Conversion from sfixed to ufixed (gives error)

Hi
I am trying to convert a signal from sfixed to ufixed, using the syntax given in the fixed package documentation:

-- Conversion from sfixed to ufixed (performs an "abs" function)
function to_ufixed (
arg : UNRESOLVED_sfixed)
return UNRESOLVED_ufixed;

My code:
SIGNAL result : ufixed(8 DOWNTO -4);
SIGNAL position : sfixed(8 DOWNTO -4);
:
:
result<=to_ufixed(position);

In quartus 9.1 the following error message is given:
Error (10476): VHDL error at hjulenkodernxt.vhd(96): type of identifier "position" does not agree with its usage as "UNSIGNED" type

Have I done anything wrong? Is it a alternative way to do this conversion?

I am not familiar with VHDL,from verilog view,in the "position",signed number would not be recognized.
And if you want to implement "abs" function ,here show you an example in verilog(16bit):
assign abs_num_diff = num_diff[15]?
(~num_diff[15:0] + 1'b1) : num_diff[15:0];

no problem for me.
Are you sure you havent got a local variable called result overriding the result you posted?

TrickyDicky said:

no problem for me.
Are you sure you havent got a local variable called result overriding the result you posted?

Click to expand...

Hi

Just for testing the conversion I removed all other code, and only used this:

Library ieee;
USE ieee.std_logic_1164.all;
USE ieee.std_logic_signed.all;
use ieee.numeric_std.all;
library ieee_proposed;
use ieee_proposed.fixed_pkg.all;

ENTITY HjulEnkoderNxt IS
PORT (TACHO0,TACHO1: IN STD_LOGIC;
Clock : IN STD_LOGIC;
distanse :OUT ufixed(8 downto -4));
END HjulEnkoderNxt;

ARCHITECTURE Behavior OF HjulEnkoderNxt IS
SIGNAL scalingfactor : sfixed(2 downto -4);
SIGNAL position : sfixed(8 DOWNTO -4);
SIGNAL pulsesSfixed : sfixed(5 DOWNTO 0);
SIGNAL result : ufixed(8 DOWNTO -4);

BEGIN
scalingfactor<=to_sfixed(1.0,scalingfactor);
pulsesSfixed<=to_sfixed(2.0,pulsesSfixed);
position<=Scalingfactor*pulsesSfixed;
result<=to_ufixed(position);--Conversion from sfixed to ufixed
END Behavior;

I still get the error message:
Error (10476): VHDL error at hjulenkodernxt.vhd(95): type of identifier "position" does not agree with its usage as "UNSIGNED" type

I copied and pasted your code - no problems.
Your error points to line 95, which is not in your cut down code. So please post the full code.

Also - remove ieee.std_logic_signed from your code. You dont need it because you are using numeric_std.

TrickyDicky said:

I copied and pasted your code - no problems.
Your error points to line 95, which is not in your cut down code. So please post the full code.

Also - remove ieee.std_logic_signed from your code. You dont need it because you are using numeric_std.

Click to expand...

My code is:

Library ieee;
USE ieee.std_logic_1164.all;
use ieee.numeric_std.all;
library ieee_proposed;
use ieee_proposed.fixed_pkg.all;

ENTITY HjulEnkoderNxt IS
PORT (TACHO0,TACHO1: IN STD_LOGIC;
Clock : IN STD_LOGIC;
distanse :OUT ufixed(8 downto -4));

END HjulEnkoderNxt;

ARCHITECTURE Behavior OF HjulEnkoderNxt IS
SIGNAL scalingfactor : sfixed(2 downto -4);
SIGNAL TACHO0prev,TACHO1prev : STD_LOGIC;
SIGNAL pulses : SIGNED(5 DOWNTO 0);
SIGNAL position : sfixed(8 DOWNTO -4);
SIGNAL pulsesSfixed : sfixed(5 DOWNTO 0);
SIGNAL result : ufixed(8 DOWNTO -4);



BEGIN
scalingfactor<=to_sfixed(1.0,scalingfactor);--Setter verdien til skaleringsfaktoren.
PROCESS(TACHO0,TACHO1,Clock,pulses)
BEGIN
IF rising_edge(Clock) THEN

IF (TACHO0='0' AND TACHO1='0') THEN
IF (TACHO0prev='1' AND TACHO1prev='0') THEN
TACHO0prev<='0';
TACHO1prev<='0';
ELSIF (TACHO0prev='0' AND TACHO1prev='1') THEN
TACHO0prev<='0';
TACHO1prev<='0';
ELSIF (TACHO0prev='0' AND TACHO1prev='0') THEN
TACHO0prev<='0';
TACHO1prev<='0';
END IF;

ELSIF(TACHO0='0' AND TACHO1='1') THEN
IF (TACHO0prev='0' AND TACHO1prev='1') THEN
TACHO0prev<='0';
TACHO1prev<='1';
ELSIF(TACHO0prev='0' AND TACHO1prev='0') THEN
pulses<=pulses-1;
TACHO0prev<='0';
TACHO1prev<='1';
ELSIF(TACHO0prev='0' AND TACHO1prev='0') THEN
pulses<=pulses-1;
TACHO0prev<='0';
TACHO1prev<='1';
ELSIF(TACHO0prev='1' AND TACHO1prev='1') THEN
TACHO0prev<='0';
TACHO1prev<='1';
END IF;
END IF;

IF (TACHO0='1' AND TACHO1='1') THEN
IF (TACHO0prev='1' AND TACHO1prev='1') THEN
TACHO0prev<='1';
TACHO1prev<='1';
ELSIF(TACHO0prev='1' AND TACHO1prev='0') THEN
pulses<=pulses+1;
TACHO0prev<='1';
TACHO1prev<='1';
ELSIF(TACHO0prev='0' AND TACHO1prev='1') THEN
pulses<=pulses-1;
TACHO0prev<='1';
TACHO1prev<='1';
ELSIF(TACHO0prev='1' AND TACHO1prev='0') THEN
TACHO0prev<='1';
TACHO1prev<='1';
END IF;
END IF;

IF (TACHO0='1' AND TACHO1='0') THEN
IF (TACHO0prev='1' AND TACHO1prev='0') THEN
TACHO0prev<='1';
TACHO1prev<='0';
ELSIF (TACHO0prev='0' AND TACHO1prev='0') THEN
pulses<=pulses+1;
TACHO0prev<='1';
TACHO1prev<='0';
ELSIF (TACHO0prev='1' AND TACHO1prev='1') THEN
TACHO0prev<='1';
TACHO1prev<='0';
END IF;
END IF;

pulsesSfixed<=to_sfixed(pulses,5,0);
position<=Scalingfactor*pulsesSfixed;
distanse<=to_ufixed(position);--Conversion from sfixed to ufixed (Line 94)

END IF;

END PROCESS;

END Behavior;

Quartus message: Error (10476): VHDL error at hjulenkodernxt.vhd(94): type of identifier "position" does not agree with its usage as "UNSIGNED" type

I can only think that you have modified or have an incomplete version of the fixed package. there should be 2 to_ufixed functions, one using sfixed and using unsigned. it looks like it can only see the latter ( protection could be an old version of the library). alternatively just do abs yourself and just type covert to a ufixed.

TrickyDicky said:

I can only think that you have modified or have an incomplete version of the fixed package. there should be 2 to_ufixed functions, one using sfixed and using unsigned. it looks like it can only see the latter ( protection could be an old version of the library). alternatively just do abs yourself and just type covert to a ufixed.

Click to expand...

I downloaded the package from: **broken link removed** so I think it should be updated.
Could you post an example of how to use the abs function and type convert? I have tried this, but I do not get it working.
I used this abs syntax:
-- Absolute value, 2's complement
-- abs sfixed(a downto b) = sfixed(a+1 downto b)
function "abs" (arg : UNRESOLVED_sfixed) return UNRESOLVED_sfixed;

you will need a temporary signal or variable.

signal a : sfixed(position'high + 1 downto position'low);

a <= abs(position);
distanse <= ufixed(a);