Hi Ulchang,
It is next to impossible to achieve zero backlog with a count rate in the kHz range. The principle is that data is transferred from the board (counter registers and FIFO) to an intermediate PC buffer. LabVIEW or your programming environment cannot poll this PC buffer quick enough to achieve zero backlog. Zero backlog would imply you are reading 1 sample at a time quicker than samples are being supplied to the PC buffer by the hardware. Even if you put the samples to read at a time to 1 and your loop has no delays and is running as fast as possible you will probably still have backlog.
If you application reads a block of data from the PC buffer, whatever is left is backlog. Thus, even if on each loop iteration you specify that your read oper
ation read the last backlog, from the last iteration to this iteration more data would have been transferred to the PC buffer form the hardware.
One way to minimize the effect is by reading a very large block of data into your application program. This means that each time your program loops, you will wait until enough data is in the PC buffer before your read completes. If this happens on each loop interation you will in theory have zero backlog but you will be less efficient with your processor since it will be busy waiting for enough data to be present on the PC buffer.
The most efficient and dynamic system is if you read a block of data from the PC memory the first time and also read the backlog. On the next iteration simply read the number of samples equivalent to the last backlog and record the next backlog for your next read. Therefore in this scenario, if you run your loop as fast as it can go, you should be minimizing the scan backlog while making the most efficient
use of the processor since it never has to wait for data to be ready.
Anyway, hope that helps. Have a good day.
Ron
Applications Engineering
National Instruments
