A Horn Subwoofer


Redefinition of the bookshelf loudspeaker


From when I became a horn enthusiast I have always dreamed of a subhorn. Real horn loading down to 25-30Hz or lower. This would be a crazy project, no matter how it was done. It would be big, really big. But I had made up my mind, one day I would do it. I just had to try...

July 2003: I moved into a new house. The time had come!

Other subhorns

A few people have subhorns. An interesting project is described by Thomas Danley, the LAB sub at Live Audio Board, see http://www.prosoundweb.com/. This sub was developed by specifying a horn, and then making a driver that would work in that horn the best way. This sub requires a special driver, but is smaller than many other horn subs.

Bert Doppenberg of bd-design has two 3.7m long subhorns in his listening room. (Click DIY Products -> Articles -> Bass Horn Design).


driver thumb



For many months I did simulations in David McBean's Hornresp horn analysis program. It can be downloaded here: Hornresp. I tried lots of drivers, horn lengths, expansion types and front/rear chamber sizes. It is really hard to make a horn go below 30-35Hz, especially if you want to get it domestically acceptable. For a 2.5-3.5m long exponential horn, the lower limit seemed to be about 30Hz. That is pretty low, as low as my old sub. But if this thing is so big anyway, why not go deeper? I wanted to squeeze as much as possible out of this design.

I considered the LAB sub, but it needs a special driver.

Then at last it seemed like I had found a nice horn, using a 15" driver from Sammi. But then something happened: ELTEK had an "offer of the week" on the most powerful Sammi 12" drivers, the SR-12A300. Two of them in a 4m long horn gave very good results...


hornresp2 thumb

fig. 1


horn1 thumb

fig. 2

horn2_thumb.gif (6325 bytes)

fig. 3

mouth thumb

fig. 4

The horn

The horn I ended up with is a hyperbolic-exponential horn with the following parameters:

Ath: 350cm2, Am: 6500cm2, length: 400cm, Fc: 15Hz, T: 2, Vrc: 100 liters, Vtc: 7 liters.

The simulated response is very good, look at fig. 1. At 40Hz, efficiency is 107dB/1W/1m. If we use this as the reference, -3dB is 25Hz and 300Hz. However, the horn is huge, about 1200 liters... the only way to make it domestically acceptable was to disguise it as a piece of furniture. Since I was short on shelves, a logical choice was to put shelves on the horn.


Building the horn

It wasn't until I started cutting the panels and glueing them together that I realized how huge the horn really was. And it grew bigger and bigger... It is (outside measurements) 208 cm high, 185 cm wide, and 65 cm deep, all in all (I can stand upright inside it!), but not all of it will be used for the horn. It took 3 people to lift it up to an upright position, it is quite heavy... I don't know how much it weighs, but it is most likely more than 100kg.

Fig. 2 shows the first steps. The front panel lies on the floor, and the inner boards are glued to it.

Fig. 3 shows a friend of mine nailing one of the angled boards inside the horn. The picture gives a measure of the size of this thing.

Fig. 4 shows the horn mouth without the bracing. The shelves will be in the space at the right side of the mouth.


Objective; measurements:

Not much to put here yet, but it measures down to 30Hz with a small peak at the 15Hz flare frequency, and another one at about 70Hz, which I have EQ'ed flat. It seems that much of it is a room mode. I will put up some graphs later I hope. One reason that it doesn't go down to the simulated 25Hz is maybe that it's not in a corner. I hope to do a simulation of the horn as built to compare with the measured results.


This is probably the most interesting part for most of you. How does this monster sound?

Accuracy: A subhorn does not have the slow response of  a vented or bandpass enclosure. The bass is tight, precise and articulated. No matter how loud I play, it never loses control. A kick drum has the right snap, the bass guitar can growl. It feels real, not like something reproduced by speakers...

Impact: "I feel the earth / move / under my feet"  When two heavy duty 12" woofers work into a horn of this size, you really get the air moving! This is the most physical bass I've ever had. I can feel it even at low volume settings. When I turn up the volume, it really shakes the floor.  And when I play really loud, it feels like someone is lightly kicking my chair... It is possible to sense the resonant frequencies of the floor. The bass is enveloping and real. Size does matter!

Distortion: The bass never gets muddy or compressed. No matter how loud or complex, it keeps going. This is perhaps the greatest difference from the old bandpass sub. At last, I can turn up the volume without the bass getting left behind. The feeling that nothing limits the SPL enhances the realism of reproduction, it is the natural, satisfying experience of increased bandwidth with increased volume. Usually, small woofers start to limit the level at 25-80Hz at around 90-95dB, which means that if you turn the volume further up, there will not be more bass, in addition to increased IMD. A separate subwoofer helps, but even a 15" driver has its limits. Of course, this horn also has limits, but they are in the 115dB range at 25Hz, 120dB at 40Hz and 135dB at 60Hz, Xmax limited. That is enough for me...

Was it worth it? Definitely! I would not trade a real subhorn for another sub, this is by far the best performance I've heard. As physical as PA bass, tighter than anything I've heard, controlled and detailed. Big, yes, but good!

4 years later

I have now had this horn for nearly 4 years, and it still provides impressive performance. But I can no longer say it is the best performance I have heard. About a year ago, I designed a horn subwoofer for sound reinforcement, and even if it doesn't go as deep as this horn, the response is smoother, and it just sounds more “right”.

Since I built the big sub-horn, I have learned a lot about how horns work, and as a result, I can now see several flaws in the design of this horn:

  1. Even if it's big, it is acoustically small compared to the wavelengths it is intended to reproduce.

  2. The mouth is too small and the flare rate too low, resulting in a behaviour similar to an organ pipe.

A horn that would out-perform this one with ease, using the same drivers, could be a 4.5m long exponential horn with 2 sq metres mouth and 350cm2 throat areas, and an 80 l rear chamber volume. This would give some 3dB increase in low-frequency efficiency, and a much smoother response, and a -3dB of 28Hz when mounted against the wall/floor boundaries, and 25Hz when corner mounted (where it would also have the required mouth size).

If you want to build a good horn subwoofer, there are a few things to keep in mind:

  1. If you want smooth response down to a low frequency, free of resonances, the horn will be big – very big – and should preferably be built into the house.

  2. The mouth should have a circumference close to one wavelength at the cut-off frequency, and not less than 0.7 wavelengths. Too many bass horn designers shorten the horn too much, and this invariably results in a resonant response. It is still far better than the usual vented or band-pass enclosure, but it doesn't take you anywhere near the performance of a full-sized horn.

  3. Use an exponential or hyperbolic-exponential (hypex, T <= 1) flare. Conical horns of equal performance will be more than 30 times larger. Small conical bass horns will act like tuned pipes, and will not give the full performance of a well-designed exponential or hypex horn. Plane-wave propagation through the horn can be assumed if the horn is placed in a corner or at the wall-floor intersection, this means that the cross-section of the horn can be calculated according to the known formulas for hypex horns without correction.

  4. If you want a horn that is not extremely big, select a higher cut-off instead of shortening the horn too much. Very good performance in a slightly limited frequency range is to be preferred over mediocre performance in a larger frequency range. Smooth power response and good output down to 45Hz will give a greater satisfaction in the long run than a peaky response down to 25Hz.

  5. Use Hornresp to design the horn. Use first the Hypex designer tool with the driver of your choice, then tweak the design to optimize performance. A rear chamber slightly larger than that suggested by the Hypex Designer tool will usually result in a smoother response. If the rear chamber is too large for your liking, select a driver with higher Vas. Check the acoustical impedance chart, and avoid designs with large peaks and troughs in the impedance. This is the load for the driver, and large peaks and troughs here indicate strong resonances in the horn. This may not be so obvious in the SPL response if the driver matches the horn well, but will affect the subjective performance. A hypex horn (T < 1) will have one large peak close to cutoff, but subsequent peaks should quickly diminish in amplitude.

Good luck!