Tapped Horn For Car

[judtoff]

I have a fair understand of horns, and a better understanding of Car audio. That said, I've seen(heard) rear 6X9s capable of over 120dB on a Term Lab(the standard mic for car audio). They were amped of course, at about 50Wrms each.

And I really doubt there is any acoustic impedance matching being done by a box that small, at least not at sub frequencies.

You cannot compare your model of what I suggested to your results. Apples to Oranges.

To make it more fair, either measure them both and compare them, or develop an accurate half space model for both and compare them.

2cuft is way too big for a pair of those sealed anyways...

Either way keep up the good work

Originally posted by mwmkravchenko


Hi there judtoff

I have tried it and here is a post to see what you get with the same drivers in a sealed box. Lets make it 2 cubic feet or around 57 litres. It's easier to use a computer to knock about designs than to make them and then measure them. Power in is 50watts.



Your maximum level is 105 db no matter how power you pump in. It's limited by the cone area and the stroke of the driver. A horn or in the case of the UNHORN gives you a more efficient coupling to the air around the box and in the room or in this case the car. To understand it simply think this through. If you ever tried playing a trumpet or some other brass instrument or woodwind you don't make alot of noise in the mouth piece. But put the mouth piece on the horn and voila you have a lot bigger sound. Same principle with a horn loudspeaker or a Tappered Quarter Wavelength Resonator which is what the UNHORN really is.

If you want lower (as I do) keep watching . I'm working in the shop right now on a set of three. Home version of the unhorn. True TRIO 8 18hz tapped horn and dual SDX7 18hz UNHORN box. Have to finish this stuff eventually.

Mark

EDIT

And if that graph is in half space, in my experience with hatchbacks, you'll get roughly a 12dB/octave gain from 60Hz down, and a static 9-12dB gain across the board. So if that graph is of it's half space response, then the sealed box could very well be louder in that hatch. Try it


Also those two drivers are tapped at different lengths on the horn..."If you ever tried playing a trumpet or some other brass instrument " with two mouthpieces at different entries into the horn path? Nope haven't tried that one...But playing Brass and Woodwinds with traditional mouthpieces, yes I've done that.

[mwmkravchenko]

And if that graph is in half space, in my experience with hatchbacks, you'll get roughly a 12dB/octave gain from 60Hz down, and a static 9-12dB gain across the board. So if that graph is of it's half space response, then the sealed box could very well be louder in that hatch. Try it

Hi Judttoff

Your right 2 cubes is big. But it gives a decent low end. and the graph is for illustrative purposes. Not a serious closed box design to optimize size. No mater how you change the box you can't get a lot louder. Whereas the horn gives you an extra 8db of gain. Go figure the power return on that. If you start SPL at say 10 watts you need 10 * 2 = 3db (20Watts) +
20 * 2 = (40Watts) + 40 * 2 = 80 watts. It is an illustration of what gain from a properly designed box can offer. The output of 80 watts power for an input of 10 watts. Not to bad.


In a 1/8th space the graph would go up by 3 db overall. You will gain some loading in the low end as you state. You do know something about car audio!

The point of the horn design is to get the most bang for the buck. There is no free lunch. Hi efficiency means narrow bandwidth. Therefore you can get 108db/watt in a car environment or 1/8th space.

But you only get usable output from 35 to 140 hz. Two octaves. Those two octaves are where most bass music exists and that's why it was designed this way. It is also a tiny box. Again part of the design goals. I have fully horn loaded designs that are a fairbit bigger and even more efficient. But this is not the pupose of this thread.

The impedance matching is evident in the change of Fs and the second impedance rise. You get the low end resonance of the horn and the upper resonance of the horn caused by the driver at the second tap point. I didn't post those graphs but if you know your horn design you will understand this concept. The driver out of the horn measurement was shown and the driver in the horn was shown. As was a detailed 5hz increment SPL measurement. Without any smoothing or marketing crap. There is nothing to hide behind smoke and mirrors. This really works and runs well of my head unit. No amp. Another reason for wanting high efficiency.

Cons? The drivers are not cheap. But they perform very well. If you want to pour in the power you have already quoted the max SPL. There have been a couple of cheaper alternatives posted. They do compromise the design in terms of output. But they are considerably cheaper. If you want to try it go ahead.

Otherwise thankyou for your encouragement.

Mark

[judtoff]

The thing is, a car is Not eighth space. Again, instead so speculating that your so called horn is more efficient than the sealed box, test it out. Prove me wrong. Preferably use a .5cuft box with the two drivers to get a Q of .7. It's not that much wood, or time to build it.

I've looked at your measured impedance graph, all I see is a 6th order bandpass. Not really any horn action going on.

We need some response graphs in half space of your horn, to compare with the Hornresp response, I bet they are nothing alike. Then you could compare it fairly with other designs.

[Don Hills]

Mark,
First, an apology: I thought you used 3/4" HDF, but you used 5/8". You have a picture in post #31 with a ruler showing the thickness, and you also mentioned the thickness in post # 72. As well as "measure twice, cut once" I also need to "read twice, post once".

Second, I have managed to draw a plan of the box that matches your photos. Lots of orthographic correction, printing and measuring...

Armed with the measurements off the plan, I have modeled both your original Hornresp design, and the "as built" box, in Akabak.

The original design model results match closely between Hornresp and Akabak. I would expect this, because Hornresp can export a design suitable for importing in Akabak, but it was a good sanity check. (Note that Hornresp will not export an Akabak model of this design because Hornresp models converging horn sections differently than Akabak.)

The "as built" model, however, produces results very different than the design model. The peak you noticed when measuring in your car is there too. It appears that the compromises you had to make (folds, flat panels, non-coincident drivers etc) have had their effect.

None of which is news, of course, and not intended as criticism of you. I emphasise that I would not be able to do better under the same constraints. All kudos to you for building it, we can model until we die but that is no substitute for making sawdust and seeing how it works in the real world. I look forward to your experiences with the new 2x DX7 box, it looks as if it will require fewer compromises.

I've learned a lot from this thread already, I hope to learn much more in future. It spurred me to research widely, and I even found a technique for cheaply and easily producing rectangular cross section horns with curved sections, which would have eased your design constraints. I'll post it in a new thread and post a pointer to it.

Regards...

[mwmkravchenko]

The "as built" model, however, produces results very different than the design model. The peak you noticed when measuring in your car is there too. It appears that the compromises you had to make (folds, flat panels, non-coincident drivers etc) have had their effect. None of which is news, of course, and not intended as criticism of you. I emphasise that I would not be able to do better under the same constraints. All kudos to you for building it, we can model until we die but that is no substitute for making sawdust and seeing how it works in the real world. I look forward to your experiences with the new 2x DX7 box, it looks as if it will require fewer compromises.

Hello Don

You sure went through a great deal of work. Probably more than me even!

You have uncovered the difference between a simulation and a prototype. As measured in post #102 by Adam ( Binary110) and his cousin:

The bass is pretty much as flat as i have heard it. Sittin in car me and the other guys were lookin at the SPL meter playing test tones in increments from 40hz to 100hz, the dB meter never swayed more than 2 dB up or down from our baseline. The car did have a hell of a resonance around the 75-80hz range, the car was rattling a bit, but thats got nothing to do with the sub, but the cabin size/resonances of the Toyota Matrix.

My measurements were done i the car with a different setup. The back seats were up the mouth of the horn was shooting into the middle of the car. When the three of us took a whack at it I had one seat folded down to get access to the box, the mouth was pointed to the rear hatch and the response was almost flat. Obviously moving around in any measurement space will affect the results of the measurement.

But in your defense Don the design is not totally faithful to the proper layout. The last three that I did have been almost exact replicas. But they are larger as you alluded to. A box this small is not easy to create. But it could be tweeked a bit to match almost precisely. Hindsight is always 20/20. I redid the design using conic paths only and it makes it much easier to work with. In a horn that handles only bass there is no real life benefit to using curves. There are only benefits up and above 350hz. Not a hard and fast rule but it has proven true for the most part.

Now you have me interested in whether or not you drew the design as I have. I found the original full scale template. So other than the 1/8th inch difference on the thickness of the material it is the original layout.

It is probably a fair statement that this design has been one of the few thoroughly measured and reviewed horns in recent memory. And a few gents have been there to push things in the right direction despite my being a bit naieve on the software end at times. GM and Bjorno are owed a debt of thanks and Don you are the only person I know of who took the time to do some real thinking. Good on you man!

Mark

P.S. Looking forward to your thread Don

[Don Hills]

Originally posted by mwmkravchenko


My measurements were done i the car with a different setup. The back seats were up the mouth of the horn was shooting into the middle of the car. When the three of us took a whack at it I had one seat folded down to get access to the box, the mouth was pointed to the rear hatch and the response was almost flat. Obviously moving around in any measurement space will affect the results of the measurement.

You're right. Given the small dimensions of a car at the wavelengths concerned you would not think a few inches would make a difference, but they do. I've seen 10 dB difference, measured at the standard SPL contest measuring microphone location, caused simply by rotating the subwoofer enclosure 90 degrees. We got 12 dB more in a small hatchback by moving the sub from its forward facing position so it faced into one of the back corners. It sounded much worse for music, but music has nothing to do with SPL contests...

In a horn that handles only bass there is no real life benefit to using curves. There are only benefits up and above 350hz. Not a hard and fast rule but it has proven true for the most part.

Look at the bottom corner of your box, after the decreasing section turns the corner. It expands to almost twice its size before starting to decrease again. I'll tweak my model so that the size does not increase, then see if it improves the performance. If so, it indicates that using curves and fillers might be advisable in small boxes. In other words, bends / corners in long, thin pipes are not a problem. Bends / corners in short, fat pipes might be a problem.

Now you have me interested in whether or not you drew the design as I have. I found the original full scale template. So other than the 1/8th inch difference on the thickness of the material it is the original layout.

I'll tidy up the drawing, scan it and post it along with the Akabak scripts.

It is probably a fair statement that this design has been one of the few thoroughly measured and reviewed horns in recent memory.

Indeed it has. This is because it challenges our preconceptions - we "know" what a tapped horn should look like, then along comes the Unhorn. It makes us think " how does this work?". I know I wouldn't have attempted such a design because I "just knew" it wouldn't work. But then I wouldn't have learnt as much as you have. Your tag line at the bottom of your posts sums it up well.

[mwmkravchenko]

Hi Don

It really is a strange thing to listen to such a small box produce bass. I modeled it and thought exactly the same thing. It can't work right? But then the urge to give it a shot won out. It is not really a horn in some ways. But in a confined space it behaves as modeled. Go figure. And it sounds good to boot!

I encourage people to give it a try you may be surprised. And as shown there is an entry level version possible.
Mark

[Don Hills]

Okay, here we go.
First, the picture of the box plan, as I measured from your plan and your photographs. I settled on about 11/16 (17 mm) as the panel thickness - your photo with the ruler showed 0.68 inches.
The 5/8" HDF appears to have a laiminate surface which would account for the difference. Using 17 mm, everything fell into place.
I know it seems pedantic to be worrying about millimetres at these wavelengths, but they do add up by the time you get to the throat of the last sections.

[Don Hills]

Next, the "horn plan", in other words the horn laid out straight, Hornresp schematic style:

[Don Hills]

Now, here are the SPL plots from your original Hornresp model and the corresponding Akabak model. Note the similarity.

Also, here is the Akabak script corresponding to your original model.

|----------------------------
Def_Driver 'CSS SDX7'
SD=128cm2 |Piston
fs=34Hz Mms=21.9g Qms=2.2
Qes=0.41 Re=6.4ohm Le=0.75mH ExpoLe=0.618


System 'Unhorn As Designed'

Driver 'D1' Def='CSS SDX7' Node=1=0=11=13

Driver 'D2' Def='CSS SDX7' Node=1=0=11=13

Waveguide 'W1' Node=10=11
STh=85cm2 SMo=556.21cm2
Len=24.2cm T=1

Waveguide 'W2' Node=12=11
STh=21cm2 SMo=556.21cm2
Len=42.8cm T=1

Waveguide 'W3' Node=12=13
STh=21cm2 SMo=302.28cm2
Len=16.4cm Conical

Horn 'H4' Node=13
STh=302.28cm2 SMo=305cm2
Len=0.1cm Conical
|----------------------------------------