Rif.: Fatal Run-Time error with Decimate function

[vix]

I have a buffer where the samples are interleaved (ch0, ch1, ch0, ch1, ....) and I need to separate the channel, so I tried to use Decimate as suggested here.

The code is really simple:

double    interleavedSamples[100], ch0[50], ch1[50];
    
    Decimate (interLeavedSamples, 100, 2, 0, ch0);
    Decimate (interleavedSamples + 1, 100, 2, 0, ch1);

but the second call to Decimate gives Fatal Run-Time Engine - Array argument too small.

This is not true, because with a decimation factor of dFactor, you can get Nout samples if the input buffer has at least dFactor*(Nout-1)+1 and not dFactor*Nout

I tried a workaround, specifying 99 as numberOfElements parameters in the second call to Decimate, but this doesn't work, because as written in the help file

size = trunc(numberOfElements/dFactor) is the size of the output sequence

and so only 49 elements are returned in ch1 buffer.

How can Decimate be used to separate interleaved samples?

[Wolfgang]

Vix,

in the second function call you should give the address of the second element of your array, i.e. &interleavedSamples [ 1 ]; this is different from your code because the elements need more than 1 byte

[vix]

Hi Wolfgang,

my code and your suggestion are exactly the same, because interleavedSamples is in reality the address of the first element of the buffer and a pointer to a double.

For this reason interleavedSamples +1 is the same as &interleavedSamples[1]

But this isn't the problem: it is Decimate that expects an array of size dFact * Nout

[Wolfgang]

Ok,

attempt number 2: Did you consider increasing your initial array by one element, i.e. using a size of 101? Only the first 100 elements are filled with data, but this should avoid accessing an element outside of the array if you start with an odd element

[vix]

the fact is that the buffer comes from a DAQmx acquisition.

I tried this workaorund:

Decimate (interleavedSamples+1, 99, 2, 0, ch1);
ch1[49] = interleavedSamples[99];

[RobertoBozzolo · Accepted solution]

You should have noted that the size error is on interleavedSamples argument of Decimate. In effect the function expects to have 100 elements to treat starting from interleadedSamples[1] element, so you must size your starting array as interleavedSamples[101]. next you'll acquire 100 measures and the second Decimate will correctly extract samples [1] to [99], leaving (empty) sample [100] untouched.