Tapped Horn Experiments

There has been much talk on many of the DIY audio forums about the tapped horn. This design was recently made popular by Tom Danley at Danley Sound Labs. Tom is responsible for designing the the best performing tapped horns to date. The bandwidth he achieves in his products is as good as I have seen. It is a real juggling act between the driver and the horn to get a broad and flat response. Tom's TH115 is as good as it gets. Interestingly enough, there are many similar designs that were developed decades earlier, such as the Jensen Transflex and the JBL Air Coupler which used the 2245H driver. US Patent #5177329 shows a similar design. I would have to assume performance was poor because there are no published response plots (to the best of my knowledge) and the design quickly fell from favour. Many people are speculating and discussing the tapped horn at length, but with few exceptions in the DIY world, has anyone made some sawdust and actually built one. Through 2006 I was involved in a thesis project which attempted to accurately model the tapped horn. Many prototypes were built and many iterations of these prototypes were tried and measured. I've put together some of our findings here for discussion.

The first tapped horn we built was this 60Hz version. This cut off was chosen for economy of timber and ease of manufacture. It was initially designed around the Eighteen Sound 10W400. This driver appeared excellent for this application, and the guess was right. Out of all the horns we built, this one had the flattest bandpass and the greatest frequency range, being essentially dead flat from 50 to 260Hz.
Here the driver is being moved up the horn, away from the mouth and throat. Red - Driver is at the mouth Yellow - Driver moved up the horn by 1/8th of the line length Green - Driver moved up the horn by 1/4th of the line length Brown - Driver moved up the horn by 3/8th of the line length
This plot shows the results of two different drivers being tested in this 60Hz tapped horn. Brown - Eighteen Sound 10W400 Red - Etone model 135 painted cone.
The second tapped horn was essentially a scaled up version of the 60Hz horn. A more suitable driver was chosen for the frequency range of interest. This horn was used as a subwoofer in my main home system for a while, and now resides in my second system. The output this horn is capable of in a package one person can carry is quite astounding.
This plot shows the results of two different drivers being tested in this 30Hz tapped horn. Red - Peerless 830500 Brown - Peerless 830515 (I now believe this measurement is not correct)
Inside the 30Hz Tapped Horn. No bracing in this picture.
Anyone who wants to make a great little tapped horn similar to this one should make a few changes. The slope of the baffle should be increased so that the throat end measurement is 48mm, giving a 150cm^2 throat. Combined with a Peerless 830564 XLS driver, I measured a flat response from 30 to 100Hz and more than 130dB output with 400W of drive, when firing from a corner. In half space the maximum output will be around 122dB at 1M and 400W. To get a feel for what this is like, it would take four 830847's in a 350 liter sealed box driven with 1200WRMS to achieve the same output as this 165 liter box with one driver at 400WRMS. The tapped horn has more output than the four driver sealed box right down to 20Hz. The Peerless 830564 is a very suitable driver in a 30Hz tapped horn, with it's 49Hz resonant frequency, good excursion and strong cone assembly. Here's a photo into the mouth of the 830564 version of the 30Hz Tapped horn. The driver will have to be mounted this way around to fit the magnet in.
Looking in the mouth of the new 30Hz Tapped Horn.
The 18Hz, 30Hz and 60Hz Tapped Horns

Predicted V's Measured Performance of Tapped Horns

There has been a lot of discussion on the diyAudio forums on the Collaborative Tapped horn project about modelled versus predicted response of Tapped Horns. In this example I will show a model of a 35Hz tapped horn, and compare it to the measured response.
Here is the predicted response for this 35Hz Tapped Horn. The 56V RMS input power level was chosen because this corresponds with the power available from the power amplifier that drives these Tapped Horns.
A drawing of the construction of this Tapped Horn. It is a simple single fold horn with a straight sloping driver baffle. No stuffing or lining was used in this construction.
The measured response of the Tapped Horn. This measurement was taken with my IMP/MLS using no smoothing, taken at the mouth. You can see that the measured response of this tapped horn corresponds very well with the predicted response generated by HornResponse. The Q of the out of band resonances is slightly lower than predicted. If the horn path was lined with absorbant foam, or similar, the Q of the out of band resonances would have been even lower, also reducing their amplitude. I chose not to line the horn path because the desired final response could be easily obtained with the DCX2496 that was doing crossover duties. In use I was unable to get more than 12-15mm of peak-peak excursion from the driver when driven with an amplfier that was close to clipping at peaks of more than 500W RMS. When trying a particularly brutal Slayer track with lots of very fast kick drum the smell of hot voicecoils became quite obvious. The thermal limits of the driver became a problem well before the 25mm peak to peak xmax.

Modelling a Tapped Horn

By far the easiest way to accurately model a Tapped Horn is to download and use the excellent HornResponse program written by David McBean. He has invested a huge amount of time developing what is arguably the best horn loudspeaker modelling software available, and it's free! If you want to incorporate features into your design that HornResponse doesn't allow, you will have to resort to a package that is a litte more flexable such as AkAbak. Many thanks to David for showing me how to simplify the AkAbak script for a tapped horn. You can copy this script and paste it into AkAbak and adjust the constants for the horn and driver to suit your proposed design and then run the simulation. If you model a few different drivers in the 30Hz tapped horn, you will see that several perform better than the 830500 used in this model. HornResponse will also allow you to export files for use in AkAbak.

| 30 Hz Tapped Horn

Def_Const |Horn Dimensions

{
a1 = 315e-4; |Area at throat (cm^2)
a2 = 320e-4; |Area at rear of driver (cm^2)
a3 = 940e-4; |Area at front of driver (cm^2)
a4 = 945e-4; |Area at mouth (cm^2)
l1 = 15e-2; |Distance from throat to rear of driver (cm)
l2 = 350e-2; |Line distance from rear of driver to front of driver (cm)
l3 = 30e-2; |Distance from front of driver to mouth (cm)
}

Def_Driver 'Dr1'

| Peerless 830500

Sd=466cm2
fs=18.1Hz
Qes=0.21
Qms=3.7
Vas=139L
Re=3.5ohm
Le=4.2mH

system 'S1'

Driver Def='Dr1' Node=1=0=3=4
Waveguide 'W1' Node=2=3 STh={a1} SMo={a2} Len={l1} Conical
Waveguide 'W2' Node=3=4 STh={a2} SMo={a3} Len={l2} Conical
Horn 'H1' Node=4 STh={a3} SMo={a4} Len={l3} Conical

Back to my home page