Building Do It Yourself Loudspeakers
Background and History
Back in college I had the opportunity to A-B test a pair $600 Klipsch speakers, which was a lot of money when I was in college, against a pair of home brew speakers that were built from Peerless drivers at a cost of less than $200. There was really no comparison, the home made speakers were leaps and bounds better. I then realized that good speakers could be built at low cost. After graduating with an electrical engineering degree and starting my career in audio I had a solid grasp on audio basics. I built that same pair of home brews with the help of Warren Merkel, who designed them. He also wrote and published a speaker design program called Perfect Box. Later in my career a colleague built speakers using Dynaudio tweeters and after listening to his speakers I knew then that I also wanted to design speakers using Dynaudio drivers. At that time I was still early into my career and Dynaudio's were some of the most expensive drivers available plus I didn't have a good grasp of speaker design yet. Through the years I listened to as many speakers as I could in the area and what I found in general was that I didn't like hard dome tweeters. Although there were a few speakers with hard domes that I did like I decided to stick to building speakers with a soft dome tweeter. So after studying speaker design books through the years I was ready to build speakers with Dynaudio drivers. I don't claim to be an expert at speaker design but I learned that it is a combination of science and art.
Clarity was not a design goal because it is inherent in Dynaudio drivers. My first and primary design goal was imaging and sound stage. My second goal was speaker size. I didn't want speakers half the size of a refrigerator. My third goal was some good low end bass which resulted in making a 3-way speaker. Lastly, I wanted these speakers to look good.
The Balancing Act - Woofers and Tweeters Working Together
Designing speakers requires a balancing act and this is where the design gets tough. I had to find drivers that sounded good together, a woofer that played down into the upper 20 Hz range, drivers with closely matched sensitivities, and result in a box of medium size. I started with the famous and world renowned Dynaudio D260 Esotec tweeter. Keeping my primary goal in mind I then chose a Dynaudio dome midrange. The dome tweeter coupled with the dome midrange should provide excellent imaging and their sensitivities were within 1 dB of each other. It is important to consider driver sensitivity in order to keep the crossover as simple as possible. When their sensitivities don't match, typically, additional components are needed to the compensate for the mismatch.
If you are not familiar with speaker design, the box size is based on the woofer parameters. Using Perfect Box software I modeled different woofers and observed the frequency response compared with the box size. I started with Dynaudio woofers but after modeling them, their box size turned out to be larger than I wanted. This was some what disappointing because I wanted all drivers to be Dynaudio. I also modeled Scan Speak woofers but their box size wasn't conducive either. I finally settled on a Peerless 10" Woofer. It was a company I was familiar with and like Dynaudio was a Danish company which made quality drivers. The box size met my goals and the sensitivity was within 2 dB of the other two drivers. Something I haven't mentioned that was a very important part of my design was the manufacturers response curves. I was looking for smoothness in the curves. Since this woofer was to be used as a mid-bass driver it also had to be smooth.
Attacking the Nearly Impossible - The Crossover
Now that all drivers were chosen the next step was to design a crossover. With all the variables associated with speaker drivers it is nearly impossible to design a crossover with a pencil, paper and a calculator. If you do, you will be tweaking them for a long time and eventually you'll spend more time tweaking the crossovers instead of enjoying music. A crossover modeling program is essential but it was not included in the Perfect Box program. Fortunately, Madisound offers a service for designing crossovers using the LEAP program from LinearX. I believe LEAP is the industry standard for crossover modeling so I used the Madisound service. After I researched many DIY speaker projects on the Net and studied the manufacturers frequency response curves, I chose the crossover frequencies without the aid of LEAP. This very important decision and different from most other designs was not to push the drivers to their on axis limits. In other words I chose crossover points where the driver was still flat even at 60 degrees off axis. This key decision is why I believe my speakers have excellent imaging and sound stage.