Yes, I can see how that could change the direction at random.
At rest though, the parallel port supplies the HC245 STEP Input with 4.5 V and the DIR Input with 108.6 mV. When the motors are activated the parallel port supplies the HC245 DIR Input with 4.5V in the positive direction and 108.6mV in the negative direction. Therefore the parallel port can supply the HC245 with more than 3.3V when it is required. So I now think it is a 5V parallel port. Also, the random direction changes only occur on the Y channel, which ever way it is hooked up between Mach3 and the Y motor.
There must be something I don't as yet see at the moment and maybe the anomaly is in the computer hardware, although the Y direction changes did not occur when the CNC was attached to the Maxpath42 circuitry .... that just blew up ..... I'll see tomorrow when I hook the other CPU up.
I did not test the voltage on the HC245 outputs with the motors running ... I'll do that tomorrow as well before I connect the other CPU
I have read blogs that state that although stepper coil wiring colour codes are specified by the manufacturer, sometimes the coils are wired differently, which, if not detected, can cause random direction changes. I have interchanged the motors though, and the random direction change always occurs to whatever motor is connected to the Y channel.