If you put two connectors, one of 40 pins and one of 36 pins, end to end, I expect you'd do fine with a power and ground pair at about positions #5 and 16 on the 20x2, and about positions #5 and 14 on the 18x2, the way you have them numbered. (In IDCs, 34- and 40-contact are standard, not 36, but you're not planning for those or even allowing them at all anyway. The board-mounted headers and sockets OTOH can be cut or stacked to any length you want.) This way no pin will be more than 1/2" from a ground pin, and the pins in the middle will have a ground pin at that distance in each direction. That would free up at least 12 pins, and 14 if you cut down to one Vcc per connector (which I find a little uncomfortable). Header pin numbering, BTW, normally goes

1 2
3 4
5 6
7 8
etc.

looking at the header from the top of the pins. If you deviate from that (as you have above), you'll want to make it clear in the documentation.

Any attempt to prepare for Terbium compatibility at this point however is still mostly speculation. I'm really looking forward to the Terbium becoming a reality; but I think I personally should not have halted my own computer-building for the last few years while waiting for a processor that we thought would be available over a year ago. Now WDC finally has samples of the 65c51 and '21 available for evaluation, although they're still working through some known bugs (like that the '51 needs an external oscillator because the onboard one has a problem). We can drool, but at some point we may need to just go ahead with what's available now.


Changing the subject just slightly, I might try to anticipate what some might be thinking. I'm on another forum where a few members are working together to design and produce a high-quality, sensible engineering calculator/hand-held computer that's a modern version of what HP produced in their glory days of the early 1980's. They're not in it for the buck, and they have day jobs and can't work on this project full time; yet several people on the forum are getting impatient and, three months into the pregnancy, saying, "Where's the baby?" and throwing little insults. They compare it to business ventures that tried and failed, as if this calculator organization had to answer to investors or even to potential customers who don't have anything invested in it. It's gotten a bit ridiculous.

I have no idea if any readers here are thinking the same thing; but I can imagine that to some, the number of design approaches considered for the Kestrel, and the number of times the direction has been changed, may seem ridiculous. This is kind of part of the process though. It's not easy trying to make sure you don't paint yourself into a corner such that some future related thing cannot be plugged in because the designer(s) didn't anticipate all the possibilities and make the system as expandable as it should have been. Add to that the challenges of making it simple and inexpensive enough for hobbyists to make their own expansion boards. I have only a little to contribute in the area of bus design, but I'll encourage and defend the effort. Past discussions for an expandable computer like this did not result in anyone actually putting the time into such a design. So if someone (Samuel in this case) is interested in tackling it, terriffic.