| ??? 09/11/07 17:31 Modified: 09/11/07 17:34 Read: times |
#144387 - looking at the wrong characteristic Responding to: ???'s previous message |
Erik Malund said:
Richard Erlacher said:
I do recall, however, that we were able to get crystal oscillators to work just fine back in the days when we allowed 5-10 cm traces between the crystal and the active components. That was, indeed, true in the days when risetimes were significant. HOWEVER, I do not, in any way, see that working with modern derivatives. I[m not so sure that's the relevant parameter. I believe it might be the impedance, though. Today's CMOS has much higher impedance than the popular "standard" TTL of 1975. While the DC load is lower, the capacitance of CMOS is higher, hence, the rise/fall times would probably not be terribly different. Further, the technology of the time that DID have fast rise times, namely schottky TTL, didn't behave as well in oscillators, though it wasn't as subject to crosstalk either, thanks to its lower input impedance. A slow risetime is not affected greatly (perecentagewise) by stray capacitances, a fast one is. Actually, stray capacitances will increase the rise/fall time. Erik I think this was more a matter of learning curve than of technology. It didn't take outfits like DEC, Burroughs, CDC, or IBM long to figure out about ground planes, power supply bypass, and component placement. RE |
