A colleague of mine made the following statement regarding when a line should be treated as a transmission line:
"When the two way flight time (i.e., twice the one-way propagation delay Td) is greater than or equal to the signal rise-time."
In other words, he means whenever the rise/fall time is less than the round trip delay of the net. Can you comment on this? I told him I would consider the one way propagation delay rather than the round trip.
Thanks for your interest in High-Speed Digital Design.
Actually the situation is even worse than either of you have postulated.
The critical line length beyond which many people use terminators varies from about 1/10 to 1/3 the length of the rising edge. That means a one-way line delay of between 1/10 and 1/3 of the signal risetime. The termination problem is influenced by the driver impedance, load impedance, and the degree of perfection which you require of your signals (that is, can you wait for the ringing to dissipate or must the first incident edge be perfect?).
One particularly bad case for ringing involves a powerful, low-impedance source coupled through a short line to a large capacitive load. With such a setup, terminations can easily be required on lines as short as 1/10 of the risetime (or even shorter). If the line delay is as enormous as either of you suggest, equal to even half of the risetime, severe overshoot and ringing can easily occur, even with ordinary logic, on un-terminated nets.
As a rule of thumb, I generally simulate all lines longer than about 1/6 of the risetime.
Dr. Howard Johnson