[Gb-ccb] Filtering connectors on the PCB

[Martin Shepherd]

Brian asked that I send multiple questions at a time, so here is a set
of 3 (albeit related) CCB questions.

1. Regarding the control and signal lines connecting the CCB and RX
   boxes, my original plan was to mount bulkhead connectors through
   the walls of the CCB case, and use shielded patch cables on either
   side of these connectors, connecting to the RX box on one side, and
   the CCB PCB on the other side. Using patch cables between the case
   and the PCB, rather than soldering wires to case-mounted sockets,
   would make it easy to remove the PCB for maintenance, and also
   reduce RFI problems inside the CCB box. Unfortunately, not only
   haven't I been able to find any suitable filtered bulkhead
   connectors, but I haven't managed to find any type of filtered
   twisted-pair connectors, bulkhead or not. For this reason I am now
   considering a slightly different approach.

   My idea is to use shielded, but unfiltered bulkhead feedthrough
   connectors, going through the walls of the CCB box, have these
   connect to PCB mounted sockets on the CCB PCB via shielded cables
   and connectors, and have RFI filter capacitors soldered to where
   the pins of the connectors are soldered to the PCB. Given the use
   of shielded connectors and cables, the only point at which RFI
   could get onto or off the cable would be via the pins of the
   PCB-mounted connectors.  However, provided that these pins were
   very short, and that the filter capacitors were either soldered
   directly to them, somehow embedded in the board as feedthrough
   capacitors, or connected via matched stripline or micro-stripline
   PCB tracks, I imagine that this wouldn't be significant. Would this
   be an acceptable solution?

2. My second question is whether anybody has any experience with
   Sabritec's line of ultra-miniature concentric triaxial
   connectors/cables (or can suggest something better)?

    http://www.sabritec.com/technotes/triaxconnectors.htm

   Given the difficulty of finding an off-the-shelf RFI filtering
   solution for the high density connectors that I was looking at
   previously for the differential signal lines, the difficulty of
   building custom filters for high-density connectors, and worries
   about crosstalk within a high-density connector, I am now back to
   the idea of using triax/twinax shielded twisted pair cables. The
   problem with this is that the standard triax/twinax connectors are
   very bulky, and I don't want to have to devote a large area of the
   main CCB PCB to 16 of them. Sabritec's ultra-miniature triax
   connectors are much smaller, so I am considering using them
   instead.

3. My final question is whether push-on connectors (like SMB
   connectors) must be avoided. If so, would this still be the case if
   a number of push-on plugs were attached to a screw-down plate? I am
   wondering about having the plugs at each end of the 16 triax cables
   that carry the detected signals, all go through a single screw-down
   plate. This would reduce the amount of space required to plug in
   the connectors, make connection and disconnection easier, and make
   it almost impossible to accidentally cross-wire the 16 cables. I
   probably wouldn't actually afix the plate to the connectors. The
   plate would probably just be threaded over the cables, and press
   down on the connectors from behind. This would allow individual
   cables to be intentionally cross-wired while the plate wasn't
   screwed down, and thus facilitate debugging.

Martin