Depends on the XO point/slope WRT predominately baffle, horn mouth width and woofer size same as with point source drivers with the caveat that a WG or horn will have dramatically more directivity for a given desired BW than a typically much smaller point source driver, greatly reducing the need for any BSC per se, with whatever is needed usually being accomplished by the XO design and/or horn adjustable shelving filter. The M19's, 846B's XOs are good examples. For < ~500 Hz XO points, HF shelving combined with some form of CD horn EQ is usually sufficient.
In short, you can't go wrong by making a horn baffle of the same width [and height if there's room] as the cab, but in most cases there's no need for it in a HIFI/HT app beyond having a highly absorbent one to reduce mouth reflections back to the driver as much as practical. By highly absorbent, I mean more like the thick fiberglass lined early M19s, 846Bs, etc. than the mostly cosmetic sculpted ones used later on.
Unfortunately, I didn't get to do much foam testing way back when and don't know the density of the two different layers of sponge material I was fiddling with back before the lightning strike, so don't have a clue what might work for even 511/500 Hz XOs; so at this point the only foam I know of that will work is something called trail pad and then only if wrapped around the mouth at the flange to form a crude ~12" deep mouth ~90 deg included flared extension which is about as anti-WAF as anything I can think of ATM, though Pano-Mike's T-shirt? experiments might be right up [down?] there too.
GM






Bookmarks