I'm still working with this customer, and he brought up something I could use some clarification on: When we spec out "relative accuracy," he took it to mean accuracy between DAC A and DAC B. What do we actually mean by this?
It can be quite difficult to understand all these terms.
Relative accuracy is another way to say Integral non-linearity. This specification applies to the individual DAC output channel. Looking at the transfer function of the output. If you were to draw a straight line between the two end points of the DAC output. In this case code 8 to 248. It is the maximum deviation in LSBs of the actual from this straight line drawn trough the end points. A common misconception is that this takes into account gain error and offset error. Unfortunately it does not. Offset and gain error can be calibrated easily by the user.
In this case there is no specification dedicated to channel to channel matching. The offset in each channel can be different meaning the customer needs to calibrate the channel individually to get the performance required. Is there a chance the customer could calibrate each channel?
Here is a useful guide to calibrating out this offset: http://www.analog.com/library/analogDialogue/archives/42-06/DAC_calibration.html
Retrieving data ...