Count pulses with DAQmx ANSI C

[esci]

I have an externally-generated sampling signal and I need to count how many pulses occur between each edge of the sampling signal, where this count will very often be zero.  This is similar to photon-counting applications that have been posted on this board, but I need to accomplish it using the C API (not Labview).  I'm using an NI-6251 card and NI-DAQ 7.4.

From everything I've read I want to use Buffered Period Measurement (with Duplicate Count Prevention enabled), but I cannot get code to work for this.

For testing I have a 16 kHz pulse train wired to CTR0 GATE (PFI9) for my sampling signal and a 100 Hz pulse wired to CTR0 SRC (PFI8) to simulate my input that I need to count.

My code looks like this:

const int   SamplesPerSecond = 16130;

float64     TimeoutSeconds = 20.0;

float64     Pulses[SamplesPerSecond] = {0};

TaskHandle  CounterTaskHandle = 0;

DAQmxErrChk(DAQmxCreateTask("PulseCount_C", &CounterTaskHandle));
DAQmxErrChk(DAQmxCreateCIPeriodChan(CounterTaskHandle,      // Task Handle
                                    "Dev1/ctr0",            // Counter
                                    "",                     // Name to assign channel
                                    0.00001,                // Minimum expected value
                                    100,                    // Maximum expected value
                                    DAQmx_Val_Ticks,        // Units
                                    DAQmx_Val_Rising,       // Edge to trigger on
                                    DAQmx_Val_LowFreq1Ctr,  // Measurement method
                                    1,                      // Measurement time
                                    4,                      // Divisor
                                    "")                     // Custom scale name
);
DAQmxErrChk(DAQmxCfgImplicitTiming(CounterTaskHandle,
                                   DAQmx_Val_ContSamps,    // Acquire or generate samples until you stop the task.
                                   SamplesPerSecond)       // Samples per channel to aquire
);
DAQmxErrChk(DAQmxSetCIDupCountPrevent(CounterTaskHandle, "Dev1/ctr0", 1));
DAQmxErrChk(DAQmxStartTask(CounterTaskHandle));

// Loop over the next statement

DAQmxErrChk(DAQmxReadCounterF64(CounterTaskHandle,   // Task handle to read data from
                                SamplesPerSecond,    // The number of samples, per channel, to read
                                TimeoutSeconds,      // timeout, in seconds
                                Pulses,              // The array to read samples into
                                SamplesPerSecond,    // Array size, in samples
                                &readD,              // The actual number of samples read from each channel
                                0                    // reserved
));
 

The ReadCounter call is blocking for 1 second as if everything is working, but the Pulses array contains all zeros.

 (Oddly, if I remove the call to  DAQmxSetCIDupCountPrevent() aquisition still takes 1 second but the values I get back are all 4960, plus or minus one. I have no idea where this number comes from, and it doesn't change if I increase the frequency of my data signal.)

I can pretty much get things working with edge counting mode, but I don't really want a cumulative count; period measurement should give me exactly what I want without any post-processing (or so the docs say).

Can anyone explain what I'm doing wrong?

[Doug_L._Bear]

Your gate is moving way to quickly at 16 kHz to be able to count anything on your 100 Hz Source.  Check out the following Dev Zone Article.

http://zone.ni.com/devzone/cda/tut/p/id/2868

[esci]

Thanks for your reply Doug.

I understand that for a "typical" period-measurement application I would have the faster signal as the source and the slower as the gate (as pictured in the Dev Zone article you referenced) so that I get the period of the gate as a multiple of the period of the source.

My intention, however, is to hook up the signals in reverse.  So for the majority of the 16000 samples I will have zero counts, but 100 times a second I'll capture an edge and get a single count.

This post seems to imply that this should work:

http://forums.ni.com/ni/board/message?board.id=40&thread.id=1974&view=by_date_ascending&page=1#M205

In fact that thread describes pretty much exactly what I want to do, except all the discussed code is in LabView and I need to do it from C.

[esci]

Upon playing with this some more, I've discovered that the default sampling signal is not PFI8, but the 80MHz timebase.  Once I explicitly set the source to my 100 Hz signal with

DAQmxErrChk(DAQmxSetCICtrTimebaseSrc(CounterTaskHandle, "/Dev1/ctr0", "/Dev1/PFI8"));

Things are working as I expected.

Message Edited by esci on 03-05-2009 06:00 PM