[Gb-ccb] Information and advice on optic-fibre ethernet

[John Ford]

Martin Shepherd writes:
 > 
 > 
 > On Wed, 13 Aug 2003, John Ford wrote:
 > >...
 > >  > Note that I plan to use metal fiber-optic bulkhead couplers to
 > >  > pass the TX and RX fibers through the wall of the CCB chassis.
 > >
 > > For best RFI shielding, you should consider using a waveguide beyond
 > > cutoff farrangement to get the fibers into the system.
 > 
 > I have been searching the web for any mention of explicitly
 > RFI/shielded couplers, but not found anything. Is the solution that
 > you are suggesting available in an off-the-shelf item, or is it
 > custom? Could you give me the appropriate part numbers please?

It's a custom thing.  A metal tube 7 times as long as it is wide, with
no metallic cables running through it.  

 > 
 > >  If you are
 > > sure your metal connectors will work, then that would be fine, too.
 > 
 > No. I don't have any idea how well they would work. They were just the
 > best off-the-shelf solution that I could find on the web.
 > 
 > > On the subject of RFi shielding, have you a block diagram of the
 > > system and the interconnections between assemblies that we could look
 > > at to pick out any RFI problems?
 > 
 > I'm working on documentation of the external interfaces at the moment,
 > so this is still a work in progress (hence the question about
 > ethernet). The following is a list of the external connections:
 > 
 >  Ethernet input/output:
 >    2 x optic fibre, using some kind of bulkhead feedthrough connector.
 > 
 >  1-PPS signal input:
 >    I am guessing that you use 50-Ohm RG58 coax to distribute the
 >    1PPS signal around the focus cabin, and have thus tentatively
 >    decided on bayonet BNC connectors.

That's a good guess!

 > 
 >  Opto-isolated digital phase-switch and cal-diode control signals outputs:
 >    I am planning on using off-the-shelf IEEE-1284 parallel-port cables
 >    for this. The IEEE-1284 standard mandates overall foil and braid
 >    shields, connected around 360 degrees to the shielded male DB25
 >    plugs at each end. These shields enclose 8 independent unshielded
 >    twisted pairs. There are also 9 other twisted pairs, but these
 >    share ground-return wires with the 8 independent ones, so I don't
 >    plan to use them. In fact I only need 4 of the 8 that can be used,
 >    so there will be a comfortable margin of 4 extra independent pairs
 >    available for potential future expansion. The standard mandates
 >    that there be no more than 10% crosstalk between neighboring pairs.
 >    It doesn't say whether this is near-end or far-end crosstalk, but
 >    the fact that these cables are used to exchange TTL signals between
 >    computers and peripherals at up to 2MHz suggests that this level of
 >    crosstalk won't be a problem.

Should be fine.  Use Spectrum Control db-25 connectors with filters
suitable for the edge rates of the signals.

 > 
 >    Alternatively I could use 4 individual coax cables for the 4
 >    control signals, as Tim has suggested, but I would prefer to have
 >    as few cables as possible.

I don't really have an opinion on this.  One reason that this might be
nice is the ability to pull off one cable for tests without disrupting
the other 3 control signals.

 > 
 >  DC power cables:
 >    I haven't got to looking for suitable connectors and cables for
 >    this yet. I understand that at in the focus cabin you have +/-15V
 >    and 5V power supplies that can be used by any instrument. What
 >    types of cables and connectors do you usually use when connecting
 >    these to other instruments? Also, how much current will we be
 >    allowed to draw from these supplies?

Roger should be able to answer this question for you.

 > 
 > By the way. I am spending huge amounts of time searching the web for
 > individual electronic components, connectors and cables, plus
 > associated data-sheets. Do you have any suggestions for a few
 > convenient one-stop shops, so that I don't have to waste so much time?
 > In particular, Newark's online catalog is horrendously inefficient for
 > finding things (and Newark still hasn't sent me the paper catalog that
 > I requested months ago).

We buy from many places.  I'm not sure I know of a Web site that's got
it all.  I like Newark, Digikey, Allied, and Mouser.  I like paper
catalogs best, though.

I suggest you use Spectrum Control for RFI filters and filtered connectors
AMP for regular connectors.

Anyone else have suggestions?

John