Here were my thought processes in coming up with my 'design':
Vintage speakers use large woofers in sealed enclosures and I like the sound of them, notwithstanding their other drawbacks. I've yet to hear a modern, ported, slim line speaker that doesn't eventually fatigue the ears in some way, although they seem miraculous for their size at first. Could it be the small long-throw drivers and new fangled ports that are the problem? In my case I could build my speakers any size I liked, so I settled on 12” woofers, finding some Skytronic 902.222 with suitable specs available for £49 the pair. Very impressive-looking, but rather utilitarian on the cosmetic front. I figured that I should mount the woofers as close to the mid and tweeter as possible in order to minimise interference effects. Conventional wisdom these days is to put sub-woofers close to the floor even if the other drivers are higher up, but I figured I could turn my woofer enclosures upside down to experiment with this, if necessary.
There's no way I could begin to contemplate submitting to the pain of designing conventional passive crossovers, and anyway they offend me on principle! The technical advantages of active crossovers are well known, and using them would provide me with a shortcut to becoming an expert(!) on speakers, in that I could try out more permutations of crossover frequencies, slope, delay, levels and filter types in days than a conventional speaker builder could try out in years.
DSP-based crossovers were the obvious way to go
ultimate flexibility, PC commodity sound card prices, no soldering involved, and the opportunity to educate myself in writing DSP audio software, adding on whatever features took my fancy as the project progressed. At the back of my mind was the idea that I could measure and correct the drivers for phase response, and possibly experiment with room correction.
With active DSP crossovers, there was no reason not to build three, or even four way speakers, thus relaxing the requirements on the drivers themselves.
As regards questions of whether the baffle should be
wide all the way up from the woofer, or tapering, the decision was made by my next shortcut: to use existing vintage speaker cabinets and just add on my own baffles, thus avoiding a large amount of woodwork and expense (raw MDF costs more than old speaker cabinets) and for maximum experimentability I wanted the drivers in separate enclosures. In fact, I ended up with mids & tweeters in a pair of bookshelf enclosures that I happened to have lying around, and the woofers in separate huge enclosures.
I had an idea that I wanted the mids to go up fairly high in frequency (say 3 kHz), so they shouldn't be too large. I saw some cheap Peerless SKO100 4” mids with polypropylene cones, like the woofers, available for £18 a pair, plus physically-small 1” Moncaor DT25N tweeters with neodymium magnets for £30 a pair that I could mount especially close to the mids. I bought these, thinking that if I were to blow any of them up I could easily afford to replace them. I could always go upmarket with more expensive drivers once the thing was working properly.
There was never any question of worrying about the finer points of driver directionality etc. I figured that a three way line-up based on these sizes of driver was fairly conventional, and that in terms of directionality I was going to get what I was going to get. Some people favour a wide baffle and will tell you why they choose it (Grimm LS1), and some favour the more omnidirectional (Linkwitz Pluto). Others favour horns, dipoles, open baffle, electrostatics. Some people like speakers that “excite the room” and some the opposite. Some speakers taper towards the top, and some don't. Some have separate enclosures for each driver (Kef 105). They all have their supporters and detractors - which maybe tells us something. My system was probably going to be a bit Kef 105-like, and also similar to tapered baffles, but not a million miles off the conventional floorstander either.