An NI 9401 was being used to log 0 to 5 volt frequency pulses from a flow meter that produces a square wave signal by means of an open collector output and a 2.2k resistor connected to a 5 volt source. Unsteady results when the controller of the flow meter was indicating a steady flow resulted in the input to the NI 9401 being viewed using a CRO. This showed a good quality signal present by also some intermittent impulsive spikes. The cable from the flow meter was around 50metres long. A simple capacitive filter at the input to the NI 9401 was tried starting with a 10nF capacitor connected to ground. The result was a much increased indicated flow rate by the output of the LabView application. However the view on the CRO showed all but no difference to the waveform with the impulsive spikes much less evident. A much larger capacitor was then tried with a result that the rise time of the square wave was now clearly rounded off when viewed on the CRO and yet the output frequency indictated on the LabView application increased almost by an order of magnitude.
It was felt that a hardware solution would be to add a filter followed by a Schmitt trigger ( in fact a comparator with feedback) so that the external noise spikes could be rejected but the rise time of the square wave signal into the NI9401 could be kept as low as possible ( i.e. 100ns or so) This arrangement worked well and the flow rates as indictated by the flow meter controller and by the LabView application were then the same.
Looking at the installation documentation on the NI 9401 (page 12) showed an example of a digital input from a switch to ground and a pull up resistor and so it is not understood why the pulse rise time into the NI 9401 was critical in any way ?
Not having access to the LabView application an external hardware interface of a filter followed by a Schmitt trigger provided a solution but is it possible that the software setup of the NI 9401 could provide a solution so making this external hardware interface redundant ?