Oliver centre-channel speaker
Gattiweb
The way is not in the sky. The way is in the heart. - Buddhist quote
Design & simulations

I've questioned the need for a centre channel for a a long time, thus far being content with a "phantom" centre channel provided by the stereo front speakers. However, I have finally decided that dialogue clarity and off-axis performance could benefit from the presence of a compact speaker located above the TV. This design sets out to put that theory to the test.

Its specifically designed to sit on top of my 50"plasma TV, being flush with the face of the TV, minimising diffraction effects. Basically I've arrived at this shape as a compromise between a horizontal WTW and a vertical WT, minimising the lobing effects in both axes,

The tweeter is the Seas T25C003, woofers are the Seas W12CY003.

The tweeter has a shallow horn, necessary to allow alignment of the acoustic centres with the woofers. Additional benefits are less diffraction and greater power handling.
The twin woofer design allows closer proximity of the tweeter (which will have its face plate routed) , while giving a cone area equal to a single 6" driver.
The pyramid shape also looks less bulky when sitting on top of a flat screen TV.

 
W12CY003 response and dimensions
T25C003 response and dimensions
Unibox simulations
Below shows the expected parameters for 2 x Seas W12CY003 parameters in a 4 litre enclosure. The sensitivity is 92dB for 2.83V and -3dB point is at 118Hz.
Cone excursion is greatest at resonance, 100Hz. With a 3mm Xmax, the maximum SPL (sound pressure level) is 98dB at 100Hz, occurring with 30W input. With an AV receiver setting of small for the centre-channel (2nd order highpass filter with -3dB at 100Hz), power handling doubles to 60W at 100Hz.
The impedance of the 2 woofers in parallel peaks at 100Hz and drops to 3 ohms minimum. With the crossover in place, I'd expect that to increase to around 4ohms seen by the amplifier.
Crossover design
For the crossover, I'm going to use a linear-phase topology as outline by Joe Rasmussen with his Elsinore project. Basically it involves using a 1st order filter for high and low pass, offsetting the tweeter, and inverting it's polarity. This seems to give a smoother frequency response in the midrange with real world drivers (which have their inherent rolloffs). It sacrifices some phase linearity above 10kHz, but here phase linearity becomes academic since wavelengths are only a couple of centimetres. For more details on this topology, see Joe's explanation here.
Here's the LspCAD predicted response using the measured driver responses and the filters described below.
W12CY003 woofer filter (note that R1012 and R1021 are inductor parasitic resistances).
T25c003 tweeter filter
System measurements
Below are the total system measurements, taken at 1 metre on the listening axis, with the speaker sitting on top of my 50" plasma TV and pushed up against the wall. i.e real world conditions.
Measured impedance below shows a nominal 4 ohm load with a 3 ohm minimum at 900Hz.

Off axis measurements

Below is the layout in my listening room. All on-axis measurements on this page have been taken with the microphone on the listening axis (at 1metre), which is 8 degrees below horizontal axis running from the bottom of the front baffle (130cm from the floor). I sit about 300cm away, on a couch that is 200cm wide, with my ear 90cm from the floor. The main speakers shown are for illustrative purposes.
The vertical off axis response shows excellent behaviour above the listening axis, but quite poor behaviour below it. I therefore decided to tilt the speaker down 2 degrees (put adding 5mm foam on the backwall support) to improve the frequency response for shorter or slouching listeners.

Horizontal off axis measurements show that the response stays smooth within 15 degrees off-axis. At 300cm from the speaker, that equates to +/- 160cm coverage, which is approaching the width of my 200cm couch.
Listening impressions

For the past few days I've been mainly listening to music in mono with no highpass filter (i.e centre channel set to LARGE in my AV receiver). Just Oliver alone to really be able to concentrate on its sound and capabilities.

The Seas W12CY003 nextel bass drivers are very nice and clean sounding, and have excellent detail, probably more than my uber-expensive 5.5" Audiotechnology drivers. Maybe a tiny hint of paper sound but who knows for sure, all materials have a certain character and its hard to say which is more accurate.

Solo female vocals on SACD sound absolutely incredible and the rock-solid image when listening in mono was literally like the performer was standing in front of me. The best centre image I've ever heard! Maybe the height of the speaker (1.3m from the floor) also contributed. Vocal coherency and central imaging is really the raison d'etre for centre speakers and these certainly lived up to that expectation. A regular stereo pair sounds quite diffuse in comparison, especially if seated off the central axiis.

I havent heard any unpleasant distortion even at very loud levels (which I almost never subject myself to) - and remember this is with Oliver alone being driven by a full spectrum signal. However my listening room is fairly small (2.5 x 4 x 3.5m). You really onlly know when the little 4" woofers are getting close to their lmits when some hardening and compression becomes apparent. Setting the AV receiver to a SMALL centre channel will obviously improve this considerably, as will the assitance of all channels in a full 5.1 or 7.1 configuration.

The T25C003 tweeter had a harsh sound at first which concerned me greatly. However after loosening it's face, moving the voice-coil slightly and torquing things back evenly eliminated that harshness completely. I guess I must have off-centred the voice-coil during reassembly after working on the face-plate. Now it's clean and refined, but ultimately lacking the air and detail of a Dynaudio d21/2.

Off axis, sitting at the extremes of my couch, things remain basically unchanged as expected from the above measurements. Only when standing, or being situated wider than the couch, did clear tonal shifts occur. Regular MTMs with large drivers and high crossover points will suffer in this test.

So to summarise, I'm quite happy with the final result. Going back to the original design goals, I have to now say I'm definitely a convert to centre-channel speakers.

part 2 - enclosure construction home