This design began as a project to create what I would consider my personal "reference" speakers. Living in a small apartment, my larger speakers take up a good portion of my room, while the extra bass really isn't needed. So, the idea was to create a very high-quality, fairly robust speaker to fulfill my everyday listening needs.

The quality construction of the Ex-Monsters accented by a lovely green 70's sofa

I ended up choosing the tweeter that I wanted to use first. I was looking for a tweeter that would be unique, and yet offer good performance for the money. I began to investigate the Bohlender Graebener Neo series of drivers, starting with the Neo3. I tried out the Neo3 in a standard dipole configuration, just to get an idea of what they sounded like. I was very impressed by their ability to play clean and effortlessly, but I was not impressed with their off-axis performance. So, I looked at the newer Neo3PDR, and found that it maintained the clarity and detail, but also offered improved off-axis performance. Additional testing found that the harmonic distortion figures were as good or better than some of the more expensive tweeters on the market.

During my testing, I evaluated both the Neo3PDR and the Neo3PDR-FP, the faceplate version. The faceplate version offers an overall flatter response from 2K on up, but the edges of the faceplate and the waveguide also induce some diffraction effects that cause a bit of a roller coaster response. Whether these dips of several dB are really noticeable is definitely arguable. If you are looking to use a low crossover point, the faceplate version will help fill in the frequencies from 2K to 4K, thanks to the slightly horn-loading wave guide. However, most importantly, in my situation I needed to maintain as small of a faceplate as possible, due to the "time aligned" design.

One thing to note is that I ended up sealing the rear of my Neo3PDR with a piece of vinyl damping material attached directly to the back of the element. I somewhat forget the reasoning behind this, but I believe that using it in a large rear chamber actually reduced the midrange output even further, making it hard to get a smooth transition at the crossover.

#264-714 BG Neo3PDR

The woofer that I chose was an older Seas Excel W17EX001 6.5" driver with a black fiberglass cone. This driver offers the low distortion motor of the Excel line, but also offers a more user-friendly and slightly warmer midrange. I feared that the newer magnesium cone woofers would not sound good with the Neo3PDR. To top this off, I had a chance to purchase some of these drivers used at a good price, so I couldn't pass it up. As I said before, the goal was to produce a reference speaker for my own use, so I wasn't concerned about the unavailability of the woofer.

The cabinet was designed to offer a time alignment of the drivers. From measuring the woofer and tweeter to compare acoustic centers, I decided an offset of roughly 1-3/4" was necessary. This offset would begin right above the woofer and go down at a 45 degree angle to the location of the tweeter. Here again, the small footprint of the tweeter comes into play.

The cabinet was constructed out of 3/4" MDF; two braces running vertically up the sides and across the top were employed. The cabinet walls and bracing were attached with biscuits and yellow wood glue. The result of the double brace is very impressive, as the cabinet is quite solid. The 2-1/2" thick front baffle doesn't hurt the rigidity of the cabinet either. The external dimensions of the finished speaker were 9" wide, 13" tall, and 14" deep across the bumped-out portion of the speaker. The difference between the tweeter and woofer positions is again 1-3/4". The net volume by the time the bracing and port are subtracted is about .55 cu. ft. The port was 2" diameter PVC and I believe was 7.5" long.

Here is a picture of the inside of the cabinet, with the front of the speaker removed. One picture is with the port, one is without. Notice also that the damping material is in place, which consists of vinyl damping sheets and Black Hole.

The first step was to make the different layers of the baffle lamination have larger cutouts on the back side. It took some planning ahead, but using a larger cutout in the rear piece of the baffle helps quite a bit. I went further to remove material from the inside of the front baffle using my router and a cove bit. Notice that I attempted to keep material around where the mounting screws would be located. See below picture.

A scalloped effect was done on the inside of the baffle to "open up" the space behind the woofer.

I added a rear chamber behind the Neo3PDR to eliminate the need for a tight seal directly around the driver. Since the raw element has a rather wierd shape, it would be difficult to prevent woofer air from leaking out around the element. Hence, a 4" PVC end cap was glued onto the back of the baffle to seal things up. If the faceplate version of the Neo3 was used, this rear chamber would be unnecessary, as the faceplate versions include a foam gasket.

The inside of the front baffle features a small chamber to protect the tweeter and eliminate air leakage.

The cabinet damping consisted of lining the walls with Black Hole 5 where possible. The areas that could not fit the Black Hole received a coating with sheets of Parts Express vinyl damping material. In this situation since the cabinet is so rigid, it probably provided minimal benefit, but the small cost probably made it worth the effort. Also note that because of the complex bracing, all of the damping material was inserted before the front baffle was glued into place. This is a somewhat risky procedure, but since I had listened for these so long in an unfinished state, I was pretty certain about the results.

The speaker was finished with Red Oak NBL veneer. This was a "rift" cut, and not the standard face cut that you find on Oak veneer. This cut of veneer basically produces a continuous quarter-sawn look with a very small and tightly repeating grain structure. The two speakers are covered with mirror image cuts of veneer on all sides. It was secured with standard contact cement. In the beginning, I trimmed all of the veneer by hand with a razor knife, but as my time to complete the project ran out, I ended up flush-trimming the veneer with a flush-trimming router bit. This worked remarkably well, and will definitely be my method of choice in the future. However, with thinner or more fragile veneers, this method may not work as well.

If you notice in one of the pictures below, I had already made the driver recess before I applied the veneer. This creates a very difficult situation in trimming the veneer around the cutouts. I knew this would be a problem when I first built the boxes, but figured that I would figure something out later. Well, after doing some brainstorming and playing around, I did come up with a solution to the problem.

I ended up making my own very small flush-trimming bit. I believe I have seen the idea before, though I am not sure where, so don't ask for a source. Again, because of my tight time frame, I had to get whatever bit I needed locally. I ended up buying a very small radius high speed steel roundover bit. This is one of the cheap ones that does not have a bearing and simply rides on a metal shaft. Because the "bearing" section of the bit was too long, I went in with my Dremel tool and a cutting disc and chopped down the length of the bearing. Next, I used a sharpening stone on the Dremel tool to turn the bottom end of the roundover into a straight cutting bit. The result was a very small yet very effective flush-trimming bit perfect for cleaning up those little driver recesses.

The finish is an off-the-shelf Zinnser shellac, the color was achieved by mixing their clear and amber varieties. I mainly chose to use shellac because of its fast drying times and less-offensive odor. I had roughly 4 days to do all of the finishing, and it had to be accomplished in my apartment in the evenings. The shellac really paid off, because I was able to do one or two coats in an evening without totally asphyxiating myself. The resultant finish is very deep and rich, and has a very good physical feel to it. I would recommend to anyone to give this product a try!

Since I did use the raw element version of the Neo3PDR, I needed to make a small faceplate to neaten up the appearance. What I ended up using was a piece of copper sheet stock that you can get at the hobby store. I cut the copper very painstakingly with my sabre saw with a fine metal cutting blade on it. I went through about 8 blades in the process, but it eventually worked out. I cleaned up all of the cuts with a hand file and sanded the copper clean. A small ring of the copper was masked off and the rest of the plate was sprayed with a flat black paint. Once the mask was removed, the result was a black faceplate with a hint of copper to accent the copper phase plug on the woofer. The entire copper faceplate was then sealed with a clear lacquer to prevent oxidation of the piece.

The crossover design for this speaker was done with a combination of software simulations and good old fashioned tweaking. For the most part, I use software to get things in the ballpark, but usually have to go through several iterations of measuring and modelling to get phase issues worked out. One of the main goals of this design was to achieve a speaker with good phase coherency on axis with the tweeter-- not on an axis angled up or down from the listening position. Here is the crossover schematic:

As you can see, the crossover is a standard parallel design. The woofer crossover features a third order design plus conjugate network and baffle step compensation circuit. I know that the use of the extra resistor as part of the baffle step is technically not correct, but I find that it does help to increase the effectiveness of the BSC circuit. The difference with and without is very slight, yielding roughly 1 dB of attenuation from 1K to the crossover point. I listened to the speaker with and without this resistor, and there was a noticeable but small difference. Using the resistor did not degrade the sound, but provided what I would consider to be a slightly fuller midbass region, and slightly more laid back midrange. Its really a matter of personal preference.

The tweeter section of the crossover uses a standard second order network, plus a notch filter and padding resistor. The notch filter is designed to remove a natural hump that occurs in the Neo3PDR raw element at roughly 10K. While the notch filter does reduce the overall sensitivity very slightly, I had extra sensitivity to spare on the tweeter so that was not a concern. A single series resistor of 6 ohms was used to do the remaining matching between the woofer and tweeter.

The above picture shows my normal crossover test configuration. The crossover was initially designed and auditioned using 18 gauge inductors, Dayton caps, and Dayton Non-inductive resistors.

In the final state of being, the speakers ended up with a top-quality crossover featuring Theta film and foil capacitors, Mills resistors, and Alpha-Core foil inductors. It's a long story...

Here is the overall frequency response of the system and the tweeter-reversed measurement. This was performed directly on the tweeter axis, showing that my goal was achieved and phase coherency at the crossover is best on the listening axis. Also, it can be seen that the final crossover point was just under 3,000 Hz. The overall response may be considered a tad bright if anything, but with this tweeter, it really brings out a sense of air.

The overall system impedance and impedance phase. The impedance hits a minimum of about 4.5 ohms, with a max of about 14 ohms, so I guess you could say that compared to some designs, it isn't too bad. Again, the impedance phase basically just shows you the magnitude of how fast the impedance is changing, which also doesn't look too bad.

The step response of the speaker. Definitely not "transient perfect" but better than many speakers out there.

Here is an on and off-axis measurement made on a different system before the speakers were totally finished. While the measurement is not all that accurate, we can still see that the system holds together quite well at 30° off axis, and more importantly the Neo3PDR is good at 30° well above 10kHz.

I really learned a couple of major lessons as a result of this project. The first is that the Neo3PDR is a very good tweeter and is a serious competitor with any tweeter under $100; heck even with some of the more expensive tweeters. The off-axis response both vertically and horizontally is not all that bad for a planar device, the relatively short length definitely helps. The Neo3PDR seems to be able to capture the potential of the planar tweeter-- a very "electrostatic" type of sound that is clear, airy, and effortless.

The second main thing that I learned was how the quality of crossover components effects the sound of a speaker. I have done component tests before, but have never really experienced dramatic differences. However, because of the detail and clarity of the Neo3PDR, I was able to clearly hear a difference between some of the sandcast resistors and the Mills resistors. I think with many tweeters, especially when being played on modest stereo equipment, the difference probably would not be so apparent. But, I had the opportunity to hear the speakers on a VERY nice system, and switching resistors from the sandcast to the Mills produced an audible difference. The sandcast resistors definitely imparted what could be considered a "grainy" or "harsh" sound to the speaker. They were very fatiguing and in fact I added tweeter padding to help eliminate this harshness, but ended going back down in resistance once I went with the Mills resistor. I wouldn't mention these results if I didn't feel that this experiment was very repeatable and that a clear difference could be determined by just about anyone.

Overall, I was impressed with the results of this project. These are definitely not a beginners speaker, but the results can meet the demand of even relatively serious audiophiles. I listened to these speakers compared to a pair of Acoustat electrostatic speakers, and they sounded remarkably similar in detail and tonal quality. At DIY2002 Dayton, they received mixed reviews during the main audition day, but on the better system later in the evening, they sounded quite good. They received favorable comments, and in my opinion were one of the better designs at the event. Of course, I may be somewhat biased...

This design was submitted by Darren Kuzma, he can be tracked down on the Tech Talk board with any questions.

Last but not least, a picture to explain why these are called the Ex-monsters. For a halloween speaker-building get-together, they dressed up like a pair of furry monsters.