Taking heed of adverse comments some had levelled at my choice of driver (the Peerless), and only having taken only one out of its box fand that only for a few tens of minutes of use I was given a fair exchange deal for a pair of Scanspeak 26W/8867 which seemed to model well with MK's software. But scarcely any improvement: it was indeed the box that was the problem! Back to drawing board. same external design concept to start, but now same volume as IMF RSPM, including same starting cross section (previous had been smaller, possibly responsible for that closed-in sound), remodelled from that starting point again using MK's algorithms. Vast improvement! However, no matter what I did with the stuffing in terms of density, profiling, materials, it never sounded as good, nor as effortless as the IMFs, and so failed to satisfy.
I gave up - at least in terms of one of original criterion, WAF. Now it was serious: I copied the IMFs. Blow computer modelling, blow even the difference in driver characteristics, yes this was simply a dimensional copy as exactly as I could measure by reaching and probing as deep as I could. The only change I made was to reduce height slightly to keep internal volume and cross sectional area the same with no midrange enclosure. Damping started off copying what I could of IMF's (but limited by available materials and the fact that I couldn't tell what was in the middle section), then tweaked. Ahhhh... the TL sound was back! As good as or better than the IMFs? I can't tell, but bass now was enjoyable, clean and extended, though certainly not identical. Until I make the second one and try some extensive listening tests I'll reserve judgement. One advantage of the new is much greater power handling capability - my IMFs use the older 'racetrack' SP6171 version of the B139, which from memory had only been rated at 15W RMS, and which is all too easy to overload if playing at elevated volumes.
Mid & HF are in small satellite enclosure on top, SACD-style, though freestanding so they can be angled independently of the bass enclosure. Drivers are the ATC SM75-150 and Scanspeak D3004-66000. Triamped using Behringer DCX2496 active XO, with Musical Fidelity P270/P170/P170 amps. The DCX2496 allows tweaking of the response to compensate for driver response - or indeed for room response. For comparison I set for a near-level response measured at listening position but boosted towards bass end in line with general shape of the in-room response of my IMFs. Overall, with the whole system one noticeable difference was a feeling that singers were in the room, which was startlingly good, though at the same time there seemed to be a harshness when the midrange got busy - both to do with the ATC mid I suspect, and will leave me wondering which of the two I prefer - again, until the new pair are done and comparable as a complete system it is hard to be definitive.
One thing that the testing has done is to reinforce how un-true much of popular 'Hi-Fi' hype really is. For example, simply moving the listening position few inches either way can produce dramatic peaks and troughs of 10dB and more in the response curve,
I yearn for the sound that my IMFs had the one time that I played them in the garden a few years ago - maybe not surprising as they were designed in 'free air', and maybe that is when they get closest to the mythical 'flat' resoponse (depending, of course, on where the listener is positioned...)