54's picture to allow for the changes in area as the bend is traversed. There was no significant difference in the results in the passband. Remember that a tapped horn only works properly over a 2 to 2.5 octave range at best. (40 to 100 Hz, for example.)
 
Thanks Don,

The very little I've played with I see less than 2 octaves of useable passband. That's ok, as I am looking very seriously at narrow passbands. I want to emphasize distortion minimization so small xmas with narrow passbands seemed to be called for. The Big concern is the interaction between passband drivers. I want a common radiator for as many drivers as possible, Helmholtz convinced me of that. However, what affect will 4 thumping 12"er's going to have on my 4" er's. I have a LOT to look at.
 
Look at Horst Moller's double horn designs:

Hornlautsprecher

They are mostly back loaded horns, but they do use the principle of two differently tuned enclosures working together.

Using a common mouth for two tapped horns doesn't save you any space, you still need just as much box volume. And if they are tuned to different ranges and their mouths are close together or common, you will get slightly worse performance because the output of one horn will be partly out of phase with the other horn at some frequencies, causing cancellation.

The only real advantage of having two overlapping passband tapped horns is that if you set the resonant frequencies properly, the electrical impedance peaks can be made to "interleave". This flattens the overall impedance curve seen by the amplifier, which can be useful when using small tube amplifiers etc.
 
BTW, what did you use to mark up the pdf?
I took a screen capture of the pdf, and then drew over it in MSPaint. Fast, dirty, but it works.

: the areas of multiple outputs can be summed into a single input. That is HUGE.
You can also vary the areas too. Say two 100cm^2 waveguides/ducts merging into a 400cm^2 waveguide/duct. The pressure change can do interesting things sometimes under the right circumstances.


The technique of doing the 180 (or whatever angle) is to create a series of waveguides apparently. Is there a rule of thumb for how fine to divide?
You just use the 90deg lines, and then radians to the point halfway between them and the corner. With lesser degree bends it gets a little looser.

It is used for simulating from a real world build, or a finished horn from plans. That is why I used it there, habit. ;) It isn't for the initial design work.

It is normally used to check your build plans after you have converted your original model into 3D/2D build plans, to make sure you did it right. A lot of people have problems bending a horn, and don't catch that something is off until after it is built, and tested. Most then still blame the software to varying degrees.

Good luck.w
 
Using a common mouth for two tapped horns doesn't save you any space, you still need just as much box volume. And if they are tuned to different ranges and their mouths are close together or common, you will get slightly worse performance because the output of one horn will be partly out of phase with the other horn at some frequencies, causing cancellation.

In the very limited amount of time I played with this idea, I was able to create a complete mess to what I thought to be an improvement in character. Try the script I originally put up. The ducting removal and conversion to the proper way actually didn't alter anything regarding fr curves.
I understand that size of cabinet is not affected with removal of the center baffle. Hoping for sound to become more single point over broader range, instead of multiple sound sources.
 
... Hoping for sound to become more single point over broader range, instead of multiple sound sources.

For bass, you may find that multiple sources are preferable to single source.
The wavelengths of bass frequencies are so long that you can place the speakers well apart from each other and still not be able to localise the source. There are several threads here on diyAudio which are devoted to the art and science of using multiple subwoofers spread around the room.

Having two or more subwoofers, each designed to cover a slightly different range, can be better than two identical subs. For example, your room might have resonances at 50 Hz (length) and 90 Hz (width). If you put a sub in a corner, you get a 50 HZ resonance and some 90 Hz resonance. If you put the sub on a long wall, you lose the 50 Hz resonance but the 90 Hz resonance is worse. If you have two subs covering 30-60 Hz and 60-120 Hz, you put the 60-120 Hz sub in the corner and the 30-60 Hz sub on the long wall.
 
Isn't there a saying about not being able to see the trees for the forest? Here I am trying to fill in valleys and flatted peaks by arranging phase within the enclosure. Need I go on? I do recall seeing a web site or 2 that has a java script calculator showing nodes and antinodes within a room for a given speaker/listening position. Given that and my trusty rta, I should be able to get data on each speakers' location.
 
Arta and Steps each have the ability to provide distortion information, based on input data from a microphone input. This is end of process stuff, though

Mathematically, there is a Volterra Series which can be used to predict distortion components based on a non-linear transfer function. This can be handy during the design phase. Difficult to use properly, however.

Is there easily obtainable a piece of software that would run a theoretical distortion analysis assuming that I can supply a "reasonable" (non-linear!)function as input? Is there a Spice module that might pull this off?
 
Questions using HornResp and EasyHorn

I used EasyHorn to calculate some data to enter into HornResp.

Things didn't go so smoothly, so some questions came up which I haven't been able to find answers to in the help file. Always the way, right?

I have screen shots, but attachment upload size restrictions are killing me. It's a pdf of about 800k. I can break file apart into multiple files, but somehow that seems to go against the spirit of the original intent of file size limits. Someone willing to snag this file as an email?

In a nutshell, I am getting beat up when entering data in the sequence outlined by EasyHorn. HornResp is complaining when I do a 3 section input, wants something other than 0 for S5( not used of course in 3 section), it complains in a different manner when I do a 4 section input, S4 is > than 1 (but its the mouth for crying out loud!), Loudpeaker Wizard doesn't have "Set S2 to variable" option.

Little things I'm sure, but being the newb, not sure what to do next.
 
Any problems phasing Tapped horn subs with upper freq system?

Since THs have so many internal phase interactions are there any problems mixing with upper freq systems? Is it a matter of time delay? Imagine the upper system with a midbass driver whose voice coil is time aligned with the TH driver. Where do we go from there?
 
Help setting up HornResp for TH

I included an attachment .doc showing the error message I get on Hornresp. I am following instructions on EasyHorn. I show both windows. I go into wizard, change size of s2 and s4 with same results. ??? What fundamental error am I making?
 

Attachments

  • Untitled 1.doc
    103 KB · Views: 60