Solution to previous problem with shift registers

[David.Nicholson]

We had a problem with connecting a made up board containing shift registers
with a PC and driving the board via a parallel cable and the parallel port.
(the shift registers were 74C165)

In clocking the data from the shift registers(raise clock, lower , read the
data bit) had a problem that we were reading "random" bits from the data
bit, when we knew that the 4040 counter we were reading via parallel load
into the shift registers had a different value in it.

The problem turned out to be that the clock line was ringing(rather like a
button might bounce) and we were getting multiple clocks from just one clock
from the PC port, and consequently missing data bit transitions. They
weren't so much random as the correct data bits only a number of them at
once!!

We found this out by reading The Art of Electronics by Horowitz and
Hill(fantastic book that), who have a couple of pages describing our exact
problem and also cautioning us doing what we found worked, which was to put
a 300 pf cap across the clock bit to ground.

Their advice was to bias a Schmidt trigger with resistors or to put a 100pf
cap in series with 100 ohm from the clock gate to ground. The schmidt
trigger bias approach was

--------5v rail --------------------------
|
R 180
|
------------0--------- R 180 ------0------o schmidt trigger( the output of
this goes to the clock on the shift register)
| |
| |
| 330pF
R 390 |
| |
| |
------------0-- ground-------------0----------------------- and this works a
treat

Thanks for those who posted asking about whether we had let the 4040 ripples
settle down first(yes)
Needless to say, we have also used the same fix on ALL our outputs(from PC),
as one of them also gates an oscillator and we may have had the same problem
with other parts of the circuit without realising it

cheers, David and Ben