B&W 802 D2 Diamond upgrade

All measurements were done with microphone distance 1m, and height 4cm below tweeter. No smoothing was applied. To remove room impact from measurements, impulse window ~4ms was applied. Absolute SPL was not calibrated.

Original state analysis

The set of horizontal FRs, 0-90degrees with 10deg steps.

This measurement shows what I have already expected. Too many energy/bump at 600Hz, suppressed midrange energy around 2kHz and the dip created off-axis. Tweeter provides very wide dispersion.

Step response

Power response and DI

I used VituixCAD to obtain Power response and DI. I expected both would not be good, but was interested how much not good.

Axial FRs of both loudspeakers

I have to say both loudspeakers measured identically, this is not easy to achieve. Mild difference at 20kHz is nothing to be concerned of, and can be caused by small change of mic position between measurements.

Impedances of both loudspeakers

As  expected from frequency response measurement, impedances also match very well.

Individual drivers measurements

Measured in box impedances of all drivers, without crossovers

Woofers impedance shows no signs of standing waves in the cabinet. Small peak at 2kHz is most probably the first breakup of the cone.

Midrange impedance is free of any resonances up to 3kHz, so this drivers would be preferably crossed over at ~2kHz. Resonant frequency Fs is 280Hz, and it turned out it had to addressed.

Woofer horizontal FRs for 0-90deg

Not much to add here. Dip at 1.5kHz is typical for two woofers measured vertically off axis.

Midrange horizontal FRs for 0-90deg

This is very nice and flat. Impedance did not suggest any problems up to 3kHz, and this was proved. There are no significant breakups, though there are multiple mild breakups at 3, 5, etc. Khz. So definitely better to cross over this midrange at 2kHz maximum.

Also dispersion gets quite narrow above 1.5kHz. It would be nice to use this midrange in combination with waveguided tweeter.

Tweeter horizontal FRs for 0-90deg 

What we see here is very nice, consistent and wide dispersion up to 20kHz. Though sudden change of directivity at 3500Hz will not make crossover modelling easy. 

Axial tweeter FRs, comparison of tweeter grid impact

I was curious to see the impact of removable tweeter grill. Basically, we can say, grill has no impact on axial response.

Woofer and port NF measurements

All measured without crossover

Red - upper woofer NF measurement, mic distance 5mm from dust cap

Green - between woofers NF measurement, mic distance 5cm from baffle

Blue - port NF measurement, mic placed on the plinth right below port

Magenta - port NF measurement, mic placed between cabinet and plinth, on the plane of front baffle 

These woofers are very well behaved, no signs of resonances up to 2kHz.

Also bassreflex curves show typical shape, and resonances above 200Hz are well damped.

Original crossover modelling with measured responses

I loaded measured drivers responses to Vituix and modelled original crossover. Nearfield responses of woofers and port were added, baffle response was factored in as well, before NF and FF merging.

Crossover images

Summary

PROs: 

- loudspeakers are perfectly build. Quality is simply superb

- piece to piece matching is very good

- all components are high quality, both visually and technically, especially midrange and tweeter drivers and their measured performance

- crossover components are all air coils, capacitors are various Mundorfs

- tweeter dispersion and behaviour up to 20kHz is great

- all the voltage rating of capacitors, nikkohm resistors, wire gauge and air coils signalize that loudspeakers were built to play loud and to be reliable

- listening in larger well acoustically treated room, on powerfull amplifier, gave big fullbodied sound. Minimizing the room reflections seems essential with these loudspeakers.

CONs:

- drivers integration and resulting axis and off-axis responses are, well...., nothing that loudspeaker designer could be proud of

- crossover is probably the simplest version that at least does something

- broad bump at 600Hz will have to addressed in new crossover, quite strange B&W left that as is

- tweeter response with simple capacitor is not optimal, moreover, there is no RLC addressing Fs impedance peak. Tweeter does not have easy life in this loudspeaker.

- midrange-tweeter phase matching is just so so, phases cross at Fc, but their slopes are different, and tweeter response gets out of phase with the midrange below crossover point. 

- off-axis dip between midrange and tweeter at 2kHz. This is the result of high midrange driver directivity, and tweeter out of phase below Fc

New crossover modelling

After listening tests and several crossover itterations it is clear that new crossovers are really step up from original crossovers. So original crossovers rest untouched on the shelf.

There are several aspects that make new crossover modelling difficult. First of all, I was aiming for mid-tweeter crossover around below 3kHz. Midrange dispersion at this frequency is very narrow, and quite opposite of tweeter dispersion, so we have two extremes here. 

Tweeter response is not easy to work with. 

Midrange phase is quite steep above 3kHz, and achieving good phase tracking above crossover seems not possible. I account this behavior to the series of mild midrange breakups, and this is visible on the phase graph.

Crossover compartment in the plinth is not large, and mid-tweeter PCB is quite small, so I try to keep crossover as simple as possible, and low crossover components count. 

So far, crossover slightly below 3kHz seems  to provide good on axis and off axis behaviour, with reasonably complex crossover.

Finished loudspeakers measurement

Generally, both loudspeakers show very good match of their frequency responses and impedances.