Resample Buffered data from AI-READ to File

[educhene]

Hi,

I have a pb with my data acquisition : I'm using a PXI-4472 AI card and continuous acquiring data to file.
Until now all channels are acquired at the same frequency : Ok. I know this card does not support multi-frequency acquisition.
But I want save data at different rate : ex : first channel at the frequency desired (ex 10 KHz), second channel at frequency /2 : 5 Khz ... to reduce amouth of data in the saved file.
I know the amouth of data retrieve et each AI-Read (ex 2048 scaled data / channel), and the frequency divisor / channel.
What's the best way (performance) to process the scaled data tab (2 dim) from the AI-READ to save it in the file (using Write File +) with frequency divisor consideration.
And how to re
ad the data in the saved file after.


Can you send me an example please ???


Thanks"

[johnsold]

Separate the channels you want to reduce and pass them to the Decimate.vi. It lets you choose the reduction factor, n, and decide whether to select every n th sample or to average n samples for each output point. Write a header to the file recording the decimation factor for each channel. When you read the file that decimation factor will tell you how to align the data.

Lynn

[educhene]

The Decimate.vi ????

What's this ??

Eddy

[johnsold]

Analyze>>Signal Processing>>Time Domain>>Decimate.vi. This VI is used to reduce the number of samples in a consistent way.

It is not included in the base package, only the Full or Professional packages of LV 7. It has been present in earlier versions also.

Lynn

[educhene]

Thanks for your help : I didn't know this Vi.

But I'm doing the same thing using the vi Tab -> Decime Tab (I'm using a condition struct depending of the divide factor).

Thanks.

Eddy

[chilly_charly]

Although possible, it is not easy to save all the data in a single file. The easiest way would be to separate the channels and to decimate the data according to your division/channel, before saving in SEPARATE files.
But apparently that's not what you intended to do.
Here is a way of storing data easily in a single file.
Assumming that you have 2 channels (works also on n channels), channel 2 being (virtually) sampled at half rate, if you replace your "unwanted" data with �empty string- but leave the Tab separators, your saved file will look something like
ch 1 TAB ch 2 RETURN
ch 1 TAB RETURN
ch1 TAB ch2 RETURN
ch 1 TAB RETURN
and so on�
Writing the file is not a problem.
Retrieving the data back is a bit more difficult, since you have to avoi
d confusing empty strings with zero values !

See the example.below for 3 channels.

If file size is really a problem, you should also consider saving your data in binary format, and not in ASCII.

[educhene]

Thanks for your solution, but I need to save data in binary format (due to perf pbs).

I've succeeded to resolve my pb, but I have done a lot off data manipulation (especially to read them from file).

Thanks.

Eddy

[cb34]

Hi there! I have the same problem with you. I am using the PCI 6052e to sample different frequency rates.

Is it possible that you can send me an example on how u did it. I am really having a hard time figuring it out. Thanks!

[educhene]

Ok,

So I'm retrieving scaled datas (max 8 channels) from AI-READ and transform them (2D tab) to save them in a binary file (using 1D tab).
To do that, I'm using the "Convert Scaled Data to Tab of Scaled Data With Divisor.vi" : it suppress data overflow for each channel (in case of divisor > 1) and save data in 1D tab.
After that, save this 1D tab in file using the "Write File+ [SGL].vi".

To read the binary file, I know the divisor parameter for each channel because I save it in the user header file.
Then I use the "Read_Datas_Acq_File_New.vi" to retrieve datas in the file depending of the Scan Begin and Scan End.

Good Luck !