As in real life, loudspeaker building can take a sharp turn and the Excellence Two – Mini Monitor is no exception.
When I listened and tested the loudspeaker with my DEQX in an active setup it sounded fantastic. It worked best when I used a 48dB/octave linear-phase cross-over with a cross-over point between 1750-2000Hz.
With that in mind I started to design and build a first prototype of the cross-over with steep cross-over slopes (4th-order LR, acoustical) and with a cross-over point around 2-2.1kHz.
Everything was looking good, both frequency and distortion measurements, but something was missing…. Yes, the loudspeaker sounded good, but the magic was gone!
I used a “Sledgehammer” Steel Laminate Inductor for the mid-woofer section of the cross-over in order to keep the inductors “DCR” as low as possible. I quickly swapped this inductor to a 12 AWG copper-foil inductor instead (expensive).
The effect surprised me. Now the loudspeaker sounded much better. Some muddiness disappeared and the mid-range clarity was much better, but still no magic! I expected more of these drivers after listened to them in the active setup.
First of all, even though these “Sledgehammer” Steel Laminate Inductors have a good reputation they will never be used as a cross-over part in one of my designs. Perhaps they would work in the woofer section of a 3-way cross-over, but I would never use them in any mid-woofer or mid-range section of a 2 or 3-way system. They have poor sonic performance in my opinion and when using a high resolution driver as the SEAS Excel, they just don’t work in this design.
What to do? Scrap the project……. Of course not! Back to the drawing table and build the v2.0 cross-over.
Ok, a steep passive cross-over for the W12CY001 mid-woofer killed the midrange resolution and even though the cap shunt over the inductor technically worked in reducing the 10kHz cone break-up, it didn’t sound good enough.
This time I simplified the cross-over even further. I used a single inductor together with a LC notch to reduce the cone-break-up at 10kHz. Now the acoustical cross-over slope is a 2nd order LR topology. The mid-woofer section has the same number of cross-over components, but the components connected in series with the signal path are reduced to one. That’s a good thing sonically.
The tweeter section only needed a minor change to fit the new design. It was as simple as removing one cap and the new cross-over only needs 8 components to achieve the desired cross-over point at 2100Hz.
The v2.0 cross-over consist of a single inductor (L1) and with the above mentioned LC (C1, L2) notch-filter for the mid-woofer section. The tweeter uses a single cap + inductor (C2, L3) and has a response shaping circuit ( C3, R2) in order to reduce the tweeters rising top-end. The single resistor (R1) is used to attenuate the tweeter level to an appropriate level and personal taste.
The cross-over is an asymmetrical filter design with a mid-woofer acoustical 2nd order topology and a tweeter 3rd order topology. The asymmetrical filter design together with the relative acoustical off-set between the mid-woofer and the tweeter makes a perfect acoustical 2nd LR cross-over filter with a 180deg phase shift at the cross-over point, which means that the tweeter must be connected in reverse polarity.
Left picture: Cross-over v1.0
Right picture: Cross-over v2.0
Left picture: System — 85dB @ 1m
Right picture: System — 90dB @ 1m
A tiny bit higher distortion numbers compared to the v1.0 cross-over, but it’s still very low.
Blue: Filter v2.0 without notch (10cm near-field measurement)
Red: Filter v2.0 with notch (10cm near-field measurement)
The notch-filter reduces the cone break-up by 25dB. The “knee” between 5.5kHz and 7.5kHz are smaller in the far-field measurements and isn’t as pronounced in the off-axis measurements.
15deg off-axis cross-over simulation.