[Discuss-gnuradio] Re: Phase accumulator

[Brian Padalino]

On Tue, Aug 12, 2008 at 5:13 PM, Sebastiaan Heunis <address@hidden> wrote:
> Brian
>
> I had a look at some of the Verilog again.  For a decimation of 8,
> decim_strobe pulses every 64MHz/4 when FR_DECIM_RATE is equal to 3, so
> this is correct.  hb_strobe will then be 16MHz/2.  I'm looking at the
> phase_acc and cordic now.  When the DDC gets set to -20MHz as is the
> case when we're using a 7901 tvrx, we have the following c++ code to
> calculate the value to write to the FPGA:
>
> compute_freq_control_word_fpga (double master_freq, double target_freq,
>                                double *actual_freq, bool verbose)
> {
>  static const int NBITS = 14;
>
>  int   v = (int) rint (target_freq / master_freq * pow (2.0, 32.0));
>
>  if (0)
>    v = (v >> (32 - NBITS)) << (32 - NBITS);    // keep only top NBITS
>
>  *actual_freq = v * master_freq / pow (2.0, 32.0);
>
>  if (verbose)
>    fprintf (stderr,
>             "compute_freq_control_word_fpga: target = %g  actual = %g  delta 
> = %g\n",
>             target_freq, *actual_freq, *actual_freq - target_freq);
>
>  return (unsigned int) v;
> }
>
> v then gets written to FR_RX_FREQ_0.  This is used for the phase
> accumulator.  For -20MHz, v = 2952790016.  In the phase accumulator,
> we have:

Not quite - it's a signed number so it's really -1342177280.

> setting_reg #(FREQADDR)
> sr_rxfreq0(.clock(clk),.reset(1'b0),.strobe(serial_strobe),.addr(serial_addr),.in(serial_data),.out(freq));
>
>   always @(posedge clk)
>     if(reset)
>       phase <= #1 32'b0;
>     else if(serial_strobe & (serial_addr == PHASEADDR))
>       phase <= #1 serial_data;
>     else if(enable & strobe)
>       phase <= #1 phase + freq;
>
> so when serial_strobe = 1 and serial_addr is the address of
> FR_RX_FREQ_0, freq gets the value 2952790016?  enable comes from
> rx_enable which gets set in master_control.  strobe comes from
> sample_strobe, which gets set to 1 in master_control.  So on the first
> clock pulse after writing to FR_RX_FREQ_0, phase will be 2952790016?
> This value keeps on getting added to phase on every clock pulse?
> Phase is 32 bits, so is the value then truncated?  I know that the top
> 16 bits of phase is used in cordic.v.
>
> So does the following happen:  phase gets added each clock pulse, the
> overflow is discarded and cordic uses the top 16 bits of phase?  At
> the next clock pulse, phase changes, so phase[31:16] that enters
> cordic is different?  Or does phase get a once-off value that is used
> in cordic.v?  The word 'accumulator' gives me the idea that something
> keeps on getting added.

The first scenario sounds like the correct answer.  This is a phase
accumulation that feeds the CORDIC which will then perform the mixing
operation.  I think you get it.

> Then another question to anyone who has used the cordic testbed.  In
> cordic_tb.v included with Gnuradio, we have `include "sine.txt".  I'm
> assuming that for a 20MHz signal that needs to be mixed down, this
> will be a text file with samples of a 20MHz sine wave sampled at
> 64MHz?  So I can just generate it in Octave or Matlab and use it to
> run the testbed?

"sine.txt" looks to be a routine that will actually stimulate the
CORDIC.  You should probably do something more interesting that just a
20MHz tone.  Even a tone at 20.1MHz might be more interesting, so you
get a 100kHz tone out from the mixing process.

You're going to have to fill in the Verilog that should be in
"sine.txt".  I wrote a math_real.v package that is in the repository
here:

    http://gnuradio.utah.edu/trac/browser/usrp2/trunk/fpga/models/math_real.v

You can use that to generate sin, cos, etc within Verilog.

Hope this helps.

Brian