[time-nuts] Transmission Line Question

[Magnus Danielson]

On 04/07/2011 12:55 AM, Don Otknow wrote:
> Well actually there's another level here which is that some of the
> lines are bidirectional data lines - each device at the end of the
> lines will occasionally drive these. However one of the devices (lets
> call it device A) can request a wait from the center device via a
> separate line so it should be possible to mediate issues for that
> device by just telling the center device to wait until the signals are
> stable.
>
> However, the other device (lets call it device B), with no wait
> request line, is probably not going to be as happy. But I think that
> (assuming that the data pins of the center device are high impedance
> while device B is driving the line and device B has a 50 ohm source
> impedance), then device B is going to see a 100 ohm transmission line
> in parallel with a 100 ohm termination, which should be fine, and at
> the other end of the transmission line (device A) a 100 ohm
> termination, which is also fine. Assuming the center device is high
> impedance, then there shouldn't be any weird reflections.

OK, now you have a bus.

This should work, as long as the transmission lines are allowed to ring 
out such that it doesn't cause interference.

Another solution is to have open-ended sides and then use Zener-diodes 
to clip out the reflections.

See PCI for instance.

Cheers,
Magnus