LabVIEW Express 7 and PXI-5122

[enekofg]

Hello.

I'm involved in a project which aim is digitizing the signals received from several(2 at the beginning) antennas. We pass these signals to an intermediate frequency of about 15 MHz.

Now we want to use the NI-5122 to DIGITIZE these signals and after that GENERATE a response.
The questions are(I know the hardware for generation has not been specified):

1.- Is the material(LabVIEW and PXI-5122) appropiate for the task?(there could be more antennas in the future) (LabVIEW running in Windows 2000(multi-task) and sharing the bus PCI of the computer with other cards also)

2.- Can we have troubles of time, in order to process the signals at these rates(30MS/s minimum,5122 is a 14bit resolution digitizer, if I'm not wrong)

3
.- Should be Real-Time used?

4.- Would the bus PXI and PCI be a problem(I refer to the transfer rate) when working with both channels and at these rates in a continuous way?

5.- In order to generate the signals, could we use the NI-5411 Arbitrary Waveform Generator?

Thanks a lot.

[Jeremy R.]

enekofg,

I will post the answers in the form they were asked:

1. The materials you have are appropriate. LabVIEW will use the NI-Scope driver to directly program the 5122 and there are several example programs to get you started. The only consideration will be channel count. The 5122 is only 2 channels, so you will need multiple boards or a switch to multiplex through each input signal. I'm assuming you want coninuous measurements, so multiple boards is probably the way to go.

2. The PXI backplane has the same bandwidth specs as a PCI bus. It will be able to handle around 130 Mbytes a second. Because the digitizers will be transferring the data 1 record, this should not be a problem. If you begin to add several boards all sampling at 30 MS/s, the
n you might have to start considering the limitations depending on how many records you are acquiring.

3. Real Time will obviously give you more control than a non-deterministic windows operating system. In this case, the most important thing is the memory buffers where the data is stored. RT would really only help you get the data into LV more efficiently, but the memory buffers will ensure that your data is acquired correctly. The 5122 is definitely compatible with RT if that is needed though.

4. Answered with question 2.

5. The arbitrary waveform generator can definitely generate the signals that you need. I'm not exactly sure what signals you want to generate, but if you can get the signal in LabVIEW, then you can download that waveform into the 5411. The main consideration here is how big is the waveform. That will tell you how much on board memory you will need with the 5411.

Hope this helps. Good luck with your application.

Best Regards,
Jeremy Ross
NI
- Applications Engineer

[enekofg]

First of all, thanks Jeremy for your answer.

1. To the first answer. Let me have a go to see if I'm right. Let's say I sample a channel at 30MS/s. As 5122 is a 14 bit digitizer, it needs 2 bytes per sample. Then, each channel would use 60Mb/s in a continuous way. If I use two channels, 120Mb/s. Am I right? So, as it is the case, I need more channels, wouldn't the bus be a limiting factor?

2. Related to the second answer. If I have to fetch the signal from the memory buffer, process it and generate an output, wouldn't the whole loop be very slow and couldn't I lose data(memory overwritten) because of this parameter? (Acquiring data is not the goal of the project, but generating a response)

Isn't then Real Time needed?(To have a little jitter from on
e loop time to another)

5. The signals I want to generate are the response(generated by LabVIEW) to the acquired signals. Then, we are talking about a signal that's changing all the time. I think 5411 cannot generate data in a continuous way. First you have to load the signal onto the onboard memory and then generate it. Am I right?

Thanks again.

[Jeremy R.]

Based upon your furthur description of the project:

1. I you are aqcuiring at 30MS/s at 14 bits, then your limiting factor will be the PCI bus bandwidth. I blelieve the spec to be 133 Mb/s. This is actually an ideal spec, so you really can't count on more than about 110 Mb/s. You might be OK with 2 channels depending on how fast you are fetching each record, but you will definitely have to be careful because you are getting close to the limits here.

2. If that is the way you are processing your data, then RT would definitely give you an advantage because the software controlled loop now becomes very time critical.

3. The 5411 can be updated on the fly to output data, but this board will not give you that much more added functionali
ty over an analog output board in this case. This also brings up PCI bus bandwidth issues because you will need more of it for the output operation as well.

Best Regards,
Jeremy R.
NI - Applications Engineer