To be truthful, all that brainstorming about hanging IDE ports (or whatever) off PDS slots or ROM SIMMs aside my genuine opinion is for 90-whatever-percent of Ye Old Macintosh applications the "correct" solution is a bridge that hangs off the SCSI port. It's the one bus that all those machines have in common, it's a pretty least-common-denominator version of it (8 bit single ended, "Fast" mode optional), the drivers are finished, and while in principle it might be possible to eek *slightly* more performance out of one of those old boxes by cutting out the middleman and hanging something off a direct bus port... really, is it going to make a difference if your SE/30's disk performance is only 50-some-odd-times slower than a modern machine verses a hundred? It's still *amn slow, son.
There's one area, personally, where I think a direct bus port design would offer real value, and coincidentally it's also the area where the level of technological expertise required to achieve success is low enough that a relatively ham-fisted hobbyist-built design has a chance of working. That of course is something for the SCSI-less toaster Macs, IE, the 128k (although fitting a driver into 128k might be hopeless) through 512ke systems. If desoldering/piggybacking on the 68000 is too daunting there's still the ROM socket approach. (There's the documented John Bass SCSI design, and the Dove SCSI boards also worked that way.) This is something that fits into a small handful of TTL circuitry, doesn't require extensive modifications, and is a solution documented to work. (Minus the design change of replacing the SCSI chip with an IDE device, microcontroller, whatever, of course.) The only other options for those machines are Macintosh Plus upgrades (either literal or a combination of expansion boards that add up to one with the attendant need for a SCSI drive or suitable replacement *anyway*), or really unobtainable items like a working HD-20 or GCC Hyperdrive.
But, eh, you know it's *impossible* to do something like they did back in 1985 with a few TTL chips without resorting to an FPGA these days...
There's one area, personally, where I think a direct bus port design would offer real value, and coincidentally it's also the area where the level of technological expertise required to achieve success is low enough that a relatively ham-fisted hobbyist-built design has a chance of working. That of course is something for the SCSI-less toaster Macs, IE, the 128k (although fitting a driver into 128k might be hopeless) through 512ke systems. If desoldering/piggybacking on the 68000 is too daunting there's still the ROM socket approach. (There's the documented John Bass SCSI design, and the Dove SCSI boards also worked that way.) This is something that fits into a small handful of TTL circuitry, doesn't require extensive modifications, and is a solution documented to work. (Minus the design change of replacing the SCSI chip with an IDE device, microcontroller, whatever, of course.) The only other options for those machines are Macintosh Plus upgrades (either literal or a combination of expansion boards that add up to one with the attendant need for a SCSI drive or suitable replacement *anyway*), or really unobtainable items like a working HD-20 or GCC Hyperdrive.
But, eh, you know it's *impossible* to do something like they did back in 1985 with a few TTL chips without resorting to an FPGA these days...