Single volume gain math

Hello,

The conversion from dB to linear gain (for both positive and negative dB):

G = 10 ^ (dB/20)

To convert the linear gain to an integer (32.0 format) variable, multiply by (2^24) which is 16777216.

Example: -20 dB G = 10 ^ (-20/20) = 10 ^ (-1) = 0.1

Integer value = 0.1 * 16777216 = 1677722 base 10, equal to hex 19 99 9A -- the same as the Capture Window shows for -20 dB.

Inside your microcontroller you'll have the binary value stored as a long integer, you just need to break this 32 bit word into four 8-bit bytes for safeloading. My C compiler has a "make8" function for this purpose.

Best regards,

Bob

Thank you Semi

Hello Bob,

I think we should use 8388608 instead of 16777216.

Integer value=0,1 * 8388608 = 838861 (B10), hex C CC CC.

C CC CD I see at the Data capture window.

For 0 dB we have 1 * 8388608 and it is equal to 80 00 00 hex.

Regards,

Przemek

Hello Przemek,

You are correct, the multiplier for the Sigma 100 / 200 chips using the 5.23 number format is 8388608. Although the original post does not mention the particular SigmaDSP in use, it appears I had figured the calculation for the Sigma 300 / 350 chips like the ADAU1452. These use a 8.24 number format -- thus the multiplier is twice what you have found for chips using the 5.23 format. We can observe the difference below, where I cut and pasted the ADAU1701's -20dB setting below that of the ADAU1452 to save space.

Thank you for bringing this up as it clarifies the difference between Sigma 100/200 and Sigma 300+ chips.

Best regards,

Bob