Converting from floating point to fixed point.

How can i convert from floating point to two's complement? I need to convert a vector of filter coefficients from Filter Designer but I do not know how.

Answer 1

[Walter Roberson]

The Fixed Point Toolbox fi() can be used to create a variety of fixed point formats.

If you need to match a particular hardware fixed point format for a dsp then we would need to know the details of the representation.

[Walter Roberson]

format long g

BitsAfterDecimalPoint = 12;
Value = -0.0126842654631061924064283630286809057

Value = 
       -0.0126842654631062

encoded = int16(Value * 2^BitsAfterDecimalPoint)

encoded = int16
   -52

decoded = double(encoded) / 2^BitsAfterDecimalPoint

decoded = 
             -0.0126953125

abs(Value - decoded)

ans = 
      1.10470368938076e-05

There is a limit to how much precision you can get with 16 bits...

BitsAfterDecimalPoint = 16;
encoded = int16(Value * 2^BitsAfterDecimalPoint)

encoded = int16
   -831

decoded = double(encoded) / 2^BitsAfterDecimalPoint

decoded = 
       -0.0126800537109375

abs(Value - decoded)

ans = 
      4.21175216869241e-06

Answer 2

[ON]

I need to convert for example -0.0126842654631061924064283630286809057 into a 16 bit signed value.

[Walter Roberson]

 int16(Value * 2^BitsAfterDecimalPoint)