In BIOS you can set certain modes of operation for the parallel port, like "ECP", "EPP", etc. When testing ATMEL's ISP programmer cable, which is also driven by the parallel port I have noticed that rise time and signal level heights enormously depend on this set mode. Experiment with these modes, maybe you have only a signal compatibility problem somewhere.

Take care, the signals of parallel port are usually TTL! So, don't use 74HCMOS, but 74HCTMOS, as an example. Only 74HCTMOS will accept TTL levels, while 74HCMOS on the other hand needs CMOS levels. Confusing this can result in strange transmission errors. Many years ago I had an application, where simply touching the cable resulted in transmission errors.

Another issue is cable termination. Parallel port signal transmission is believed to be very slowly, so in many applications there's no proper cable termination. Well, to maintain TTL compatibility at the receiving end the well known standard cable termination with characteristic impedances at both ends must even not be installed, otherwise the output signal would be halved at the receiving end. But what should be used is series termination at the driver end like shown here:



Q5 and Q6 of 74HC541 are driving two lines of the port cable. You see 100Ohm series termination resistors inserted. The 100Ohm resistors at the input of 74LS540, on the other hand, are only for protection purposes, serving some current limitation and a little bit of low pass filtering. These resistors are no termination resistors, because then they had to be connected to signal ground! The LM393 doesn't need series termination because due its limited speed and output signal rise time reflections are hardly to expect.

Take care that all partners of transmission are using proper series termination, with only one characteristic impedance (resistor) at the driving end. Neglecting this can result in enormous cable reflections, ringing and cross talk.

Kai