Serial signal in DAC

angelo Guest 2014-09-16 09:10 #3804995

Rate this post

•	▲ useful ▼ not useful

Hi everyone,

I'm writing a code to simulate a 12-bit dac. The behavior is quite 
simple:

A serial digital input receive datas ( 32-bits word with the 12 bits to 
convert + don't care bits + command) and the 12 bits word is converted 
every 32 clk.

My first try was to create a signal data_input: std_logic_vector (31 
downto 0) and convert only the 12 bits I want. But this not exactly the 
right thing.

I thought a clk counter is necessary and others things but can't find 
out...

Can you give me some advice to get over it please?

Reply Quote Edit Move Thread sperren Anmeldepflicht aktivieren Anpinnen Delete topic Thread mit anderem zusammenführen Report post

J. S. (engineer) 2014-09-16 11:39 #3805204

Rate this post

•	▲ useful ▼ not useful

Basically you will need two clocks since the left side domain is totally 
analog and independent from the sampling domain. Practically you will 
work with ps resoultion in order to feed your DAC with data. Furthermore 
you will need a kind of anti aliasing filter for the incoming data 
stream to simulate aliasing which will occur in reality. Next steps are 
non linearity and noise.

Reply Quote Edit Delete Report post

Lothar M. (lkmiller) Moderator (Company: Titel) 2014-09-16 12:00 #3805238

Rate this post

•	▲ useful ▼ not useful

angelo wrote:
> I'm writing a code to simulate a 12-bit dac.
> ...
> and the 12 bits word is converted every 32 clk.
Into what?

> signal data_input: std_logic_vector (31 downto 0)
a 32-bit input doesn't match
> A serial digital input
What input do you have? What output do you need? What is the actual 
problem?

Reply Quote Edit Delete Report post

angelo Guest 2014-09-16 12:23 #3805268

Rate this post

•	▲ useful ▼ not useful

I wrote this code in a first time:

library IEEE;
use IEEE.std_logic_1164.all;
use ieee.numeric_std.all;
entity dac is
  generic (data_width : integer;
     SDI_width  : integer
     --REF        : real
  port(  REF   : in  real;
    SDI   : in  std_logic_vector (SDI_width - 1 downto 0);
    SDO   : out std_logic_vector(SDI_width - 1 downto 0);
    SCK   : in  std_logic;
    CLR   : in  std_logic;  -- low level active
    CS_LD : in  std_logic;   -- '0': CS / '1': LD
    LDAC  : in  std_logic;  -- low level active (always '1' here)
    Vout  : out real
end dac;
architecture archi of dac is  
signal data      : std_logic_vector(11 downto 0);
signal data_real : real;
  data <= SDI(15 downto 4);
  P1: process(SCK, data)
  constant Tclk : time := 10 ns;
    data_real <= (real(to_integer(signed(data))));
    SDO <= transport SDI after Tclk*32; 
    if (CLR = '0' or CS_LD = '0') then
      Vout <= 0.0;
    elsif (SCK'event and SCK = '1') then
      Vout <= data_real/(2.0**data_width) * REF;
    end if;
  end process;
end archi;


The problem is I have to connect the signal SDI to an 1 bit signal. 
Which is not possible if SDI is a std_logic_vector.

Reply Quote Edit Delete Report post

Lothar M. (lkmiller) Moderator (Company: Titel) 2014-09-16 12:51 #3805306

Rate this post

•	▲ useful ▼ not useful

angelo wrote:
> Which is not possible if SDI is a std_logic_vector.
SDI is not a vector at all! It is a simple 1 bit std_logic input.

And together with LD it is very clear what has to happen here:
1. LD is set to '0'
2. with SCLK bits from SDI are shifted into a 32 bit register in the 
DAC
3. with LD set to '1' the register is evaluated

So it may look like this:

    SDI   : in  std_logic;
    SCK   : in  std_logic;
    CS_LD : in  std_logic;   -- '0': CS / '1': LD
    signal   inreg : std_logic_vector (31 downto 0);
    inreg <= inreg(30 downto 0) & SDI when rising_edge(SCLK) and CS = '0';    -- shift data in, assuming MSB first
    data <= inreg(15 downto 4) when rising_edge(CS_LD); -- transfer interesting part of vector to analog processing

Got the idea?

Reply Quote Edit Delete Report post