Hi guys,

I just upgraded from the 1701 to the 1452, and an having to rebuild all my blocks.

Is there a reference, by block, for what numeric format each block requires as input and produces as output? I feel like I'm wasting a lot of time trying to figure out what format each block requires. I'm sure a reference is available, but I'm having trouble finding it on the wiki and the help.

Suggestions?

thanks a lot,

Dave.

Hello Dave,

The short answer: You're having trouble finding a one-stop reference of numeric formats used by all the blocks, because there isn't one. However, the needed info is available, we just need to dig a little. First, let's review what you already know from using the ADAU1701:

All SigmaDSPs are fixed-point processors -- by necessity, the "decimal" format is set by design. Thus for the -1701 (and many others), the standard format for audio and other decimal values is five bits before, and 23 after, the binary ("decimal") point. Since this is hardwired into the chip, it's described at page 30 in the ADAU1701 data sheet. Many of us were first introduced to number formats when we first dragged a

DC Sourceinto a project. The standarddecimal formatfor a given chip has several significant properties:It defines "full scale"-- The decimal value+or-1.0is the clipping point for all I/O, although signals within your project can make use of the headroom provided by the bits before the decimal point.It establishes the multiply identity value-- Anything multiplied by1.0decimal remains unchanged. Thus a linear gain of 1.0 is a straight wire (0 dB gain).The other important number type is

integerformat, which simply represents all the binary bits as a whole number. In the ADAU1701 (etc.) integers are 28 bits wide, so it's called28.0format. Because the binary point shifts 23 places to the right from decimal to integer format, the integer 8388608 (that's 2^23) is a mere 1.0 decimal. Integers are used primarily asindexes for multiplexers and such.Unlike previous SigmaDSPs, the

ADAU1452handles 32-bit data words. Itsintegerformat is thus32.0.The chip's designers settled on a standard

decimalformat of8.24-- providing one more bit of full-scale resolution than the older DSPs, as well as three more bits of headroom. The -1452's number formats are described beginning at page 77 in the ADAU1452 data sheet.Block descriptions in the wiki generally specify their expected input and output formats -- either decimal or integer. Problem is, most wiki entries were written before the ADAU1452 arrived -- so the terms "decimal" and "5.23" are used interchangeably. -1452 number formats are also unrepresented in the wiki entry at Numeric Formats [Analog Devices Wiki] . You'll need to think "8.24" and "32.0" on your own with the ADAU145x ("Sigma 300 core") chips.

The good news is that,

zero is zeroin any format -- also, many logic blocks (except paradoxically the AND/OR gates, which are decimal) are designed to ignore format. They simply respond to zero as false, and any nonzero number as true.Best regards,

Bob