This construction was made before I started my diy-loudspeaker website, and I never found the time to write anything about it. The drivers here are among my few favourites and the W15CY001 drivers were acquired for the Point75 dipole speaker and the HIQUPHON OWI was bought only because I wanted to own a pair of these - probably - best ever made 19 mm domes. Thanks to Oscar Vroending at HIQUPHON.

The Point75s were sold to fund the Acapella adventure and it was only recently I again bought a pair of the W15CY001 midbass drivers. A shop in Copenhagen had a sell-out of SEAS drivers.
I had a mail from Michael Boyd expressing his surprise of the magnificent qualities of the W15 driver, but he found the overall balance of the construction well on the bright side - and had I ever tried the BBC voicing of this speaker? Well, I hadn't and looking at the old measurement I have to admit the overall balance is very flat and maybe voiced to be too much of a real monitor.

So, time to heat up the soldering iron and take a second look at this construction. The W15 and OWI were placed in my 7.5 litre test cabs and the OWI had a lot of Gaffa tape to cover a wider routing from other constructions. Test cabs are still 19 x 30 cm and driver placement is as seen on construction details below.

First of all I wanted to see if the original crossover (V1, not shown) could be modified by simple means so that replacing expensive crossover components could be avoided. By adding a resistor (R2021) in series with the 6.8 uF capacitor in the low-pass section, the 1-3 kHz range can be attenuated 1-1½ dB, having an immediate impact on the forwardness of the sound. Tweeter obviously needed some further attenuation, thus the 8R2 (R1011) is replaced by 10R and a 3R3 (R1061) is added in series with the tweeter.

Left: The treble level is now 1-2 dB below midrange level and provides a more natural balance. The response here is merged at 350 Hz with a nearfield measurement of bass driver and don't pay too much attention to the level below this point. Nearfield measurements do not tell the true level of bass at 1-2 metres distance, but it does tell where the bass driver starts rolling off.

Right: You may think that adding 3R3 in front of the tweeter changes the point of crossover and mostly it does, but here it goes well and the phase-tracking between drivers are kept in good shape. Reversing the tweeter polarity produces a deep suck-out at 3 kHz.

Left: For those interested, here are the response profiles of the drivers without crossover attached. The W15 is a doing fine and the cone break-up starts above 5 kHz. Excellent for a 5" magnesium cone driver.

Right: Significance of notch-filter to reduce peaks around 8kHz.

Left: Significance of mid notch filter.

Right: The most critical part of the crossover is the 0.30 mH coil in the 8 kHz notch-filter. This really has to be 0.30 mH. Red = 0.30 mH, green = 0.33 mH and blue = 0.27 mH.

Impedance of speaker construction. A reasonably easy load on the amplifier.

Hello Troels,  
I implemented the all changes to the W15CY001/OW1 project.  Briefly, these are my findings: Smooth as silk (beautifully integrated) from top to bottom, unbelievably accurate and effortless mids, and razor-sharp, slightly recessed imaging!  These things come so close to ESL performance on vocals that it is scary.
Frankly, I was pleased with your initial suggestions for revisions but your final changes really do the trick.
I will only take your word for the fourth-order version sounding cleaner because this version is simply amazing.  How amazing?  I "rediscovered" my whole Ella Fitzgerald collection over the last week.   Thank you so much for this OUTSTANDING project.
Michael