Mike,

You have, I believe, some serious misconceptions regarding the relevant technology.

The old 4000-series CMOS is considerably slower than the HC/HCT logic. The HC/HCT logic is considerably slower than the AC/ACT logic that's also available. All this stuff is slower than the currently popular field programmable logic in the form of 5-volt-compatible CPLD's. Lower voltage programmable logic, while not relevant, is faster yet.

However, you're trying to do something with a 1980's device that operates at, essentially, 1 MHz when fed from a 12 MHz clock. Lightning speed is unnecessary.

You've left a couple of issues open. First of all, what role does the 18.432 MHz oscillator or crystal have in your system? You can't use a crystal of that frequency in a crystal oscillator configuration on the HMOS MCU, and, in order to get to a suitable baud rate for standard asynchronous communication, you'll have to use T2 in your MCU just to generate common baud rates with that speed even if your part does work at that rate, which it shouldn't. Ask Erik about how that affects things.

Now, remember there's a minumum period which you must never exceed in the oscillator timing. It's not a fully static part, though the CMOS versions are.

Remember, too, that the timing of your CMOS/HCMOS/ACMOS logic depends quite a bit on the load, particularly the load capacitance. That's specified, so you must read the datasheets for your various components. At rates near 1 MHz, you can probably ignore the propagation delays of your signal paths, and the associated capacitance. However, if you choose to use CMOS address latch and, perhaps, data buffer, you must provide pullups, say, 10 K-ohms, but perhaps as much as 390 K-ohms, to ensure that a one on P0 is recognized properly by your logic. If you use TTL, you'll have no such concerns, since you're then driving, essentially, an emitter, or a diode cluster that's already pulled up. A few LS devices have PNP inputs, which rely on the emitter bias to set the input (base) voltage, which pulls up the inputs. CMOS doesn't do that.

Do you have an oscilloscope available? If not, you're going to play HELL figuring out whether your circuit is working as you wish.

Have you, at least, simulated the logic?

RE