I'm using a 2 shift registers in series, connected via 5 feet of automotive lighting cables to an Industruino, and am getting weird results when I test. When I turn all the outputs off, the last pin normally stays on.
When I mocked it all up on a breadboard it worked fine.
Is it the cable I've used, the distance, or something else?
EDIT: I have a 13 way cable going between the shift registers and the Industruino. This carries the 3 pins for the shift register, 12V, 5V, and 2 earths. Is it the 12V which is distorting the data?
Some thoughts: Are there .01µF caps. at each IC? A larger 'lytic cap. for the remote board? Do you still have Vcc in spec for the remote mounted registers?
Just having 12V on a wire isn't a problem. Switching current on that line though, will naturally induce some voltage in the parallel wiring. Over a five foot run it could well be enough to cause spurious triggering.
You need differential line drivers and line receiver chips at each end. And as dougp pointed out you need proper decoupling on the shift register chips.
Other things to try is slowing down the clock signal.