Conversion to Fixed Point

Hi All,

I am a novice to the business of Fixed Point. I have read various papers
on implementation of Fixed Point and understood many concepts but have a
basic doubt in mind. 
Actually I am trying to convert an existing filter code available in
floating point to fixed point. The program accepts floating point values
from a data file and outputs a data file after filtering. I am wondering
how to convert this input data to fixed point. As the data file may change
it has to be converted inside the program only. So to convert, do I have to
multiply the number directly by powers of 2 depending on the Integer Word
Length (as we can not use shifting left or right with floating point, the
multiplication is the only option) or is there any other way which will
not include multiplication at all. Similarly while generating output how
can I convert back to floating point without dividing the number by power
of 2. I have this question because if actual multiplication is done then
it will be done in floating point only. Then I wont be able to generate a
code which will fully work on Fixed Point.

Sorry for such a big mail,

Kindly Reply,

Thanks and Regards,
bd.


		
This message was sent using the Comp.DSP web interface on
www.DSPRelated.com

http://home.earthlink.net/~yatescr/fp.pdf

http://home.earthlink.net/~yatescr/fir.pdf

Hope you will get all your answers
-sm