other stuff I was thinking about:
- we should reserve an input or two for "manufacturing" jumpers
so that s/w can run some internal diagnostics and board test stuff
- save a few bits to read the SIMM ID bits, so that s/w can know
what size DRAM is installed in each bank and 'do the right thing'
- a name for the %^*%(* thing, how about 'Foghorn' (after Foghorn
Leghorn of course) If that is too Americentric, feel free to suggest
something else. Waldorf Salad? LOAM (Linux on a MIPS)?
- for serial/parallel ports on board, how about a 16C552 - two
16550 uarts and a IBM compatible centronics parallel port with 8 bit
cpu interface. Made by TI and others. Also don't forget those
pesky line interface chips.
- we should come up with a power budget to make sure we won't be
overtaxing a standard PC power supply.
- does someone have a heat chamber to test this thing at elevated
temperature for a while? It may be GPL h/w, but it doesn't have to
- think about cooling options for the CPU. The 33MHz 79R3051E in the MJ
package runs pretty hot - I'm sure the 40MHz 3081E will really cook,
especially with more cache on chip. I would plan on a heat sink and/or fan.
Don't burn your hands :-)
- for debug, put some headers around the CPU for each pin. If you only
run the local bus to the 3730 it will be hard to hook a logic analyzer to
either the 3730 or the 3081 without using expensive test clips.
- we should budget for appropriate filters and such for EMI. If waldorf
wants to sell this commercially it will have to pass VDE and/or FCC
(or CISPR) and/or other agency approvals. Does Waldorf have a
site where they can scan the unit? Or are they going to do their own
layout and only test their version?
I don't want to rain on anybody's parade, but if this is to become a
real product we should pay some attention to these real-world issues