memory

[marcelappb]

Hi,

I have PCI 6132. I need use it with 1MS/s and acquire 1 minute of datas, using 4 channels, and record this, until a condition of trigger.

I believe that I need use my computer's memory, because I think that the PCI 6132, using 4 channel, I can put just 4MS in my acquisition. If I could put 6MS of samples, this could be resolved.

So, I need use continuos acquisition, and record datas with the step about 1 micro seconds (1MS/s) and a window with 1 minute. And repeat this process for a long time, respecting a trigger level or pulses of GPS. And I don't need a pretrigger.

Sory my Engish. I'm Brazilian.

Thanks.

Marcela.

[Been bitten by LabVIEW]

So what is your question?

A

[marcelappb]

How can I use this board with my necessity?

what you recomend?

If I put this board in continuos acquisition, I will lose some datas?

For example, If I put a rate of 100 kS/s and a sample of 100kS, I will have a window with 1 s. But I need 1 minute. If I put this in a loop, I can acquire this without lose datas?

To supply my acquition, what you recomend? Some example that I can use?  It`s not so simple, because I need use more memory than I have in my board.

I used some examples about continuos acquisition, but i had problems with buffer and memory.

My datas have high frequency, and i need use trigger and 4 channels.

[Been bitten by LabVIEW]

Am I correct that your unit S is samples?  If I am understanding you correctly, you are saying that your board can store 1 seconds worth of data so you need to read it in and buffer it on the computer it is plugged in to?  I'm not familiar with the board you are refering to but sounds like you will need to poll/wait on some event trigger to then capture the data and then store it to some queue which can be read by another consumer process in parallel or after the minute has finished. 

Will you loose data if you loop?  Depends on the specs of the board and computer, but probably not if it is fast enough.  I've not used LV for a high throughput data stream like you are refering to, but it should be possible.

A

[DFGray]

I would recommend using the board in continuous acquisition mode and streaming data directly to disk.  You will want to use a producer/consumer design, with the acquisition in the producer loop and the file save in the consumer loop (search the LabVIEW help of these forums for more info on this).  If you stream to disk using either raw binary or the TDMS file format, you should have no trouble keeping up with four channels at 1MS/s, assuming you stream the raw binary and not the scaled data.  Your raw binary is I16 (actual digitizer is 14 bits), so two bytes per sample.  At 1MS/s on four channels, this is 8MB/s to disk, well within the range of most hardware.  If you stream scaled values, DBLs, you will be streaming four times as much data - eight bytes per sample vs. two.  This brings your data rate to 32MB/s.  This is possible on a modern system, but you have to do everything right.

So, short answer, you should not lose anything if you stream to disk correctly.  Use a producer/consumer architecture and either raw binary or TDMS as the file format.  Do not write as ASCII or you will probably fail.

Let us know if you need more help.