Wow! This is particularly fascinating stuff 'cause as some of you will appreciate, I was part of the team responsible for the S-50 (and all the subsequent Canon Audio speakers). I won't go into any of that story here 'cause I've covered it in the blog posts (
https://musicandmiscellany.com) that you've been kind enough to link to. What I will do though is comment on the measurements and some of the discussion they've raised.
First, although I always had my doubts about the use of the the parasitic cone (the S-50 design was mostly frozen when I joined Canon Audio in 1990), I really don't remember the top end response of the S-50 being quite as uneven as the sample measured here. It was never great of course, but I'm pretty sure it was better than this. Similarly, the distortion performance was better – that 1.5kHz - 5kHz mess wasn't present. I wonder what the history of the particular speaker is and whether they've been well looked after? The parasitic cone shown in the photograph has definitely gone a strange colour (it would originally been the same colour as the cone) so I wonder if its mechanical properties have degraded over the years? Sadly, considering my decades working with speakers, I don't really know anything about the ageing of paper diaphragms and how their characteristics change.
The resonance of the dome is an odd one. There's a rubber o-ring at the join between the dome and the baffle, in place both to provide an air-seal and to damp the dome bell resonance, so I wonder if it has perished? It's also possible that the screws holding the dome in place (one on the underside of the baffle, two on the top under the central plug) have come loose - that might well be the case if the o-ring has disintegrated. Replacing the o-ring is, by the way, a pretty simple job.
Along with the HF unevenness, the measurements reveal what was of course the fundamental problems with the Canon "off-centre acoustic mirror" (and the many examples of axially aligned conical mirrors): Firstly, the mirror loads the driver over a narrow mid-band (I forget exactly where it was but I think it was in the high hundred Hertz) where the ear is really sensitive, so even if you can EQ it flat(ish), which was done on the S-50, it still tends to result in a characteristic colouration. Secondly, even though the mirror is actually a surprisingly effective dispersion control device, wherever you listen/measure you're always going to have a degree of direct sound arriving first, and that results in all sorts of audible comb-filtering effects. Despite the fundamental problems though, as described in the subjective assessment, the S-50 had some really interesting qualities. It wasn't really "hi-fi" in the sense that it produced an accurate reproduction of the source material, and that was never really the intention, but it could be a genuinely engaging and entertaining listen. I kind of wish I had a pair.
Lastly, having expressed doubts about the parasitic tweeter in my blog posts, I was contacted not long afterwards by an S-50 owner, who like me, had been wondering about the possibility of fitting a contemporary dual-concentric driver. Since the original KEF dual-concentric patents lapsed a few years ago there's been quite a few similar drivers launched, so the dual-concentric S-50 idea is now rather more feasible than it was in 1990. So, once lockdown is over and things picks up again the plan is to give it a go. I'll report back.
Thanks again for the S-50 measurements and chat! You've made my day.
Phil
PS. So sad that Allen Boothroyd, who did the S-50 industrial design, isn't around to see folks still chatting about it - he left us a couple of months ago. It was a project he was very committed to and very proud of.