[Gb-ccb] Caltech Backend Telecon Monday 07 July 4pm EDT

[Martin Shepherd]

On Fri, 4 Jul 2003, John Ford wrote:
> Richard Lacasse writes:
> > Martin Shepherd wrote:
> > > ... assuming that by opto-isolation, we are talking about optic fibres,
> > > rather than opto-isolator chips and wires,
> > > ...
> > All I had in mind were wires and opto-isolator chips.  A differential
> > driver would drive a shielded twisted pair and at the far end there
> > would be a termination network and opto-isolator chip.  No current
> > has to flow through the grounds or power supplies of the end with
> > the opto-isolator chip, so shield currents are minimal.
>
> This is also what I had in mind.
>
> The optical fiber approach would also work, but would probably cost
> more.  Maybe not too bad if we used the plastic AMP fiber system.  The
> waveguide beyond cutoff ports in the RFI boxes are the way to go in
> that case for shielding the penetrations.
>
> I don't think we can isolate the 2 chassis, so it's probably not worth
> the extra cost and trouble to use optical fiber for digital control
> signals, rather opto-isolators are probably the way to go.

Understood. However this leaves me puzzled.  Now that it has been
all-but-decided that the shields of the analog signal cables are going
to connect the two chassis, I don't see any reason to opto-isolate the
digital signaling cables at all, unless optic fibers are used. Surely
LVDS cables would generate less RFI than unbalanced cables driven by
opto-isolators, and the existing plan calls for using LVDS, not only
to reduce RFI, but also because the tri-state nature of LVDS will give
us the ability to remotely relinquish control of the receiver to
another backend.

Martin