The parts are PECL, 300 ps Tr and Tf, 20mA drive, 100 ohms differential
impedance. In order to do this I need to solve two things:
1) Controlling Zo through the interconnect
2) Minimizing conductor and dielectric losses through the interconnect.
Regarding item 1), it appears from connector manufacturers that running
differential pairs at these speeds through connectors can be done with less
than 5% crosstalk and Zo mismatch. (Example, 5row , 2mm connectors)
Regarding item 2), it appears that running 50 ohm stripline traces using 15-20
mil wide lines in FR-4 may be acceptable for line lengths under 8-10", but for
lengths greater a more exotic material may be needed (like GTek). Gtek has a
tan delta of .01 compared the FR-4 tan delta of .02.
Does anyone have any experience with either item 1 or item 2 above?