Whats important in this case is not necessarily controlled impedance.
You need to pay attention to three things: your edge rate, delay through
the connector and your gnd. You want to make sure that the delay through
the connector is less than 1/3 that of your edge rate( to avoide reflection);
you must include the vias/pads of the connector's escape via in your model
since this will add to your delay. Also ECL normally references to gnd and therefore
you also need a gnd in the area. However, since you run differencially your gnd should
not see too much current(common mode) through it. In short, its not necessary to
have controlled inpedance if you follow the rules above and make sure both
differencial lines goes through similar delay environmet in the connector.
From HANS_MELLBERG@non-hp-santaclara-om4.om.hp.com Thu Mar 20 12:20:02 1997
Date: Thu, 20 Mar 97 09:13:28 -0800
Subject: Re: Controlled Impedance Connectors
AMP is already working with ECL designers for Z-controlled connectors.
______________________________ Reply Separator _________________________________
Subject: Controlled Impedance Connectors
Author: Non-HP-colin.campbell (email@example.com) at HP-Boise,mimegw2
Date: 3/20/97 12:15 AM
The current design that I am involved with requires the transfer of ECL clocks
and data ( probably differential ) at rates > 50MHz between two adjacent PCB's.
I therefore require to find a suitable controlled impedance connection system.
The problem is that the connectors must be surface mount and operate over the
military temperature range.
Can anyone help?
GEC Marconi Avionics