... among other things. It compensates for slowly changing inputs as one might encounter when interfacing digital hardware to analog signals, and has considerable hysteresis (like your furnace thermostat) so it doesn't oscillate and is highly immune to random noise.

As I suggested, if he takes those 12-volt inputs from the relays and divides them through, say, a series pair of 1300 and 820 ohms, he'll have, nominally, his 2K-ohm resistance from 12 volts to GND, yet his Schmidt-trigger will see a transition from about 4.8 volts to GND, and if he puts a 1 nF cap across the 820 ohm resistor, the risetime will be slow and relatively bounce free. The Schmidt-trigger will swallow (mask) minor (100 mV) noise or ripple on the input, since its threshold for positive-going signal is considerably higher than its negative-going threshold, in fact, nearly a volt. The threshold doesn't move around much, as it's ratiometric (in the case of the 40106) and tied to the supply.

The original description says he doesn't care how much the transition is delayed/shifted in time. I'd guess it will work fine, and deosn't need an optoisolator at all.

RE