In the fall of 2010, my friend April and I took an Audio Engineering class (last class ever with Professor Leach, one of the best professors at Tech). It was very interesting, but it was only a theoretical class. Our final project was to design a speaker enclosure based on some woofer measurements provided to us. While interesting, we wanted to build some actual speakers. So we signed up for a one hour special topics class and were able to design and build some awesome speakers.
The objective of the project was to design two pairs of loudspeakers with a flat frequency response and a stylish exterior. The speakers would be two-way systems, with one woofer and one tweeter in each cabinet. Originally the project called for acrylic to design the cabinets, so we planned to use a closed, unfilled box design. We thought it would be really awesome to have some transparent speakers and a fun challange to make them work properly (there would be issues with vibration and such). However, after realizing the price of acrylic was far too expensive (it would have cost about $300-400 for a pair of speakers), we decided to go with a traditional wood cabinet design with filling and a port.
Next, we had to select a woofer to design our enclosure around. The woofer was selected based on its appearance, value and performance (since we were originally using an acrylic box). An 8” Goldwood driver was selected, which while not a top-of-the-line driver, still had decent performance and looked cool at a decent price. When the drivers arrived, measurements were taken to confirm their advertised specifications. However, the measurements didn’t match, even between four woofers that were tested. Two of the drivers had an fs (resonance frequency of the driver in an infinite baffle) around 45 Hz and the other two had an fs around 55 Hz. We paired the drivers up for each set so that they matched.
Additionally, simulations based on the driver parameters showed that the frequency response of the woofer would fall off around 700 Hz, which would not be adequate for a two-way system. A midrange would be required to have a flat response. We had already purchased materials so we decided to press on anyways. Fortunately, after the enclosure was designed, measurements showed that the woofer actually didn’t fall off until around 2kHz.
The enclosure was made of ½ inch MDF. The required internal volume for the woofer was calculated to be roughly 1 cubic foot. This includes subtracting the volumes of the internal bracing, woofer and tweeter internal volumes. All four speakers were made from a single 4’ x 8’ sheet of ½ inch MDF. Despite the calculations being made under the initial assumption that the box would not be filled, the box was filled anyways. There was no noticeable difference in the frequency response between the filled and unfilled box. The back panel is removable with screws, and a weather stripping seal prevents any air from escaping when sealed. The rest of the edges are sealed on the inside with silicone caulk.
The traditional approach to cutting holes in the MDF for the woofer and tweeter is to use a hole-saw. We didn't have many wood tools available to us, but I did have a friend in the mechanical engineering department who knew how to use the waterjet machine! That machine is usually used to cut through metal, but we were able to cut some nice holes in our MDF without ruining the wood by keeping it in the water for as little of time as possible. We had fun learning to use a router to make the box nice and smooth on all of the edges!
Our designs are actually slightly different. I configured my woofer to have the same spacing between the left, bottom and right sides, then centered the tweeter between the woofer and the top. April put her woofer and tweeter at a fixed distance, and centered both of them in the middle. It was a subtle difference, but we both had strong feelings about it!
The external box dimensions were:
The crossover network is what separates power between the drivers. Usually, a low pass filter will be installed on the woofer and a high pass on the tweeter. The crossover network is very simple in our speakers. The tweeter purchased for the system falls off around 1600 Hz, so the crossover frequency was selected to be at 1800 Hz (frequencies below 1800 Hz will be filtered out). The woofer actually had a frequency response that falls off after 2000 kHz on its own, no low pass filter was required. For the tweeter, a first order high pass filter was selected to match the slope of the woofer’s frequency response. A 15uF capacitor was used. The tweeter was connected in phase with the woofer.
A 4 dB L-pad was also used to attenuate the tweeter’s signal (to balance the levels between the woofer and tweeter). The L-Pad design consisted of a 3 ohm resister was used in series and a 3 ohm resistance in shunt. This L-pad resulted in a very flat frequency response.
The following components were used for each speaker.
The total cost for the speakers was around $150 for each pair of speakers. This includes all of the above components, the MDF, paint, wood varnish, rubber feet, wood glue, silicone caulk, and various screws.
Below is a frequency response graph measured from the speakers. We used TrueRTA and a measurement microphone (which has a flat frequency response itself) to take these measurements. An ideal speaker will have a perfectly flat frequency response, which means it reproduces all frequencies at the same magnitude as the source audio. Most speakers tend to fall off at the low end, since lower frequencies require more power and larger woofers.
We think the response is very good, as it's quite flat overall. Compared to cheap consumer speakers, this response is amazing! The speakers sound incredible and can handle a lot of power, so when paired with a good receiver, they are quite loud!
I went on later to measure the frequency response of some other speakers as well, so you can see that these speakers are quite comparable to high-quality studio monitors.
Writeup: April 2011