Re : Sub sampling method

[neil84]

Hi all,

I'm using a PCI6259 device with desktop labview RT.

My loops run at 100us

as the card samples the real signal a 1Mhz, there is a subsampling between the hardware and my loop, I need to know the way that the signal is subsampled: is it a decimation, an averaging, filtering, ...

thank you

[chat-relotte]

Hi neil84,

I'm not sure I understand what you are asking.

How are you doing your acquisition? Are you acquiring one sample at a time in a loop running at 100µs? If you are doing it this way, you are not able to know when the sample is taken between the 100µs timeframe.

If you want to acquire a sample every 100µs, you need to use the continuous acquisition mode and define the frequency at 10kHz. Then the card is using the 1MHz clock to create its sample clock at 10kHz which ensures that you are getting a value every 100µs.

I hope this helps,

[neil84]

Are you acquiring one sample at a time in a loop running at 100µs?

_______________________________________________________

Yes,

as the hardware samples the signal at 1Mhz and my loop acquires the data at 10kHz, there is a subsampling between the two.

I need to know the method of subsampling:

-decimation: that means that every 100 samples, I acquire only one (1MHz-10kHz)

-averaging: tha means that the program acquires the mean of every 100 values

-filtering: digitally filtering at 10kHz and decimation

What I need to know is what happens really between the PCI card and the labview VI

[Dennis_Knutson]

The hardware is not sampling at 1MHz if you are acquiring a single sample at a time. Your hardware would sample at 1MHz if that is the rate you specify and you request multiple samples. There is no decimation, averaging, filtering, etc., going on unless you use a decimation, averaging, filtering, etc., function. Are you asking how to do one of those?

[neil84]

Thank you for your answer,

__________________________

There is no decimation, averaging, filtering, etc., going on unless you use a decimation, averaging, filtering, etc., function

__________________________

So I'd like to know what exactly happens between the real physical signal and the data that I use in my labview program.

__________________________

Are you asking how to do one of those?

__________________________

That was not my initial question but I'm interested in knowing that

thanks

[Dennis_Knutson]

Not sure what you really want to know. Is it how an A/D converter works?

[neil84]

no, that is not the puspose of the topic

traditionally the scheme of data acquisition is like that:

physical signal, anti-aliasiong, high frequecy A/D conversion, subsampling to effective program running frequency, sending to program

As you say that there is no subsampling, did this means that the hardware A/D conversion don't run always at the same frequency?

if this is true so the anti-aliasing filter don't have the same frequency also. is this correct?

If the A/D convertion and the anti-aliasing filter have fixed frequency so there must be a subsampling method to convert the sampled signal at the A/D converter frequency the an other sampled signal at the programe frequency.

I hope I was clear this time

[Dennis_Knutson]

The A/D conversion always runs at the frequency you specify if you are requesting multiple samples. When you take a single sample, there is obviously no frequency except what you set in your acquisition loop and that has nothing to do with the hardware. The 6259 does not have an anti-aliasing filter so that question is sort of moot. There are some boards that have anti-alias filters and I believe that some of them are settable. They are placed before the A/D so the sample timing has nothing at all to do with the cut off frequency of the filter. Not sure how they would work with a non-hardware timed acquisition like you are using now. Why don't you simply use hardware timing?

In your definition, don't understand at all where you get this 'subsampling' idea. Remove that and also replace 'high frequency A/D conversion' with 'A/D' conversion at a settable frequency'. In general, you can specify a sample rate of anything up to the max of the board. You seem to have some misconceptions of NI DAQ boards.

[Dennis_Knutson]

Correction - Since the anti-alias filter is placed before the A/D, it is not at all related to how you are doing the sampling.

[neil84]

When you take a single sample, there is obviously no frequency except what you set in your acquisition loop and that has nothing to do with the hardware. The 6259 does not have an anti-aliasing filter so that question is sort of moot.

__________________________________________________

Ok,

Yes, I forgot that there is no anti-aliasing filter on 6259, however I have installed 10Khz (from NI) anti-aliasing filter before the board.

This is not good as the signal is sampled at a lesser frequency than the filter.

So the question is:

How to implement an A/D conversion at 1Mhz and subsample the signal by the averaging method?

thanks