Woofer

The Seas L22RN4X/P 4Ω (H1208-04) woofers were a prize I had won from another DIY speaker competition several years ago, donated by Madisound. (Note: the pic below is the normal 8Ω variant)

Due to the aluminum cone breakup (see the crossover design section for raw response), I always imagined using them in some sort of a 3-way, but an opportunity hadn’t risen. I’ve heard the 10” L26 variant as well as W26 Excel series and was, unfortunately, unimpressed. As a result of these factors, they were stuffed in a box, went through a move, and became forgotten. I rediscovered them during the search through my inventory. They met the criteria, modeled well in a reasonably small box, and they were free. Fine. Let’s give them a shot. I had some 10” Scan Speaks laying around as well, but I wanted to try to make an 8” work first (generally better control of power response).

These 4Ω variants appear to be some sort of a special, limited sample run from Seas, which may be why Madisound donated them to the DIY competition as a prize. When I pulled them out, I thought, “At least they’re 4Ω, so that’ll be a nice boost in sensitivity to work better in a two way.” I would later be disappointed to find out that there is actually no sensitivity improvement compared to the 8Ω (OK, they're maybe 0.5 dB better). They run about 83 dB @ 2.83V. Yeah, that sucks. That was really frustrating news. It turns out that the 4Ω variant has a bit more Mms than the 8Ω (48 g vs 40 g), which seems to have counteracted the improved current flow. 

The significant inductance of the L22 did not inspire confidence in the L22 motor design or in a result with low distortion, but alas, they were free 😊, and every decision is a balancing act of pros and cons. The 8Ω variant is spec’d at 3.76 mH on the data sheet; the -04 preliminary data sheet I found didn’t have inductance specified. Based on matching curve in the Response Modeler tool, it looks to be about 2.90 mH for the 4Ω variant. 

Tweeter

The North D28-06S tweeters (from now defunct North Creek Music Systems) were purchased used from a friend that was cleaning out his closet for a big move. I had forgotten about these in my closet as well and was pretty excited to find them! I really like these tweeters. They have an excellent motor design, with a unique damped rear chamber, sporting a 1mm Xmax. These are some personal favorites and perform way above their price point (used to sell for $156 – for a matched pair).

During experimentation phase, I had tried to shoehorn several other tweeters with the L22, including Morel MDT30, Peerless DX25BG, Scan Speak 9500, SB Acoustics SB26STCN, Dayton RS28, and more. The preliminary modeling that I had done with these other tweeters made for a difficult pairing with the L22. The L22 was the problem - the response falls off very early. Elegance was not on the horizon. Things finally came together when I found the D28s in the back of the closet. They had ever so slightly more low end in their response to work with the L22. 

The D28 also happens to have a silver faceplate, which resulted in a serendipitous match to the look of the L22. 

Woofer Network

The LP target shown below is 1400 Hz Bessel 4nd order. The response clearly doesn't really fit that. The beginning of the knee most closely resembles a 4th order LR until the strange break at ~2 kHz. So, it's a bit of a mess. I really wanted to keep a Bessel target for the tweeter, to give it a softer kneee and easier time in the low end extension. So, I just tweaked the woofer network in the result until the driver phases lined up pretty well with a nice notch. And this is where things ended up. 

The LP filter presents as an electrical 3rd order. An L-R contour is used up front to conquer the baffle step dilemma. It ends up being a 3rd order electrical filter with some damping on the shunt capacitor to soften the knee of the curve a bit. To match the curve, more would have been necessary, but following the target exactly didn't result in as good of an overall response.  The “Zobel” isn’t exactly that – it’s overly large with a higher-than-normal damper on it (8 Ω) rather than typical target of "nominal" impedance (4 Ω in this case). The value of this compensation along with the actual filter was a bit of a dance to get things to play nicely with the intention of trying not to present a terribly uncomfortable overall system load to the amplifier. The end result of the system impedance stays above 4 Ω the entire time, save the small dip to ~3.5 Ω around 90 Hz from the woofer’s raw impedance.

To keep series resistance as low as practical, I used 18 awg laminated iron core inductors. On the second series inductor, I unwound a 1.0 mH iron core inductor and ended up with half the DCR of a 0.51 mH 14 awg air core inductor at 30% the cost of the large air core. Since I already had one iron core in place, my midrange distortion was already going to take a hit. This was an all or nothing decision. 

The capacitors in the woofer circuit are all NPE type, save the 1% precision 6.8 uF in the notch filter to ensure correct placement of the narrow notch. The series 4 Ω resistor in the contour filter is two 8 Ω 10W resistors in parallel for a combined 20W rating. More probably would be better. All these resistors are standard 10W sand cast low-inductance wire wound type. 

Tweeter Network

The target tweeter response is 4th order Bessel @ 1400 Hz. This low of a target is a bit overzealous for the D28 (or any standard dome tweeter for that matter), but this theme challenge didn’t exactly expect tweeters to be lounging around. North Creek recommended at least 1500 Hz @ 12dB/ova slope. The softer knee of the Bessel (compared to BW4 or even LR4) also helps reduce output of the tweeter. 

The tweeter crossover is generally a 2nd order electrical network with the addition of a small inductor out in front to be able to tame the rising top end startign at ~8 kHz a bit. I included an oversized “Zobel” network to damp the resonance peak and get it the response to cooperate a bit better. This was cheaper and less complicated than a full series notch damper, and it still accomplishes enough damping of the peak to do what was needed. An L-pad approach may have also worked (parallel resistor can soften the resonance bump quite a bit), but I got a bit more control this way and an NPE type cap here is rather cheap.

For those that are concerned about such things, the series resistor was a 12W Mills metal oxide type. The main filter cap I used was a Clarity Cap CSA series 250V. Inductors are standard bargain 20 awg air cores. As with the woofer circuit, the resistor in the “Zobel” compensation circuit is a 10W sand cast type. The cap in the “Zobel” is NPE type. 

Crossover Board

Here’s a pic of the assembled crossover. I adopted this method of peg-boarding with ring terminals and brass fasteners from George at North Creek, which is a nice tie in. 

Note: Yes, I’m aware that the inductor layout is not ideal for the two in bottom left. This was the "let's get this thing assembled" stage and wasn't being critical enough at layout. Troels shows some measurement impacts here to inductance with coils in various positions. Turning the 0.5 mH coil on the on end was the eventual resolution after I noticed it in the cabinet, but no pics available. 

Baffle

I created a baffle based on the dimensions of the TW-0.75 cabinet to just drop it in, taking advantage of the BREX system (baffle replacement) of their prefab cabinets. I love this feature to be able to repurpose the cabinets and try different things out. 

See the file below for more detail. 

I came across some leftover pieces of recovered 140-year-old barn shiplap siding (that’s what the selling claimed, anyway) that I had used to build an authentic barn door for our office. It was severely weathered pine, had some remnants of red paint (before going through the planer), and surely looked the part. I found a couple scraps of ¾” plywood and glued that to the back of the barn wood to stiffen things up. While beautiful, this old barn wood is very soft and heavily weathered. The plywood added a lot of needed strength. I didn’t want plywood edges exposed (the rabbet on the TW-0.75 cabinets is only 15/32” deep), so I made the plywood doubler on the back 1 inch smaller on each side and wrapped a boarder of the barn wood around the plywood. 

Below is the glue-up for the baffle. Each baffle was two pieces butted together, seem vertically down the middle. I intentionally left some of the weathered look to come through and feature in the end.

After routing the holes and rabbet, I used a reverse 45 deg router bit (thanks Javad!) to provide some relief around the woofer cutout.

Main Cabinet

After building the above baffle, about a month slipped by. Ambition was low. While intentions were good to make some progress on the 4-way project, I decided to focus on completing one project at a time. Working on the 4-ways would have delayed this project for several more months, and that would have made a tight timeframe for the competition in April. So, I ended up building a complete cabinet from scratch for the Barn Finds rather than just putting a custom baffle on the Dayton pre-fab boxes. Since I was making my own cabinet, I took the opportunity to enlarge the internal volume to a full 1.0 ft3, a preference to get even more low end out of these drivers since I had plenty of margin under the 1.5 ft3 limit for the competition. 

In keeping with the limited cost impact goal, the box is made from standard, low-grade, void laden, 5-ply ¾” plywood… because I a sheet laying around in the garage from another project. 

The walls of the cabinet are single panel thickness, save the front baffle. Externally, it comes out to 10” W x 17” H x 15” D, 1.0 ft3 internal, fb tuning @ 30 Hz, f3 models around 33 Hz using a 2” ID port @ 8 1/8” long.  

Reinforcement gussets were added to the corners to have material to screw the baffle down to. Since I was no longer tied to the TW-0.75 cabinets, I added an extra brace across the middle to screw the baffle down and better compress the foam gasket. An additional brace was added in the middle of the side panels for extra rigidity. Pocket holes helped facilitate faster assembly than having to wait for glue to dry. Clamping 45 deg corners is also challenging.

Damping and Assembly

The crossover boards were screwed to the bottom of the box. F-26N ¼” wool felt damping material was staple-gunned to the sides, top, and back walls; the back piece is draped down across the top of the crossover for good measure.