Re: LVDS over non-twisted flat ribbon cable

On Wed, 16 Jul 2008 22:49:45 +0200, Rene Tschaggelar <none@xxxxxxxx>
wrote:

John Larkin wrote:
On Wed, 16 Jul 2008 10:47:30 +0200, Rene Tschaggelar <none@xxxxxxxx>
wrote:


I'm lacking the experience of LVDS over flat ribbon
cable that should be twisted, but is not. Different
pairs will couple with the next and with the outside
depending on frequency and length, but by how much ?
The motivation for the question comes from the rather
limited number of possible crimping locations a
twisted ribbon cable provides. Whereas a flat ribbon
cable can be crimped at any arbitrary millimeter.

I'd be using it for short runs, below 0.5m alltogether

Rene


What's your data rate, and what parts will you be using?

That length will be fine for most things short of the ultra-speed
stuff. It's recommend 3 or 4 wires per LVDS pair, with a ground
between a pair and anything else

gnd a+ a- gnd b+ b- gnd ...


The really fast stuff might get nasty over a half meter of standard
ribbon cable.

John

Thanks John,
It'll be 20MHz to 60MHz or so. The parts will be
65LVDS1 @ 65LVDS2. They'll synchronize different
subsystems in a distributed setup.

Rene


Not super fast, so you should be OK. Be careful to keep the pair skew
down by routing equal length traces in/out of the ribbon cable
connectors. But it's not really hairy... 100 ps or so shouldn't break
anything.

Anybody know the differential impedance for a regular Scotchflex type
cable in this config, diff pairs with outside grounds? I'm too lazy to
TDR it.

John



.

Relevant Pages

  • Re: LVDS over non-twisted flat ribbon cable
    ... John Larkin wrote: ... depending on frequency and length, ... It's recommend 3 or 4 wires per LVDS pair, ... Not super fast, ...
    (sci.electronics.design)
  • Re: LVDS over non-twisted flat ribbon cable
    ... depending on frequency and length, ... I'd be using it for short runs, below 0.5m alltogether ... What's your data rate, and what parts will you be using? ... It's recommend 3 or 4 wires per LVDS pair, ...
    (sci.electronics.design)