Have u guys heard about the truncation error?

I am now working on the project about digital PWM in the FPGA. But i encountered the problem: the output of 1 block is 24 bits and this signal will be in the input of another block, which require input 12 bits. So it meant that i need to change the output 24bits to the 12 bits input by somehow.

I read something about the fixed point arithmetic . They mentioned about get rid of 12 bits LSB of your 24 bits , but u will encounter the truncation error or something.

So you guys have any ideas about that ? and about the truncation error , got any recommendation for me to read, so that i can analysis the error.

If anything doubt , i can explain again.

Thank for helping me.

Clark

I would recommend the sections about "quantization error" in the book Discrete Time Signal Processing by Oppenheim and Schafer. Alternatively, this topic should be covered in nearly any digital signals or digital communications text.

Simply, if the expected dynamic range of your signal does not exceed approx ~65dB (for 12 bits), you shouldn't need to worry. If it does exceed 60dB, then you will need to look into techniques like noise shaping. If your signal uses all it's 24 bits of dyanmic range and all the available bandwidth at the system sample frequency, then you are more or less stuck.

thank a lot . I will find the book and see how .

Thank and happy x'mas