Hey YK,
Have you considered using one of the data lines on the other ports (C & D, or 2 & 3) to generate your pulse/trigger when your N samples have been acquired? This would be much faster than having software wait for an event and then generate your pulse. If you use the onboard clock, or even if you use an external clock, you could generate 100,000 samples, where all of them would be 0's and then on the last sample a 1. This would in a sense create your trigger signal when it is done acquiring the 100,000 samples. The generation and acquisition would both work off of the same clock source. If you want it to have a little delay after the 100,000th sample, you could acquire on the rising edge of the clock, and then generate on the falling edge of the clock. Or you could generate 100,001 samples, which would allow you to trigger a sample after the acquisition has completed, but if your acquisition clock is only 100,000 pulses, then this would not work. So I would recommend generating on the last 100,000th sample.
For your question about how the 6534 transfers data to the PC, the board does DMA the data to system memory, not to your hard drive. You have to save the data to disk if you want it to go to your hard drive. Also, there is onboard memory on the 6534 that the data goes into first, and then it gets sent across the PCI bus, but that is pretty fast, so it is negligible to how long it is taking the data to be acquired. One thing you could do to "time" your acquisition is to use my suggestion above with the generating 100,000 samples along with your acquiring 100,000 samples, but this time you could output a 1 as the first sample and the last sample on a data line, and then with a scope measure the time between the rising edges, but you would have to add 1 samples worth of time to that measurement. Or if you are doing a continuous acquisition, you could also do a continuous generation, only have a 1 on the last sample of your 100,000 generated samples, and then measure the time between the rising edges, as this would give you the exact time between each 100,000 samples acquired. If you are doing continuous acquisition, and need a continuous generation example, a good one to check out would be this example that uses the onboard memory so you don't have to stream data from the PC to the device while it is generating, which will also save PCI bus bandwidth.
I hope this helps. Please let me know if you have questions on this suggestion. Thanks, and have a great day.
Regards,
DJ L.
