That might be ideal . . . but . . .
Thanks much for keeping me on this train of thought for so long in PM and now again here. I have come up with a couple of thoughts with the project simmering on back-burner thanks to you.
#1 We have examples of a very simple, single PAL (+whatever?) NuBus adapter for the IIsi from the folks at Radius in hand which "just works" in the SE/30, albeit at only 8MHz. In this most simple of implementations, the PAL halves the 20MHz clock of the IIsi in order to produce the 10MHz clock required by NuBus.
#2 What I've been thinking is that someone with the chops to take a look at the timing mechanisms of NuBus would be the key to this predicament.
#3 The first release Macintosh II's NuBus (pre-NuChip) discrete component implementation is well documented. There are ample examples of that board available for analysis. What needs be done would be to de-couple the 20MHz to 10MHz clocking loop from the PAL of the Radius adapter to align it with the 16MHz clock of the SE/30 in the same (or similar) manner as was accomplished in the first rev. of the 16MHz Macintosh II.
Head's not clear ATM, so the above is a give or take kinda screed and I'll be away from my library for another week at best. Maybe
@Bolle or
@joethezombie can chime in with more info in the meantime on the Radius adapter? Likely with far more more technical observations than I can manage. I'm more an idea kinda guy, not a technically competent type of hacker in this day and age. I'm sure someone could implement the NuBus adaptation in CPLD/whatever should they wish to take a crack at it.