Harold Hallikainen wrote:
> Sorry I neglected to mention that. Since it is the slew rate and not
> the period of the clock that is important in dealing with
> reflections, I did not feel it was important at the time. Of course,
> a high clock rate would disallow a slow slew rate.

And conversely, a slow clock rate like your 2MHz provides ample room for
limiting the slew rate.

With a 2MHz clock we can low pass filter at 20MHz without having to think
about it much.  Let's see what that comes out to.  You want to put enough
deliberate capacitance on the trace such that it dominates the unintentional
capacitance, which can then be ignored.  I'd say 47pF is about the minimum
for this.  With 47pF and 20MHz you get a resistance of 170 ohms.  The PIC
output drivers will have some impedence, and the trace will have some
unintentional capacitance.  Of course we left a generous margin by aiming at
20MHz low pass rolloff.  Still 120 to 150 Ohms in series with the PIC pin
and 47pF to ground right after it (not at the other end of the transmission
line) should work nicely.

Now do a sanity check in the time domain (again, it's useful to analyze
these things both ways, each gives you different insights).  170 Ohms and
47pF produces a time constant of 8nS and a 90% settling time of about 18nS.
That's only 7% of the 250nS between clock edges, assuming a symmetric clock.
I don't know how fast your PIC runs, but as long as you have 25nS between
any two edges of the SPI bus, everything should be fine (From the bus
reflections point of view.  Of course your chips have to require a setup
time shorter than that, but that's unlikely a problem).  In other words, you
can probably write the data to be clocked then the clock in successive
instructions and get away with it.

The time domain analisys also points out how things could be cut tighter if
needed, like if you really had a much faster clock.  In other words, the 10x
from clock frequency to LPF rolloff frequency is a lot more room than
needed.  Of course we picked it for that reason, and since that only comes
to 20MHz there is no need to try and cut it closer.  20MHz is probably still
a order of magnitude below where the nasty stuff happens, so we're good all
around.

So this setup makes sense both from a frequency domain and time domain
perspective.  Go do it.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist