well, i m no expert in this but why don't you try a timed read of the hardware instead of a transition wait

I think you could have much more effiscient result as there could still be falures even with a phi=pi/2

as I see it (and remember i'm no expert)

you have

sig 1
___ ___
___/ ___/
sig 2
___ ___
___/ ___

both transition occur at the same time

you could easily have multiple
sig 1 high
sig 2 low

when both sigs are in transition state (hope this is more than 2% english compatible)

if you time your method

ie time beetween reads = 4 * trans time you eliminate this problems, but reduce the use of your flowmeter to a known flow range

the state machine approch is in theory the good way to go, so this would have aparently no other inconvenients than reducing range

(btw if your fluid is a good light transmitter, and not totally homogeneous in temp or concentration ligh could take some real weird trajectories, not even speaking about light diffusion and chemical sensibility to some particular frequency, ie diamond and X ray glaring light green)


For my own knowlege is there system with (phi .ne. pi/2)
i believe a pi/3 would be better or are signals more like

sig 1
_ _
_____/ _____/ _____
sig2
_ _____ _____ _
_/ _/ _/

here phi=pi/2 is just perfect

regards,
Benjamin