I had been quite interested in several different midbass drivers including models from Vifa, Seas and Peerless. Of particular interest were the Vifa P17, the Seas P17REX and the Peerless drivers using their sandwich cones. Neither the Vifa nor the Seas modeled particularly well on the lowend (I like my bass) in my opinion. On top of that, the Seas was a bit pricier than I wanted to go. That left the Peerless models. The new HDS line seems to be quite popular, and appears to have quite good construction. From what I've gathered, the CSX line is the same as the HDS, but without the cast frame. Rumor has it that the CSX frames suffer from a resonance around 400Hz, but that applying some doping to the frame can reduce or eliminate that problem. So I opted for the cheaper 7" CSX driver, anticipating good results. From literature and comments on various forums, I anticipated neutral and detailed mids, and "dry," solid bass, with very low distortion.
Next came the tweeter considerations. The number of tweeters available in my price range seemed much better than the midwoofers, so I had a little more weeding out to do here. This time the most appealing manufacturers were Vifa, Seas and Morel. After receiving several very positive reviews on the Seas 27TFFC, I decided it would be a good mate to the Peerless midbass.
In an ideal world, I would have taken detailed measurements of each driver to get accurate T/S parameters, and taken accoustic measurements to use in a crossover design program. This is not, however, an ideal world! I felt quite fortunate that there was another speaker builder in town that had some good measurement software and equipment. Unfortunately, while attempting accoustic measurements, we succeeded in blowing up part of his hardware (sorry Kai!). Rather than risking more damage to his equipment, I decided to give up on that front.
One of the thoughts I had was to use the published frequency response and impedence data to design my own crossover. There are tools available to trace in FR and impedence data and export them as .frd and .zma files. These can then be imported to other programs for crossover simulation. I used Speaker Workshop to simulate a crossover using the published data, and it looked pretty good. The whole time, however, I had a little concern floating around in the back of my mind that the published curves weren't very accurate. I also didn't have a good way to compensate for my baffle size. This is when I realized that my best option would be to have Madisound do a Leap design for me. I wasn't excited about shelling out the money for the design, but I also didn't want to have $200 worth of drivers attached to a crossover that made them sound horrible. I figured the Leap design would atleast be close to what I want, and certainly closer than I would get using textbook formulas.
The untweaked crossover is shown below. The only value I changed was the inductor in the highpass section. I wasn't able to get the .28mH inductor from Parts Express that was specified in the original design. I had to go with a .27mH inductor instead, but that difference is only a few percent, which shouldn't make too large of a difference. The lowpass section is comprised of a second order electrical LP filter formed by L2 and C3. Rbs and Lbs are used to counteract the baffle step rise. This filter drops the overall sensitivity of the system by about 3-4 dB, but allows the overall response to be much flatter when the speakers are placed away from the rear wall. It is still recommended to keep the speakers within a foot or two of the wall to maintain reasonable bass reinforcement. The tweeter section is a third order electrical HP filter formed by C1, C2 and L1. Rp1 and Rp2 form a fixed "L-pad" attenuation circuit to match the output of the tweeter to the woofer. The target responses were set at 4th order accoustic at about 2300Hz. Below the crossover are the response plots predicted by Leap showing the overall response, the individual driver responses, and the impedence profile. For the most part, these are true 8ohm speakers. The impedence drops in the upper octaves where the tweeter's 6ohm impedence dominates. The minimum is only around 5 ohms, which shouldn't present a problem to many amps.
Crossover Schematic (click for a larger version)
I wanted to try a unique enclosure shape, and the original design I drew up was considerably more complex than what I ended up using. Among the oddities of the original design were more bevels and rounded edges, along with side porting. This design would have been fairly doable, but I later decided I wanted to veneer the cabinets, so I scrapped the idea. As you can see, the only cosmetic touch is the angled/beveled corners on the front, which was fairly simple to cut on the table saw, using a simple jig. I'm sure the accoustic effect of the bevels is minimal, so it could probably be eliminated or replaced by a large radius for those who would prefer a different look. The bevels were a little more difficult to veneer than a rounded edge may have been, but I like the look. In order to accomodate the bevel, I used a double thick baffle (1-1/2"). This also served the purpose of providing a solid mounting point for the drivers. Because of the very thick baffle, I put a chamfer on the back side of the woofer opening to allow better airflow out from behind the woofer cone. I didn't want the woofer being restricted by the "tunnel" created by the thick baffle. The rest of the enclosure is just 3/4" MDF. All of the joints are just glued butt joints, which are really as strong as anyone would ever need. I find that I'm able to positon the panels better by just using glue and clamps. Additionally, I don't have to worry about the "bulge" that can result from not drilling a large enough pilot hole for the screw. Internally there is a brace that ties the front, back and sides together, and is located between the woofer and tweeter. Because the enclosure is so deep, this still left the lower portion of the enclosure a little unbraced for my liking, so I added a few small braces between the sides to tie them together. The lower portion of the enclosure is also lined with 1/2" carpet pad, which does a nice job of damping internal waves, preventing them from reflecting back through the cone.
The enclosure was designed using the published specs. The final internal volume is just over 19L (which is around .67ft^3). The external dimensions are 16"Hx8.5"Wx14.25"D. This gives internal dimensions of 14.5"Hx7"Wx12"D (remember that the front baffle is 1.5" thick). The enclosure is tuned to 34Hz, which gives an f3 cutoff around 40Hz, which is quite good for a 7" driver. I used a 5" long, 1.5" diameter port that has a large flare on the outside edge. I don't notice any port noise, however I avoid turning up the volume to excessive levels. If you plan to play them really loud, you may want to move up to a 2" diameter port, 9.5" long. I wouldn't recommend going any larger in diameter, as the length becomes quite long (a 3" port would be on the order of 22" long!) The drawing below should provide all the necessary information to build the enclosures.
In building these speakers, I really wanted them to turn out "right". I did a number of extra things to make sure they were as good as they could be, without being excessive. First, I used high quality parts. The 1.2 mH inductor is a 14ga aircoil, which offers very low DC resistance. Using an aircoil for the 2.0mH inductor would have been extremely expensive, so I used a high quality Erse iron core coil. The DC resistance is again extremely low, and The Erse coils are claimed to introduce very little distortion or saturation, compared to other iron core inductors. All of the capacitors are Dayton polypropylene. Parts Express was out of 15uF capacitors at the time, so rather than step up to the more expensive Solens, I opted to just use teo 7.5uF caps in parallel to make up the 15uF caps needed in both the high and lowpass sections. The same was true with the 1.5 ohm resistor in the baffle step circuit. I used two 3 ohm resistors in parallel there. Paralleling these components helps several things, in theory. First, any inductance in either the caps or resistors will effectively be cut in half, which is a good thing. Also, the power handling is doubled, over a single component. This isn't particularly important for the caps, but for the resistors, which could potentially see large amounts of power, this is a great benefit. I also wired the speakers internally with 12ga OFC cable. I figured it couldn't hurt, and I certainly didn't want that to be the weakes link.
Another nice feature is the bi-amp terminal. While I don't plan to bi-amp the speakers any time soon, it does allow the possibility for the future, as well as the option to bi-wire, which may be more realistic in my case.
The final touch that was the veneer. I was able to get a good deal on some bird's eye maple from Tape Ease for a very reasonable price. The quality of the veneer was exceptional, and really adds a distinctive touch to the speakers. I still haven't decided on a color for the speakers. I am toying around with the idea of a translucent smoke gray stain, which I think would be outrageously good looking. We'll have to see what unfolds down the road.
After the Competition
Now that DIY2001 has come and gone, here is some additional information:
I was a little disappointed with how the speakers measured at the competition. They have a broad dip across most of the midrange, and the high end rises like crazy. This is something I might have expected from textbook crossover values. I think these speakers NEED to be placed fairly close to a wall, also. When auditioned in the large room and placed several feet from the rear wall, the baffle step showed it's ugly head and the bass output seemed almost non-existant. I've experimented in my listening room, and I feel that good positioning would be about 18" from the wall to give some bass reinforcement, while still maintaining good imaging.
On the positive side, I feel that the speakers image very well, and despite the midrange dip, they have good detail across the board. Now to get the crossover fixed up to balance things out...
I'll update the crossover in the near future, so check back sometime. Here's a link to my DIY2001 web site. Kai Stark also has the official Midwest DIY2001 site, which includes the judges comments.