! ...I didn't notice that, thanks a lot. I did some measurements on my 3 Qantek samples, and found the same 46%/54% duty cycle, the high level being the shortest one. The MX045HS-2C CTS oscillators I'm waiting for should be better (45 to 55% claimed symmetry).

So that asymmetry is costing about 10%, which means a 1.8MHz clock might be acting a bit like a 2MHz clock half the time. Doesn't sound too big a step to me - depends of course on how much margin there is in your design.

Well, I didn't really make any measurements but my first tests showed that this board successfully ran up to 2.5 MHz and began failing at 2.6 MHz. At that time, I used a not-so-good signal generator. Since then I bought an SDG1010 arbitrary waveform generator. It should be an opportunity to use it now, since I'll get a 50% duty cycle without any trouble. I'll do that on Wednesday (waiting for a long multiprecision computation to finish).

BTW : The parts are rated at 2 MHz (ACIA, CPU) and 4 (VIA), with the RAM and EPROM being 120-nS ones. I don't think that my rudimentary address decoding scheme is critical : 2 29-nS(max)/12-nS(typ) gate delays on the I/O select signal, 1 for ROM and RAM selects.

Wednesday news - quite satisfied ! here's why :

I set up the waveform generator to a 46/54% duty cycle clock (in order to harden the tests) and feeded my 65C02 with it. I inserted enough delays in my display routines to allow the HD44780 to catch on. Then, running a multiprecision computing routine with a cyclic status display, I began to increase the clock rate. Everything worked well up to 4 MHz ! (the CPU is only rated to 2) I went on increasing the frequency, at 5 MHz the program still ran but the 4-MHz VIA began showing an erratic behaviour making keyboard entry difficult. All that done with 250-nS EPROMS. The RAM tests I did after were all "pass". Not that bad...

I didn't think I had such margins, but I pulled from the basement the previous generator I used 2 years ago : unstable both from the frequency and phase point of view, so back to the basement and forget it)

I'm still waiting for the 2 MHz oscillators, but by now I wish I had a 4 MHz CPU and ACIA the next step, surely !

Use of genuine WDC parts will fulfill that wish, as they are officially rated to 14 MHz.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Right. I don't think the asymmetric clock is a significant performance factor in caculi's system, as he's not running it fast enough for timing to become critical. I just happened to notice it while perusing the data sheet.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Your thoughts are totally consistent with my tests not surprising since you've got much much more experience than me.

By the way, concerning WDC parts : it's precisely what I had in mind. As soon as I receive from Mouser the oscillators I ordered, I'll place a new order !

One more time thank you