3 Replies Latest reply on Jun 4, 2011 7:50 PM by jenoc_293011

    Delsig11 documentation or Toy story

              Ant the beginning of documentation there is a line: "... It is based on a 2.6V full scale input range centered around a user selected AGND, when the reference selection in the global parameter window is set to +/- Bandgap..." few line later... " The conversion equation governing the 11-bit Delta Sigma ADC is shown below. The equation shows that the input range is limited to Vref. The use of this equations is illustrated in this example:" First sentense say the measurement range is arounf Vref, the second, that it's limited to Vref. Does it make sense? Another example from the same doc first line: "Data format available in 2’s complement" And few lines later: "Left-justified data is stored with the upper 8 bits in the most significant byte and the lower 3 bits in the upper 3 bits of the least significant byte. The lower 5 bits contain residue of the decimator." Where is a sign bit or it's not a 2's complement anymore? Regards Robert   
        • 1. Re: Delsig11 documentation or Toy story

          few lines later:




          The result of the calculation is referenced to AGND. For a ADC data value of 1500 the Voltage measured can be calculated to be 0.60V:"


          2's comlement reading can not be 1500 because 11bit range is +- 1024





          • 2. Re: Delsig11 documentation or Toy story

            Another example:


            Delsig11 according to documentation return 8-bit value.





            • 3. Re: Delsig11 documentation or Toy story

              I don't see all that in the documentation I have. Could be you have some older documentation with mistakes that have been edited since then.

              • Pretty sure it's 11 bits. That sounds like a typo or copy/paste error, or a misunderstanding.
              • The range is +/- Vref.
              • The sign is in the upper bit(s). An 11-bit, two's-complement value is going to use 10 bits for the magnitude and 1 bit for the sign. An 11-bit unsigned value will just use 11 bits. Either way you get 2048 possible values.

              A two's-complement, 11-bit value that's right-justified in a 16-bit field (what you'd usually see) will set all the unused bits to the same as the sign bit. So your values range from -1024 (0xFC00 or 0b11111100000000000) to +1023 (0x03FF or 0b0000001111111111). The actual value is 0x400 to 0x3FF.


              Left-justified values are the same values shifted left by 5 so they have a 16-bit magnitude. I've never needed that myself but I assume it's handy for some situations.


              As for the example that includes 1500, I have to agree that's weird. Those examples suggest that the value retrieved will be unsigned with an offset of 1024, but the documentation clearly states elsewhere that the values are two's-complement, and even say the upper 5 bits are used to sign-extend. Again, I suspect the tech writer copied another document and didn't notice the discrepancy when he edited it.