Re: SI- Termination Comment

Fred (fred@contec.contec.COM)
Wed, 24 Apr 1996 16:58:26 +0800

RC termination for this kind of topology will not work. Anyway you should not have been surprised
to find substantually reduced voltage levels. 100pf and 100 ohm are way to big. I would have
tried something like 60 ohm and 5p. Anyway the classic way of solving this problem assuming it
can be solved without layer stackup changes would be in one of two ways. If you do not care about
what the waveform looks like at the source end then try a 35 or 40 ohm resistor in series placed
close to the source. This should slow down the driver a bit and greatly improve the signal at
the receiver ends. Alternately you can try a combination 330, 220 resistors tied to gnd and vcc
respectively at the last receiver. This should also dramatically improve the signal with some
degration on the high and low dc levels. If you don't like the dc levels you can play with the
resistors some. As for the ringing you might try a prototype test case to see if the ringing is
real or an artifact of simulation. You may want to ask your friendly ECAD apps engineer what
methods they are using for transmission line analysis. Using lumped LCR elements does not work
for long lines such as you have. An artifact of this type of simulation is ringing. The ringing
can be reduced in amplitude by increasing the number of lumped elements but depending on the
rise time is cannot always be totally eliminated. Sorry for adding to your headache but as these
rise/fall times decrease the headaches increase not necessarly in a linear fashion. By the way
in the case of the dual resistor combo if there is reflections above say 5.6 volts or below
-.6 v then you can also use diodes in parallel with the resistors. You will have to use caution
in your choice of diodes. Diodes with high capacitance values are very undesirable at this stage
in the game and can only complicate things. Anyway I hope this helps a bit.

Best Regards,

Fred Balistreri
Contec CAE