>>I am using a 16C56 part to measure pulse widths
>>(in the 0 to 10 ms range). My problem is that I
>>am getting unreliable input pin detection within
>>software. I am driving the pin with a test 50 percent
>>duty square wave, and the measued pulse is correct most
>>of the time, but on random occasions the software indicates
>>that the pulse is slightly longer than it really is.
>>
>>I am driving the input pin via an 4N25 optocoupler, with
>>the input pin connected between the collector of the phototransistor
>>(with the emitter grounded) and a 4.7K pullup resistor.
>>I believe the software is working correctly, and the problem is
>>With the hardware implementation.
>>
>>Do I need any special filtering on input ports to reduce
>>false triggering? Am is missing something??
>>
>>- Bruce
>>
>>

I just checked the 4N25 data sheets, (had to go to the office anyway).
The motorola data sheets show that with RL=1K the turn off time has a
minimum of ~10us around 3ma led current. but rises either side of that
wheras with RL=100R the minumum off time is 4-5us at a led current of
7 ma or so. And only slightly faster with RL=10R.

So I suggest that you first check the led current is around 7 ma and
then change the pull up to 100R.


Ray Gardiner, Shepparton, Victoria, Australia       ray@dsp-systems.com
Technical Director DSP Systems               http://www.dsp-systems.com
Private e-mail to:-  ray@netspace.net.au