Zaph|Audio
Horn loaded Seas 27TDFC
What happens when you horn load a dome tweeter?
Jan 22, 2006 - Added part 2
Introduction
I had seen some normal tweeters mounted in a configuration like this before but nobody went into detail about the results. I had to find out for myself. We start with initial flat baffle tests. These are on an infinite baffle:
Frequency Response - Flat Baffle - 0, 15 and 30 degrees
The ABS horn assembly came from MCM, and is part number 54-580. It's milled down enough to remove the threaded compression driver mount and make the hole 1.5". When laid on a flat surface, that gives the horn slightly less than 1.25" of depth. You'll know you milled down far enough when you get to the 2nd rib from the outside. I've you've machined into that, the hole will be too big. The 1.5" hole is just enough to fit the Seas dome and the small curved part of it's flange right next to the dome surround.
Back view of modified horn
Side view of modified horn
The tweeter is then clamped into place from the back as shown:
Rear View Horn Mount
Then we take the same set of measurements:
Frequency Response - Horn Conversion - 0, 15 and 30 degrees
And finally, to compare the minor differences in the impedance curves:
Zoomed in impedance curve comparison
There is only a minor difference in the impedance curves, with the horn loaded version about 20 hz lower and about 1 ohm higher. Not much to speak of really. The real difference is in the frequency response curve. Suddenly, we have a broad 6db rise in response between 1.5khz and 6kHz. Your first thought is probably what the hell am I going to do with that?
Then the idea popped into my head: What happens to the numbers if I shape the response back to normal? So just for the hell of it, I added a single 3.3uF capacitor and ran the numbers again.
Horn conversion with a 3.3 uF series cap
Hey, now maybe we're on to something. I've got almost a perfectly behaved 2nd order target slope at 1.9kHz, and I did it with only one component. Now let's check the harmonic distortion:
Harmonic Distortion with a 3.3 uF series cap
Thanks to the added directivity, the tweeter is effectively working more efficiently in the range up to 6kHz, and after shaping the response, distortion went way down in that area. The net frequency response and SPL are similar to the flat baffle after shaping the response but there are improvements elsewhere. I think we have a tweeter design that could work well for 2nd order crossovers. The bonus is that the horn moves the effective acoustic center back, placing the listening axis straight ahead without the need for assymmetrical slopes or delay networks.
So, There does appear to be some benefits for this particular arrangement:
Before you go around claiming that I've been converted to liking horn systems, be aware that this is not really a horn, it's more of a waveguide. Most of the positive elements of dome tweeter performance remain, without most of the negatives of compression drivers on traditional horns. I believe that a horn conversion of this type will not work as well with a long throat horn design, as that pretty much requires a compression driver, and lots of other problems are introduced that are typical of horn systems.
Your results may vary, but feel free to expand on the ideas presented here. I may give this conversion a shot with a 2nd order system someday. (edit: I did, see the Waveguide TMM system on this web site)
Part 2
In this section, I've tried a similar modification to the 8" and 10" Parts Express waveguides.
8" Parts Express waveguide, modified to fit a Seas 27TBFCG
The above response curve is looking a little ragged, with large variances in the off axis response. 0, 15 and 30 degrees are shown.
10" Parts Express waveguide, modified to fit a Seas 27TBFCG
10" Parts Express waveguide, modified to fit a Vifa D26NC55
Likewise, the curves taken in the 10" waveguide don't look that great. While the first curve uses the Seas 27TBFCG, I was curious if some of the issues may have been caused by the grill or phase shield. The second test was done with a Vifa D26NC55 to be sure. Similar issues result. While off axis variances in the top octave are not very important, they are much more audible as they occur further below 10kHz.
The popular question is "will these allow a lower crossover point compared to a flat baffle?" The answer is yes. Looking at the boost, we get roughly 8-10 dB more efficiency in the 1-2kHz range for both the 8" and 10" waveguides. Despite the larger size of the 10" waveguide, the boost is not substantially higher and it only moves the center down about 200 Hz. Down around 600hz, the boost has diminished to about 5dB. In my own estimate, these waveguides lower the usable crossover point to about 1kHz with a LR4 slope. The slope can be achieved but will likely require a larger number of shaping components.
Looking at these results, it must be questioned whether the ragged and uncontrolled top two octaves are worth the benefits of the lower crossover point. For passive crossover systems, I say no. I could potentially see a benefit using one of these with a DEQX however, but those are pricey and uncommon, and they don't necessarily solve response problems on multiple axes.
I believe the Parts Express waveguides are a little too deep and narrow for the usage described here. They begin to show some of the response and energy storage problems common with horn systems. In summary, The only gain seen here with the PE waveguide is with a slightly lower crossover point, while we loose some of the positive gains of the smaller MCM version.
Page done by John "Zaph" Krutke © 2005
Also visit -Zaph|Audio-