Tapped Horns and the Dr Who Experience

[PrestonTom]

Dr Who (Mike) and others who are interested,

You had done a DIY clone of a Tom Danley inspired tapped horn subwoofer, As I recall the measured response did not look as good as the Hornresp simulation, but you attributed part of that to some room interactions during the measurement.

For those of you unfamiliar with this, the design capitalizes on horn loading (to an extent) both the back wave and the front wave of the driver. The mouth is undersized to a practical dimension (which leads to ripples in the spectrum) and the two path lengths are asymmetric (which counteracts some of the preceding problems). All of this (plus the judicious use of resonators inside the cabinet) has the net effect of decent sensitivity, and great low frequency extension in a box that is a manageable size.

What was the conclusion? Is it a keeper? Are there obvious mods you might make on version 2?

As I recall you used drivers that were chosen due to convenience. I have looked at some of the simulations and concepts and I think it is a very, very intriguing approach. BTW, a store-bought version will cost a few thousand. In my application, I am trying to give some bandwidth to some Martin Logan electrostats that can't let go.

-Tom

[soundbound]

Have you looked at this man's website? Some information there on tapped horns he built.

http://www.cowanaudio.com/

[PrestonTom]

Have you looked at this man's website? Some information there on tapped horns he built.

http://www.cowanaudio.com/

Yes, and thanks.

I have followed his projects with great interest. There is long series of threads over at DIYaudio on his (and others) projects. It is what go me interested. Also Chris (on the Klipsch forum) installed the Danley DTS-20 sub (Danley was an originator/implementer of the tapped horn design). The reviews were very positive. However, the store-bought version is rather pricey. I am interested in an affordable DIY version to supplement some electrostats that I have. I am also interested in the design and construction issues. I think this may end up being a great solution for a number of folks. Besides, you get to play with Hornresp when you do the simulations.

-Tom

[Don Richard]

Tapped "horns" are interesting devices but are combination enclosures that behave more like transmission line enclosures down low. LF distortion is higher than with an exponential horn.

[PrestonTom]

Tapped "horns" are interesting devices but are combination enclosures that behave more like transmission line enclosures down low. LF distortion is higher than with an exponential horn.

Yes, the distortion is there (although most do not post those results). But wrt to an honest-to-goodness horn, you now have a cabinet that is a much more managable size. As usual, there is no free lunch.

-Tom

[DrWho]

Oh cool, tapped horns - are you seriously looking to build one Tom?

Have I posted the data on the forum before? If not, you might find this interesting...

Here's the measured response. It's an average of two different mic positions in my room and smoothed to 1/3 octave. The horn was in a psuedo 1/8 space and I tried to pick two different mic positions that would minimize the effects of the room's modes when averaged together.

I think some of the rise around 35Hz is real, but some of that is an averaged room mode and then the dip at 55Hz is for sure an anti-node, so not real. The dip at ~23Hz and the rise again at 21Hz is real, and I think is related to the Fc of the horn...it almost looks like it has too sharp of a Q, so it's not loading properly at 23Hz or something.

Here's the predicted response from Hornresp:

One thing that's interesting is that the predicted spike at 150Hz isn't as large in the measured response. The measured response is showing an 80Hz 24dB/octave lowpass that is generated by my receiver, so the reduction isn't as large as it might seem at first...but it's still below the 0dB line.

[DrWho]

Let me know if you want any other measurements...or if you want, you could download Room EQ Wizard (it's free) and I can send you the raw files so that you can play around with it in any view you want. I've also got waterfalls available, but they're extremely hard to read when the measurement was taken in a room.

As far as my subjective impressions...

One thing I totally underestimated was how much pressure would actually be generated inside the horn...especially when cranking it up to 130dB at 21Hz [H]

I originally built the horn with no internal bracing and it vibrated crazy style. I opened it up to add some bracing and it really helped a lot, but wasn't enough. Sadly, the way I constructed the horn prevented me from adding more without constricting the horn path so I've never added more. I did open it up one more time though to add some carpet padding along every parallel surface on the inside. This dramatically reduced the ringing we were hearing - I'm still not sure though if it's related to the vibrating walls or if there are standing waves on the inside - it's probably a bit of both.

So doing my best to ignore the vibration happy horn, I spent a couple weeks listening to music with it as my subwoofer. I used a pair of Ascendant Audio Assassin 12" subwoofers that I originally had in 3.5 cubic foot ported enclosures operating over the same bandwidth. On songs that didn't trigger the resonances, the tapped horn sounded exactly like the original ported subs. It was the weirdest thing ever. The only difference is that the tapped horn would go about 10dB louder.

Sadly, I was looking for lower distortion - not playing louder so I was a bit dissappointed.

One attribute to the design, which was part of what I was wanting to test, was that the horn mouth had the same cross-sectional area as the combined Sd of the drivers. I am pretty certain that if the mouth were made larger, that you'd see a wider Q at the Fc and it should get rid of that dip at 23Hz. The problem is that as you make the mouth larger, the rest of the horn gets larger too (since you gotta keep a constant expansion)...so it's not exactly a free lunch.

Also, the drivers I used have stamped steel baskets and you could totally hear them ringing. Sadly, the throat of the horn is too small to flip the drivers around.

So if I had to do something different, I would try to go with a horn that used a single driver. This would make the horn not as wide, which would allow you to make the throat deeper and flip that driver around. Also, the effective horn length seems to line up with the inside dimension of the horn path...so the shortest distance one can trace through the horn - not the path length travelling through the center of the horn. I also had to make a second bend at the bottom of the horn. I would try real hard to get the front of the driver firing straight out the horn mouth (like you see in all the drawings), but that requires a proper driver. I had to make the second bend because the horn wanted to be tapped before the mouth because of the drivers I was using. I had a way of doing it without a bend, but then it would have been too tall to fit in my room. But with that bend, I think I was able to attenuate the ringing of the woofer baskets more...

[DrWho]

Oh...one more thing.

I've got some other things I want to try with the tapped horn, but they're ideas I don't exactly want to share on a public forum. I know Danley was doing something different with his tapped horns to increase their bandwidth from 2 octaves to 3 octaves, but I think I've got another idea that will have no limit to the HF performance (except of course the effects of the bends and the HF corner of the drivers).

If you're willing to experiment a bit, then feel free to drop me an email and maybe we can work on some of these ideas together.

[Ricci]

Thanks for sharing the info Mike. I too have been intrigued by the tapped horn design, but I've never really had the time to play with it yet. If I may ask, what did the external dimensions end up being for the assassin 12 tapped horn? Do you feel like the sound would've been cleaner and less noisy if the enclosure, and drivers were better, or was part of it the tapped horn design itself?

[PrestonTom]

Thanks for the follow up Mike. I think the response actually looks pretty good. Some more questions if I may and these may help others also.

1. Did you choose two drivers in order to smooth out some of the peaks & dips in the response (Hornresp shows these differences also, using other drivers)?

2. In the DTS-20 version (Danley's) he does not actually have the back of the driver firing directly into the mouth. Rather the rear wave fires into the last section of the final flare. IOW, the rear wave gets a bit of loading also, I think - since I have not seen actual plans. However, most of the clones do not do this extra step. What are your thoughts?

3. In the Danley version he actually has two PVC pipes between (I think) the two "chambers". The ends are capped on one side, so presumably they (the two sides have different dimensions, I think) function as 1/4 wave resonators in order to knock down some spectral peaks (I assume). Did you try such a detail, if so can you get much tuning from them? I assuming they are similar to Helmholtz resonators in that you can tune the frequency by adjusting the length or volume and adjust the Q by using some fiberglass fill.

Thanks,

-Tom

[DrWho]

Thanks for sharing the info Mike. I too have been intrigued by the tapped horn design, but I've never really had the time to play with it yet. If I may ask, what did the external dimensions end up being for the assassin 12 tapped horn? Do you feel like the sound would've been cleaner and less noisy if the enclosure, and drivers were better, or was part of it the tapped horn design itself?

External dimensions ended up being "large" []
68.5" x 16.5" x 29"
(h x d x w)

To be honest, I think the tapped horn for sure has its own problems. Some frequencies ring more than others (which is unrelated to the cabinet vibrations) and sometimes it just sounds like a chuffing port - so there's gotta be too much air velocity in there somewhere. That said, I think better drivers would bring it to a whole new level - if nothing else, a better driver would allow one to go with a single driver instead of two, which would make the cabinet narrower, but deeper (overall volume would be close to the same though).

[DrWho]

Thanks for the follow up Mike. I think the response actually looks pretty good. Some more questions if I may and these may help others also.

1. Did you choose two drivers in order to smooth out some of the peaks & dips in the response (Hornresp shows these differences also, using other drivers)?

2. In the DTS-20 version (Danley's) he does not actually have the back of the driver firing directly into the mouth. Rather the rear wave fires into the last section of the final flare. IOW, the rear wave gets a bit of loading also, I think - since I have not seen actual plans. However, most of the clones do not do this extra step. What are your thoughts?

3. In the Danley version he actually has two PVC pipes between (I think) the two "chambers". The ends are capped on one side, so presumably they (the two sides have different dimensions, I think) function as 1/4 wave resonators in order to knock down some spectral peaks (I assume). Did you try such a detail, if so can you get much tuning from them? I assuming they are similar to Helmholtz resonators in that you can tune the frequency by adjusting the length or volume and adjust the Q by using some fiberglass fill.

Thanks,

-Tom

1) Yeah, pretty much. With one driver, the smiley face curve through the passband becomes more dramatic. However, if you find yourself a good driver, then you only need one for things to model well. I think I've got a short list here somewhere that I can post when I find it. For subwoofer applications, the Peavey 18" Low Rider is on my list of next to purchase (since it models awesome in all sorts of horns and it's cheap).

2) You can actually model moving the driver to different locations in the tapped horn - it doesn't need to be just as the very end. At first it seems like moving the driver up the horn will increase your LF extension in the model, but you gotta be careful that you don't inadvertantly make the horn longer. In other words, you need to make two changes when you move the driver around in the model (though I can't remember what they are off-hand).

What I've found is that any movement away from the throat causes the response to become more peaky. I think the same happens when you move away from the mouth too, but it's been a while. There is also a tradeoff between bandwidth and efficiency as you play around with this parameter too. My design ended up with the drivers right at the throat, but away from the mouth a bit. You really just gotta run the numbers through the model to determine where you want the compromises. I think one thing I noticed is that you can kinda cheat the 2 octaves and start pushing 2.5 as you move the driver further from the mouth - I think it had the effect of lowering the Q of the peak at the bottom end of the response. Making the mouth larger has that effect too.

3) I haven't done any recent research on Danley's new designs, but I know he is very adament about being able to achieve a 3-octave bandwidth instead of being limited to only 2-octaves. He was very secretive about how he obtains this, but said he would reveal all once he knew when a big speaker brand purchased one of this speakers for reverse engineering. There is also a lot of talk about how the idea of a tapped horn is already patented, so nobody knows how he can claim a pending patent for the design...

If you look at my simulation, you'll notice that there's a huge dip right around 100Hz and the LF corner is around 25Hz (2 octaves). You'll probably notice as you bang away at the model that you'll never be able to get rid of that huge dip at 100Hz (or wherever 2 octaves above your LF corner ends up). I've been led to believe that the dip is the result of straight up destructive interference between the front wave and the rear wave (since they have opposite polarity).

After seeing the pipes in a pic I saw posted of the DTS-20's back hatch open, I began to wonder how they might have been used to prevent the destructive interference, since that's really the only way the bandwidth could be extended beyond 2 octaves. Danley talks a lot about transmission-line theory and how taps are used in electrical circuits. I've not cranked through any numbers (something I intend to do when I have the time), but there are ways to tap circuits with reactive loads to improve performance at certain frequency ranges.

I've also got another idea about how to improve the tapping of a horn, but I think it might be a patentable idea so no way will I share it in public. At first I thought that was what Danley was doing, but after seeing all his other tapped designs and hearing him talk about it I think he's doing something else. Hopefully I'll be landing myself a job here in the near future and will have the resources to test my theory. The math is a bit beyond me right now so I'm totally interested in discussing some ideas in private with those that enjoy sadistic number crunching more than I do - otherwise I'll just be busting out some rule of thumb calculations and doing the good old trial and error approach.

I hope this helps. I'm by no means an expert on this whole tapped horn thing so I hope I'm not being too misleading [] I would really really like to hear the DTS-20, or really any other tapped horns out there. I actually built my tapped horn with the intent of throwing it away when I was finished (but saving the drivers). I really only did it as a baseline to see how a mediocre implementation might perform and then use that as a baseline to compare against future projects.