In general is there a IC technology that contributes less to jitter than others? Or does nobody really know?
Yes, there are some factors that contribute, including:
1) dielectric absorption
2) losses (bandwidth limitations)
3) impedance mismatch or discontinuity
4) conductor skin-effect
5) conductor molecular structure
This means that the optimum cable will be exactly 75 ohms impedance, use perfect crystal thin conductors and air or vacuum dielectric.
This of course is very difficult to build, so you can get very close with expanded Teflon, perfect crystal silver or gold etc.. This is what I use in my Bitmeister. Very difficult to assemble as well.
Also what about propogation delay in IC's? Seems to be a big variance in different technologies. But does a 3 nanosecond delay versus a 100 nanosecond delay matter with I2S signals?
Propagation delay for cables is important only if it is not uniform across frequency. This occurs due to items 1-5 above. If these are addressed, its not a problem.
If you are talking about integrated circuits, PD is very important. It is directly related to slew-rate. Slower slew-rates mean uncertainty for the receiver to detect the transition. Faster is better, but you must take care that transmission-lines are addressed and that power di/dt does not cause ground-bounce.
I am looking to level shift a I2S signal from 3.3 volts to 5 volts. Lots of choices out there. Any suggestions?
Fairchild tiny logic "Z" series.
Steve N.