A Combined Center & Subwoofer

I recently made a center/subwoofer loudspeaker for my surround sound setup using some spare parts I had lying around from some old loudspeakers.

Since my living room is quite small, I decided to make the subwoofer a combined TV furniture and loudspeaker, thereby saving some space.

The enclosure is sealed and the filter is based on a series connection of the loudspeakers;
a quasi-stationary filter!.

The center/sub/TV bench.

Use the links below to read about the system design.

My 5.1 DTS system.
The Loudspeakers (Spare Parts).
The Loudspeaker Enclosures.
The Filter Network.

Preface: My 5.1 DTS system

Like many others I am the happy owner of a DVD player and enjoy watching movies and concerts in surround sound.
The 5.1 DTS system requires 2 front speakers, 2 rear speakers, one center loudspeaker and one subwoofer.
My apartment does however not provide the ideal dimensions for such a space demanding setup. The rear speakers should ideally be fullrange (like the front speakers). The same goes for the center speaker. The subwoofer, however should only represent the lower frequency range (.1 is for "below 100 Hz"). The rear speakers should provide a fullrange reproduction, but since our sound localization is definitely more sensitive to sounds originating in the forward semisphere than the backward semisphere I decided that a good pair of satellite speakers would do.

Schematic of my setup arrangement for surround sound reproduction.

The Loudspeakers (Spare Parts)

In my opinion deep bass is not necessary for the reproduction of voices and therefore I chose to use a midrange and a tweeter. I had a magnetically shielded tweeter from an old B&O loudspeaker set lying around and I also had two Vifa units (Vifa M13SG-09-08) which are also magnetically shielded. Therefore, it was obvious to integrate the center speaker in a TV furniture.
From that same B&O loudspeaker I also had an 22cm Seas bass driver and this proved not to be significantly deteriorating to the TV if mounted sufficiently far back relative to the screen. Therefore, I decided to combine the center/subwoofer in an integrated TV furniture.
Since I have no idea about the Thiele-Small parameters of the tweeter and the bass loudspeakers these had to be measured/calculated. This can quite easily be done if the mechanical free air resonance frequency is known. By measuring the resonance frequency and the -3dB points around the resonance frequency and afterwards performing the same measurement but with a known weight applied to the diaphragm the mechanical compliance can be isolated. The acoustical parameters were measured using my acoustical measurement system.

The Enclosure

A schematic of the initial drawing for the enclosure can be seen below.

The initial drawing of the center/sub/TV furniture
The enclosure material should be 22 mm MDF which is a very dense and acoustically "dead" material. The top and bottom panels should be angled 45 degrees so they could be fitted with the sides, the front and the back panels. The front baffle sides should be angled 47 degrees and the back panels should be angled 43 degrees thereby leaving a little place at the junction point for easier assembly.

Like it always is with my diy projects, I end up choosing another path than first established; either due to limitations regarding materials or limitations regarding my skills as a hand craftsman. In this case, the deviation was caused by a lack of materials rather than skills and I ended up with the implementation illustrated below.

Implementation of the enclosure (seen from the top)

A 10 mm foam layer is applied at all visible surfaces of the enclosure and serves mainly two purposes:
Primarily: errors in the enclosure finish are easier consealed but the furniture will also act as an absorptive element inside the room (rather than a reflector). The intention is to cover the enclosure (with the foam applied) with a thin layer of acoustically transparent sheet, which in my opinion will raise the WAF significantly.
The WAF (Wife Acceptance Factor) is in my opinion ok as long as the sound quality is not sacrificed for the purpose of goodwill, and in this case it will possibly narrow the radiation pattern of the tweeter, -perfect for a center loudspeaker. The front view is illustrated beneath.

The center/sub/TV furniture in frontal view
As can be seen, I changed the location of the tweeter compared to the schematic. In the implementation, I located it asymmetric on the baffle, which provides as less distinct baffle step, i.e : a flatter frequency magnitude response.
The enclosure for the center is approximately 12 liters and the enclosure for the subwoofer is approximately 40 liters.
I have used 40 mm very dense fibre board for the front and back panel and 15 mm MDF for the remaining panels. The cavities are separated using 15 mm MDF and Poly-filla was used to seal the cracks. As it can be seen, only the top and the bottom of the enclosure are parallel, which means that the risk of standing waves is largest in the median plane.

To overcome this potential problem, I braced the top and bottom with a think rubber layer glued to the surfaces. Light Rockwool (now called Stenuld) was used for damping material in both cabinets and this is primarily located at the sides and in the center where the particle velocity is largest. The damping material provides an increased mechanical compliance and loss of the loudspeaker drivers aswell as attenuation of standing waves/reflections.
Effectively: The damping material lowers the bass resonance peak and attenuates standing waves/reflections inside the enclosure.

The bas driver unit was mounted in the bottom of the enclosure.

The Filter Network

I have decided not to apply active filtering for the subwoofer. My surround amplifier (Sherwood Newcastle 956R) can deliver 140W into 8 ohm loads in 7 channels, and I therefore assumed that no extra power was required. It could be fun to implement a switch mode amplifier in an active filter for the subwoofer and thereby have the freedom to change the subwoofer phase in order to optimally integrate the listening room.
Unfortunately active filtering is quite expensive, so that will have to wait.

There is not much point in applying passive filters (that introduce energy loss) since the signal is already bandlimited to low frequency sounds. The only purpose of a passive filter would be to try to invert the phase for very low frequencies and thereby compensate for the lower frequency rolloff (180 degree phase trurn) of the bass driver. Unfortunately, HUGE coils and capacitors would be required for that purpose, so instead I simply used no filter for the subwoofer.

The Filter for the Center Speaker

As mentioned, I have used a 1 st order quasi-stationary filter for the center speaker. I did'nt find it nescessary to add components to this filter although the impedance is far from constant. That means that this a true quasi-stationary network with a guaranteed unity voltage gain. To my knowledge, this configuration is rather rare, because most people are conserned about the resulting "ugly" impedance.

But consider this: An audio amplifier is a constant voltage source. Of course this is only true to within limits, but do not fear a lumpy impedance curve. The amplifier only has a hard time when challenged at low impedances and extreme phase turns/high Q peaks resonances. Smooth impedance rises/falls are not difficult loads for the amplifier.

I think the misunderstanding about / fright of ugly impedances comes from the analytical design approaches related to typical parallel filtering schemes, where the models only work for purely resistive loads.
(loudspeakers are NEVER purely resistive by the way)

During the design phase, I used analytical models of the impedances of the drivers (ofcourse based on measurements).

Impedance of the drivers

Analytical model of the driver impedances while mounted in the enclosure.
Ofcourse I included an approximated analytical model of the acoustical response and trimmed using the typical "trial & error" approach.

Acoustic output of the drivers

Analytical approach of the drivers acoustical output.
From the measurements, I found that 3500Hz was a good place to cross the drivers (midrange & tweeter). The sensitivity of the tweeter is 89 dB SPL which is only slightly higher than the 88 dB SPL sensitivity of the midrange drivers, so it would be obvious to connect the midrange drivers in series thereby preserving the 88 dB SPL sensitivity (total of the two units). I considered whether a baffle step compensation should be included or not, but in fact the baffle dimensions are too small for the baffle step to matter. Besides, the baffle dimensions are uncorrelated and the loudspeakers are mounted asymmetrically on the baffle, whereby a possible baffle step would be smeered quite well.
During the simulation phase, I found out that the inductor had to be very small in order to keep the crossover frequency high and still maintain a reasonably attenuated circuit, so I determined to fixate the 0.27mH 1mm air core coil and fit the capacitor to this inductor.
I was fortunate to have quite a few capacitors lying around from former loudspeaker projects: 2.2uF, 3.3uF 3.9uF, 4.7uF, 6.8uF 8.2uF, 10uF, 15uF, 22uF and 33uF. Based on the simulation I preferred the 22uF, but I liked the sound of 10uF better (both capacitors are 5% Audyn 400 V capacitors) The simulation showed more ringing (underdamped) whereas the 22uF gave a smoother coherence (around 90 degree out of phase summation)

The used filter

The filter that I ended up using.
It sounds much more forward and dynamic with the underdamped system which I think is a general rule of thumb regarding underdamped networks: Transients are better reproduced at the cost of frequency magnitude constancy.
The electrical transfer function of the center speaker

The electrical transfer function of the center speaker

Electrical transfer function of the center speaker.
The acoustical transfer function of the center speaker

Acoustical transfer function of the center speaker.

Impedance of the center speaker.