Hey guys, lots of you are questioning this point, and it was a little obscure.
- Miswired Midrange
Let me explain. On multi-way (2 or more) systems a crossover designer has to carefully match the slope and phase of drivers together at the crossover in order to achieve the smoothest possible hand-off. The goal is that your (for instance) tweeter and midrange are in complementary phase transitions, like Fred Astaire and Ginger Rogers, as one goes counterclockwise the other clockwise, but the end effect, like Ginger and Fred, is a smooth transition from tweeter through mid, so you cannot "see" or hear the crossover. That is, there should be no evidence in your frequency response of that point. In a perfect world, the crossover points are invisible off axis but honestly many high-end (giggle) speakers achieve only a variety of success there. The gold standard for this matching is when you reverse a driver you should get a deep null. (Note: Not 100% of crossovers are designed for null tests (see S. Harsch for example) .
In the Sabrina as measured by Miller, the crossover points are pretty obvious.
I use the tweeter to midrange as an example, but the same happens at the mid to woofer, woofer to sub, etc. and this happens regardless of crossover slope.
Should you accidentally ship a 3-way speaker with an inverted midrange, both filter points would demonstrate significant nulls. Here’s an example of a simulated "null test" I was using for my own 3-way:
The nulls happen because at the mic position one driver is pushing and the other pulling, cancelling each other out. In simulation and testing, deep nulls when you invert one driver is a good thing, it shows you have gotten ideal phase matching across your crossover points, but it is the opposite of desired end-state.
It is similar to what happens when you accidentally flip the speaker wire at one end. The bass vanishes due to cancellation, but in the case above it’s a very narrow band.
In my example, above, the nulls are more pronounced and the lines more clear than many speakers would show because I’m using high order, symmetrical filters. What you see in the response of the Sabrina review is what I would expect if the midrange was mis-wired and the filter slopes were lower and non symmetrical. (lower and assymetrical is not a critique, btw, but ends in shallower, broader nulls), or the crossover points were only somewhat misaligned.
This may be exactly what they want to achieve, but if I was about to publish a review I’d have reached out to them first to ensure this was intended, and validate my measuring method, it's possble Wilson specifically designed them to be listened to at a distance further than measured, which would affect the results. It is really hard for me to see how they’d spend so much time and effort reducing distortion, and getting very custom drivers, only to produce a very non-neutral response in the end. The audibility of the frequency response is going to be much much more than the claimed distortion reductions.
Missing from this review as well, compared to JA’s practice, is the impulse response which might have told us much about the coherence of the crossovers. I’d suggest reading his old measurements for background.
It is also true that these speakers have a sound that buyers will chose or not chose based on how it makes them feel. I’m just... well... << sigh >> Over time I’ve heard and seen Wilson models with lots of character and less character and this is definitely a speaker with character all it’s own.
