Period Measurement Issue

[kb3kj]

I am using one of the counters in a PXI 6221 board to measure the period of a pulse train to implement a tachometer. Everything works fine when connected to a signal generator on the bench but when used with a more noisy signal with occasional glitches it will sometimes stop generating new data. 

The initalization code is:

DAQmxCreateTask("FInputs",&FreqHandle);
 /* add a frequency channel to read RPM */
err = DAQmxCreateCIPeriodChan (FreqHandle,FIRST_6221"/ctr0" ,"", 0.0001, 0.1, DAQmx_Val_Seconds, DAQmx_Val_Rising,  DAQmx_Val_LowFreq1Ctr, 0.000001, 4, "");
DAQmxCfgImplicitTiming(FreqHandle,DAQmx_Val_ContSamps,0);
DAQmxSetBufInputBufSize(FreqHandle, 1000 );

DAQmxStartTask(FreqHandle);

Then the following is called every mSec when a reading is needed:

err = DAQmxGetReadAvailSampPerChan(FreqHandle, &num_avail);   
if (err != 0)
        {
        DAQmxGetErrorString(err, err_msg, 100);
        printf("From GetRead... %s \n", err_msg);
        }
        
if (num_avail == 0)
        {
        /* no new periods measured, dont update speed so previous data used again
         * clear bit in status
         */      
        status &= ~STATUS_TACH_PULSE_RECVD;
        no_pulse_cnt++;
            
        if (no_pulse_cnt > 1000)
            {
            no_pulse_cnt = 0;
            DAQmxStopTask(FreqHandle);
            DAQmxStartTask(FreqHandle);
            }
        }
else
        {
        err = DAQmxReadCounterF64(FreqHandle, DAQmx_Val_Auto, 0.0 , data_array,1000,&num_avail , NULL);
        
        if (err == 0)
             {
             /* sum all availible periods */
            for(i = 0; i < num_avail; i++)
                 {
                 Period += data_array[i];
                 }
               
            if (Period > 0.0)
                 {
                 /* calculate RPM from average measured period between pulses from sensor */
                 *speed = PERIOD_SCALE*num_avail/Period;
                 }
             /* set flag to indicate pulse(s) received */
            status |= STATUS_TACH_PULSE_RECVD;
            }
       else
            {
            /* error condition, use max speed */
            *speed = 9999;
            DAQmxGetErrorString(err, err_msg, 100);
            printf("From ReadCounter...  %s \n", err_msg);
            }

     }

When the problem shows up, num_avail from DAQmxGetReadAvailSampPerChan will remain at 0  even though

there are pulses on the input.  No errors are detected.

I have implemented a timeout (based on the no_pulse_cnt variable) that Stops and Starts the task when this

happens. This gets things working again but I am looking for a cleaner solution that prevents the issue or allows me to detect it immediately.

Looking in the Archives I have seen some comments about issues which can happen when there are no

timebase pulses between input edges.  Could this behavior be triggered by that?

Thanks,

Jim

[P.Sanders]

Does this problem occur using one of the DAQmx text based examples?

-Ciao 😉

[kb3kj]

I have looked at the examples, but none seem to do exactly what I am looking for as far as providing the latest period data on demand.   I am doing this under the real time OS so I cant simply run the examples anyway.  

In the meantime we have improved our external signal conditioning circuitry and the problem has effectively disappeared.   I would still like to improve the robustness if practical though.

Thanks,

Jim