I found some time today to build up the crossovers for my friend’s new home theater speakers. I am by no means an expert on custom crossover design, and the small array of free tools I have at my disposal are all I know to design a better-than-textbook crossover for a pair of loudspeakers. It is probably the most complicated aspect of DIY speaker design, the least understood, and the most affecting of how a speaker will sound in the end. Even enclosure design takes a backseat to the criticality of crossover design. Just about anyone can build a box and toss in a port and be happy with the results. And most people do exactly this (myself included) using techniques that are basically textbook enclosure design. Lucky for most of us DIY’ers, textbook enclosures are amazingly accurate and perform almost exactly as modeled or simulated. But crossover design is an entirely different animal, one I am just barely trying to tame, one speaker project at a time.
This is basically how I designed this crossover: I started with the drivers; in this case a pair of 7″ Dayton aluminum cone woofers and a Morel MDT-20 soft dome tweeter in an MTM configuration. PE offers CLIO plots of the Dayton woofer for both impedance, phase, and frequency response which is great, it takes one less step out of the process. Since there are no data files for the Morel tweeter, I had to rely on capturing the data a different way, and that’s by using SPL Trace. With just a picture of the FR and impedance of the tweeter, I generated data which was then imported into a crossover program, along with the woofers.
So what’s the best free tool for designing crossovers? My favorite is Passive Crossover Designer available from the guys over at FRD Consortium. I loaded the impedance plots and FR plots into PCD and just started tweaking around. For these speakers I started with a x-over point of 1.8kHz and then hit Initialize Textbook values just to get a starting or reference point. Then from there I like to move things around just a bit until the combined summation response look as flat as possible. I tweaked around with the values for about 2 weeks before finally coming up with the final crossover values. Even these are probably not ideal, and the combination of possibilities is quite numerous.
The final crossover turned out to be pretty much a textbook 18dB/octave Butterworth crossover for the high-pass section right at about 1.9kHZ (with no padding) with a slight tweak on the inductor to flatten out the FR a little. The low-pass section started out as textbook but I ended up with a much larger shunt capacitor. I’ve found that in a lot of cases the cap required on the woofer must be a lot bigger to help roll off the cone break up modes quicker. Also the addition of the series 0.20mH inductor helps drop the cone’s break-up modes even lower without altering the crossover point, and thus providing the 18dB/octave Butterworth slope. The Zobel network is also basically a textbook Zobel with a 25uF cap in series with a 4 ohm resistor and wired in parallel across both woofers. The schematic is shown below. Very simple, nothing fancy with standard values which are easily available for purchase.
So I’ll update with more details and measured results of these crossovers. As of this writing the speakers have actually been installed in my friend’s home theater and we’ve already watched bits and pieces of a few different movies and they sounds fantastic. We were both really happy with the overall sound. But more on that later! So this is it, the crossovers completed.