Illu3

This project took me a lot of time and crossover iterations to get it sounding right. After some trials, I had to put it aside, and give it a time to get energy to try again. But after that all it paid off. The most difficult part was to make crossover for 18WU due to its inherent shape of response at low frequencies. This shape is similar to 18WU aluminium version and also to Peerless NE265W and NE315W. When crossover is around 300Hz it is not that problem, but with crossover above 500Hz, it was difficult to find proper bafflestep compensation - woofer coil value, and to adjust midrange high-pass filter. After some attempts I decided to remove NF part of measurement from crossover modelling. With new crossover version it took right directions and only some adjustments of midrange and treble level were needed. Also lowering of bass-midrange crossover frequency down below 600Hz proved to be the right step.

At the beginning I used D3004, but I was not sure about the sound, and same happened with Satwavess project. For both, I swapped tweeters to R3004 versions, and all was suddenly right. So, although R3004 has quite poor off-axis dispersion, I do not detect it as problem during listening, and I very appreciate its mild smooth yet detailed sound that has certain top end sparkle making the listening enjoyable and more importantly not sounding dull.

There is also some, probably vertical, standing wave in cabinet, showing as small discontinuity at 220Hz. This was partially mitigated by damping material in cabinet. After that, listening did not confirm any issues with that.

With this project it was also good opportunity to try new method of obtaining measurements for crossover modelling. For a long time I have used common mic position, knowing, that it has accuracy problem with constructions where woofer is further from midrange. I followed Vituix v2 concept of measurement on individual drivers axis and filled the right numbers in XYZRT fields for all drivers. After that I could compare the results of both method. Overall axial FRs matched well, there we just minor differences. The major difference was in phase tracking between woofer and midrange, the phase difference was ~30deg, so I had to adjust crossover to address that. And this was expected.

Drivers measurements

All measurements for crossover design were done with gatting 4.4ms, using Kimmo Saunisto method of measurement on individual drivers axis, not common mic position method. 

Crossover modelling

Crossover is after all iterations simple, I removed some RLCs that made some things looking better, but they did not prove beneficial after listening tests. It is pleasure to work with drivers that does not require heavy hammering and complex crossover to achieve required acoustic response. In this case they all fit very well together, crossover-wise.

Results are displayed in CTA-2034-A standard. Listening distance 2m. Vertical measurements are included.

Listening distance 2000mm

As crossover is all LR2, mid drivers has to be inverted, and this is shown on the left. What happens, when midrange is connected with +to+ polarity, is shown on the right. 

.....and just for a comparison......

Horizontal and vertical polar maps with contour lines with 3dB steps. That kind of vertical polar map is reality of multiway loudspeaker with physically separated drivers.

Photos

The first version was tested with D3004 therefore some photos with that tweeter.