Zaph|Audio - Design Contest Results

Zaph Audio Design Contest Results

May 27, 2010 - Initial posting

Introduction
With every batch of ZA14W08 woofers that go from production to Madisound's warehouse, I pull a bunch for quality control testing. I always have quite a few of these extra test subjects since I don't put them back in stock. I also had a couple of test pairs of the excellent SB Acoustics SB29RDCN-C000-4 small flange neo tweeters on the shelf. With the idea of giving them away, the design contest was born.

The designs were for a vented 2-way in a ZA5 enclosure and a 3-way with the woofers selected and supplied by the designer. The contest was open to all designers worldwide, both newbies and veterans alike. Measurement files were provided for the ZA14W08 and SB29 for the designers to work with, in both FRD format and Soundeasy format.

There were roughly 10 entries for the 3-way and 50 entries for the 2-way.

Due to the number of 2-way entries, some were evaluated without being built and listened to. All were modeled however and evaluated within Soundeasy first. A few were thrown out fairly quickly due to unmanageable impedance impedance curves, rough response curves that didn't match modeled response curves sent in, crossover topologies loaded with unneeded components, and so on. But many looked great and required a lot more investigation, mostly into crossover frequency choices and the balance between tweeter power handling and suppresion of the woofer's breakup induced harmonic distorion. I built and listened to what I considered the best of the entries, and ran some additional tests. The 2-way systems were heard by me in a 17x20 ft room on 24" stands, the back of the cabinet about 18" off the wall and 4 feet from the closest side wall. While that room was the specified listening area for the design, a few were designed with a leaner tonality and some also were heard in a smaller room to investigate their performance there.

The 3-way design was not built or listened to by me because obviously I disn't have the selected woofer or enclosure on hand. Instead it was an "on paper" design exercise and evaluation. As such the 3-way entries contained a lot more design information and write-ups about design choices. I gave the winner the right to tweak the design to perfection after they built it. Of the 10 entries I received, there were no losers. All appeared to be interesting, unique and well thought out. I welcome the designers to share these designs because I certainly hated to pick only one winner from this group.

I was chatting with Madisound about the difficulty of selecting a 2-way winner when I narrowed it down to the last few designs. With 3 extra pairs of tweeters and a couple extra woofers supplied by Madisound, I was able to select and reward a few runner up designs for the 2-way contest. Thanks to Madisound for the extra speakers to hand out. It felt good to reward great design work.

2-way contest winners

Overall 1st place winner:
Francisco Javier Huerta - LR4@1.6kHz [Design Image]

What we have here is a deceptively simple design. In fact, at first look, you can't help but yawn at what looks like yet another basic LR4 design. But don't be fooled. In my eyes, everything came together perfectly and the fact that it's a simple LR4 design is just a bonus. If I could sum up this design in one word, it would be balance. It has the perfect balance between tweeter power handling and suppresion of the woofer's breakup induced harmonic distorion, all while having perfectly smooth response and a very tall vertical listening window.

[Image] - Harmonic Distortion @ 90 dB

[Image] - Harmonic Distortion @ 100 dB

[Image] - Harmonic Distortion @ 110 dB

I need to briefly explain what's going on in these HD sweeps. They are not quite the standard methodology used in my driver tests, though the 90 dB plots are close. To get the system performance, they are taken semi-near field but closer to the woofer axis to properly show breakup induced harmonic peaks. As such, the fundamental droops a bit right above the Fc because of the relative distance to the tweeter, but this is the most accurate system HD sweep that I can get in a non-anechoic environment. Now, the Behringer ECM8000 is very clean at 90 dB. However, it's own inherent distortion starts to become an issue at higher levels. So, for each increase of 10dB, I have to back the mic up a distance to keep the level at the mic down. So, as I back the mic up for the higher levels, we can see the increase in level related distortion. But along with that, we get more room interaction. That is why the increased level HD sweeps get "fuzzier". It's still clear enough to see distortion trends however. Keep all this in mind for the high level HD sweeps above and below.

With Javier's design, tweeter distortion does not increase much at higher levels, while some of the 5th order and most of the 3rd order harmonics are filtered out. Nice. I couldn't really ask for more. Actual system response measurements matched up exactly with the modeled curves.

Low crossover LR6:
Alexandre Chamagne - LR6@1.1kHz [Design Image]

Alexandre's design is the one that pushed the tweeter the hardest, even more so than the LR2 system below. The design uses a brute force method to get several things done, and as such there are a lot of components, some that are large values. Not exactly a cheap option to build. Don't be fooled by the greatly amplified Y axis frequency response curve, the response was buttery smooth. And with the low crossover point, off axis did not change at all anywhere in front of the speaker. The system was a pure point source, and power response was a flat line aside from the tweeter's rolloff. At lower levels, this system without question sounded the best. Alexandre's crossover does something unique, it tames the upper peak of the box tuning. Without that peak there, the large inductors in series on the woofer give less rise just above the 2nd box tuning peak. To know what I'm talking about here, observe the transfer function of a woofer with a large series inductor. On the bottom end, there will always be some peaking just before the woofer starts it's rolloff. With this design solving that issue, it equates to a leaner tonality, one that still sounded good in the test room, but also sounded good in my much smaller home office.

[Image] - Harmonic Distortion @ 90 dB

[Image] - Harmonic Distortion @ 100 dB

[Image] - Harmonic Distortion @ 110 dB

At higher levels the tweeter did begin to show a little strain, with 2nd order creeping up, and higher level harmonics beginning to show up between 1 and 2kHz. Still, it was good for most listening and still got seriously loud. It also showed what a tank the SB29 really is. I'm impressed. This design was very close to being the winner but a slightly more conservative LR4 provided higher output. Still, if a person's listening level requirements aren't outrageous, and/or the system is used in a smaller room, Alexandre's design is the winner.

Second order:
Jay Kim - LR2@2kHz [Design Image]

This design is a valiant effort at a second order crossover, something not often attempted with metal cone woofers. Jay was able to get close enough without using a complex delay network, and power handling of the SB29 was not a problem at all. I often hear designers say that a breakup node has to be X dB down to work well. 30, 40, 50dB or whatever. This is not really correct. A response curve only has to adhere close enough to the target response to get the job done, and from there the breakup induced harmonics are the only real issue to contend with. Jay obviously understood this in his design.

[Image] - Harmonic Distortion @ 90 dB (oops, mic off woofer axis)

[Image] - Harmonic Distortion @ 100 dB

[Image] - Harmonic Distortion @ 110 dB

The only reason this design worked well was because the ZA14W08's breakup node was high enough, and the breakup induced harmonic peaks for the 3rd order and 5th order are not too bad to begin with. The design sounded great and didn't have nearly as much issue as some larger metal woofers with higher HD and lower breakup nodes. (Try a Seas L22 in a 2kHz Fc 2-way for example) So, there is some HD peaking in this design but it's barely audible. The only reason this design did not take first place is because I had to be really picky, and with drivers this small and their short center to center distance, the smoother power response of LR2 was a bit less of a benefit and it was still closer to a point source. Make no mistake, this design sounded incredible though. It's one that really tested my limits of what I can hear in non-linear distortion. This design got a lot of time in this test group. Quick comment, as you can see, I accidentally took the 90 dB HD plot a bit off axis. It's obvious to me, but if you weren't aware, harmonic distortion does indeed roll off in much the same way frequency response does off axis. With speakers where non-linear distortion is an issue, listening off axis may help.

Conservative LR4:
Jay R. Taylor - LR4@1.8kHz [Design Image]

This is probably the design to use if you use a subwoofer and listen at obscenely high levels. 1.8kHz with a tweeter built like a tank is about as safe as it gets. Conservative LR4 with this driver set was particularly hard to choose for a winner. This driver combo has a natural tendency design itself into this sort of LR4 system. It was difficult to pick out JRT's design because there were 20 other designs almost, but not quite just like this one. Still, after a lot of thought and a lot of listening, I picked this excellent implementation out of all those designs. Building this design you could hardly go wrong. While this design was effectively 4th place, it was still a winner in my book.

3-way contest winner

Congratulations Dan Neubecker. I've handed the 3-way design win to him. Very nicely done and well thought out design, and very unique on top of that. Dan's design uses the SB Acoustics 8" woofer in a transmission line configuration created with assistance from Paul Kittinger. The midrange is mounted near the top edge of the enclosure with the tweeter below. Good design decisions appear to be made, and I am impressed by the level of detail. Dan has a post here with details on the design.

Zaph|Audio - Design Contest Results

Zaph Audio Design Contest Results

Buy Zaph|Audio Drivers, Crossovers, and Kits at Madisound