On Tue, 17 Apr 2012, Dave McGuire wrote: Well, I /am/ psychic. But it was also discussed here about a year ago. -- David Griffith dgriffi at cs.csubak.edu A: Because it fouls the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail?

From: Ethan Dicks: Tuesday, April 17, 2012 7:35 PM I remember talking about it, but I'd have to look at my archives to recover what was said. I do have a drawing of an R202 on a 1"x2" board, with a 20-pin right angle header instead of the 18-pin edge connector. I had also done a drawing for an adapter, which would allow you to debug by replacing individual modules in a real -8 or -8/S with their SMT equivalents. It looks like at some point I started an SMT R210 drawing, but I must have gotten distracted, as it never got done. Yep. I lean toward doing the interconnect because of the "cool factor" of all the tiny little boards looking a bit like a miniaturized version of the original. I think it would be more expensive, but easier to debug. Of course, you probably end up with a "big board" for the backplane, but I can actually generate that auto-magically from drawings of a full-size backplane. Still, one would have to pay to get it manufactured and then to populate all the connector sites. Plus having all the module boards fabricated. One thing I am pretty sure of, is that you don't want gold edge connectors. Those are just way too pricey, and they tend to also move you out of the places that will let you send them a panelization. (Instead, I suggest the use of gold-plated right-angle headers to connect to the backplane.) Come to think of it, I do remember something about trying once upon a time to figure out how many 8"x10" panel layouts would have to be sent out to end up with all the right module quantities. I think that might have been for an 8/S design, though. As I say, I'd have to look in my archives. Got a front panel? Those are the really expensive part, I believe. I suspect one would spend about as much getting a nice front panel made as for the CPU guts. (Assuming standard 2-sided "prototype quality" service for the boards, and quantity purchases of a few types of SMT transistors, diodes, and such.) Oh, and another issue -- unless you redo all the component values, you'll end up generating the same heat in a much smaller volume. I'm not enough of an expert to understand the tradeoff of power against speed, but I doubt if one can get the speed without the heat. At some point you'd also want to drive the bus cables, and that will definitely need some power. OTOH, replacing the core memory subsystem will probably save a bunch of power. Vince