Problem with DSP_fir_cplx() from DSPLIB 3.2.0.1

Other Parts Discussed in Thread: TMS320C6657

Hi,

I'm struggling to get the DSP_fir_cplx() function to produce any reasonable output. I'm currently testing it with a real, symmetric filter and with a very basic input signal:

short h[32] =
{
  -57, 0,  -32, 0,  289, 0,  251, 0,
 -827, 0, -969, 0, 2488, 0, 7049, 0,
 7049, 0, 2488, 0, -969, 0, -827, 0,
  251, 0,  289, 0,  -32, 0,  -57, 0,
};

short x[48] =
{
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
  1, 1,
};

short result[16] = {0};

Here's the way I used to call the function (`old way', DSPLIB 3.1.0.0):

DSP_fir_cplx(&x[30], &h[0], &result[0], 16, 8);

And here's the output that I get:

result =
{
  1, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
};

This is not symmetric so clearly it's wrong. I noticed that in the latest header file (DSP_fir_cplx.h) the API has changed and that now I should be calling DSP_fir_cplx() like this:

DSP_fir_cplx(&x[30], &h[32], &result[16], 16, 8);

In this case, however, I get the following:

result =
{
  0, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
  1, 0,
};

 which is also non-symmetric, hence wrong. Moreover, according to the reference C files (DSP_fir_cplx.c and DSP_fir_cplx_cn.c) the API hasn't changed! This is all getting a bit confusing.

Does anyone anyone know the right way of calling DSP_fir_cplx() for my simplified input data? Otherwise I'm not able to use in more realistic settings.

I am testing it with a TMS320C6657 board, but this seems only mildly relevant. I also made sure that all the input is properly aligned.

Thanks!