QPSK HDL Example Data Represantation

Hi all,

I am trying understand the basics of QPSK Transmit and Receive Simulink HDL Design

The thing I noticed here is that both receive and transmit I and Q data paths are defined sfix16_En14, which is 16 bit word length with a 14 bit fraction length. Is there any special reason to use it in this way? Because, I would use int16 input and outputs for the block and do some logic operations to fit it 2s compliment representation as defined by Analog Devices?

Is there anything I am missing?

Any help would be appreciated

Thanks in advance.

Best,

Parents
  • +1
    •  Analog Employees 
    on Jan 20, 2021 4:38 PM 2 months ago

    sfix16_En14 was used to convert the data representative data range to [-1,1] to make it easier to manage downstream and is a more common range for communication systems than ADC codes.

    I would recommend looking at some of the MATLAB fixed-point examples and tutorials to understand some flows on how to manage fixed-point types. This can be difficult to understand in larger models since most choices and tradeoffs are up to the designer.

    Because, I would use int16 input and outputs for the block and do some logic operations to fit it 2s compliment representation as defined by Analog Devices?

    There are no issue using int16's (0 fractional bits) as long as that representation makes sense.

    -Travis

Reply
  • +1
    •  Analog Employees 
    on Jan 20, 2021 4:38 PM 2 months ago

    sfix16_En14 was used to convert the data representative data range to [-1,1] to make it easier to manage downstream and is a more common range for communication systems than ADC codes.

    I would recommend looking at some of the MATLAB fixed-point examples and tutorials to understand some flows on how to manage fixed-point types. This can be difficult to understand in larger models since most choices and tradeoffs are up to the designer.

    Because, I would use int16 input and outputs for the block and do some logic operations to fit it 2s compliment representation as defined by Analog Devices?

    There are no issue using int16's (0 fractional bits) as long as that representation makes sense.

    -Travis

Children