Can the NI PCI-6602 handle continuous counting of pulses at 22 MHz without missing any pulses?

[ash_alex]

Subject: Confirmation Needed before purchasing: Continuous Pulse Counting at 22 MHz with NI PCI-6602. 

Hi

I am currently working on a project where I need to count pulses from a Module, which outputs pulses (15nS) at a frequency of 22 MHz. I am considering using the NI PCI-6602 to perform this task and would like to confirm its capability to handle my requirements.

• Data Handling: I need to read and record the pulse count for each 100 µs interval continuously for a duration of 5 minutes.
• Data Transfer: The counter values need to be transferred to the PC efficiently without missing any pulses, ensuring that the transfer time does not delay the counting of subsequent intervals.

Can the NI PCI-6602 handle continuous counting of pulses at 22 MHz without missing any pulses?

Will the DMA transfer mechanism ensure efficient data transfer to the PC such that there is no delay affecting the next counting interval?

Are there any specific configurations or considerations I should be aware of to achieve this setup?

Thank you for your help!

[Kevin_Price]

22 MHz counting:  Spec'ed to count at 80 MHz.  If your "Module" puts out a clean pulse signal, you should be ok.

15 nanosec pulses:  The 80 MHz counting spec would correspond to a 12.5 nanosec period, so it must be able to handle 15.

DMA: the 6602 has its own (unfortunately very tiny) onboard FIFO.  As it latches count values, they get pushed off into this FIFO and DMA delivers them to PC memory from there.  Pulses do not get missed due to this data transfer.

However, the long-term bandwidth of 6602 counting tasks is a bit more fragile due to:

- the very small 2-sample hardware FIFO

- the PCI bus, which to my understanding must negotiate access and bandwidth among other PCI devices.  Most specifically, a PCI video card could cause trouble (though I don't think I've seen one of those in quite a long time).

100 µs sampling intervals mean a 10 kHz sample rate.  My recollection from very long ago, with much older PC hardware, was that things got more questionable in the low 100's of kHz sustained sample rate for counter tasks.  So I'd be cautiously pretty optimistic for your case.

Having said all that, I think you also ought to consider the much newer PCIe-6612, which is much more reliable for long-term acquisition due to a much bigger FIFO and dedicated PCIe bus access.  The main sticking point is that it wants a 20 ns min pulsewidth and your "Module" delivers 15.  If your Module can't be adjusted, I'm sure there are many simple inexpensive chips that could turn your 15 ns pulses into 20+.

    It's worth some effort to avoid making new DAQ hardware purchases based on the PCI bus these days.

-Kevin P