29 Nov
2009
29 Nov
'09
5:13 p.m.
dwight elvey wrote:
.. although there is the speed or power tradeoff to be made with resistive pullups.
A number of early Japanese calcs were made from a logic combination of SSI IC
NOR gates and discrete-diode AND gates to make AND-NOR logic. Quite flexible in
that the AND gates could have as many inputs as needed, canonical state
machines were easy. The calcs are packed with hundreds of diodes and a few ICs.
Some other early IC calcs used only half-a-dozen-or-so IC types; 2,3,4-in gates
and dual-FF for example.
The HP2116 machines used only around a dozen IC types.
Could be interesting to look at the first IC-based PDP-8 for how many IC types
were used.
The '12 is an OC 3-input NAND. I like 3-input gates better because
they lower the part count in circuits like J-K flip flops. If you
admit diodes to the mix, there's no reason that 38s or even 06s
wouldn't work just as well.
RTL and DTL were great when it came to wired-ORing. That 903 wasn't
a bad choice for the AGC.
--Chuck
Hi Chuck
Good point diodes and o6s makes the ultimate in
low tec computing. Diodes can be had for less than a
penny each. One resistor and a pile of diodes makes
a reasonable gate. All that is needed is an inverting
and buffering function. The 06 fits the bill.