The problem I described with the long cable was in fact a problem with my laptop PC (Everex, 266 MHz Pentium), present also for the short cable I first used, but with the longer cable the margin became negative. The other PC:s I have at hand behave normally.

It turns out that in this PC, the input I use, the "Paper Out" line of the printer port, is shunted by a network consisting of a diode in series with a 10 uF capacitor. (And probably some more things, this found just by measuring from the outside.)

To give an idea of sizes, the capacitance of the 70 meter cable is 10 nF, i.e. 1000 times smaller.

The reason for this weird circuit of the laptop is probably that the printer port doubles as a connection for an external floppy drive.

For a positive pulse to get through, the signal source has to charge this capacitor to 5 volts, which takes some time, depenting on how much current you drive. No problem for a printer, as presumably the paper won't get in/out in a few milliseconds, and can allow the maybe 0.1 second delay for the paper out alarm. The sensor output current is limited to 30 mA, and the interface has series resistors in both ends of the cable for some degree of lightning protection. This combination was dimensioned to drive the cable capacitance of the long cable, but for something equivalent of 70 kilometers of cable - not! The delay became around 8 ms, while the pulses from F2A models are about 4 ms long.

After I made the long cable I tested the connection at home just by moving my hand in front of the sensor. In spite of being a T/R pitman my fingers are a bit slow, so I got the expected indication and assumed it worked.

Now I made a test with smaller series resistors, and then I got it to work with pulses mimicing the ones from F2A models. Not a favored solution, though - sort of shock treatment for the capacitor and diode.

As there might be other laptops using the same solution, I'll look at using another signal line of the printer port. The attractive thing with using the "Paper Out" signal is that the sensor could coexist with a printer with no other disturbance than that the paper out indication won't be seen by the PC.

Göran Olsson