I'm surprised that they've never had to deal with this issue before, though.

Hallo Craig,

schemes like the following



really make me angry! To control a micro only by a simple cable, without any buffer at micro side is just ridiculous. And I would never trust anyone who tells, that this will work or had worked! And believe me, the helpless try with 100Ohm resistor and 22pF to form some sort of characteristic impedance matching at SCK line proofs, that they indeed HAD trouble, otherwise they wouldn't have implemented this 'termination', right?

The magic thing with characteristic impedance matching is the word 'matching', means whatever you put at the ends of cable must match the characteristic impedance of cable. And you must do this with every line! And, of course, you should carefully focuse on what cable to use! Just simply connecting a 100 Ohm resistor in series with 22pF capacitor to some line without defining the sort of cable to use, without taking care of the other lines, without stating about the length of cable, etc. is just ridiculous!
The consequence is, that the one, who accidantly uses the same sort of cable like 'developer' will probably be successful, if at all, while someone else who uses a different kind of cable will very probably fail.

It was many times reported, that this AT89S8252 is very susceptible against smooth edged signals and against noise, especially on SCK line, and that a buffer at micro side is needed to guarantee slew rates to be within acceptable parameters and to prevent noise from being superimposed on signals.

It's a good idea to use twisted pairs for all signals, where each signal is paired by GND. And it's even a still better idea to keep the cable as short as possible...

Kai