Original question:
[This is a follow-up on my previous query about offset error (OSE) and zero scale error (ZSE).]
Owing to the fact that linearity is calculated using a reduced code range (somewhere between 10% & 90% FSR), and offset error is the difference between measured and ideal output voltage in the linear region of the transfer function, does it mean that we have an offset error (OSE) per code?
Thank you very much.
Hi Abe,
I suppose you could say that offset error is present at all codes, but I would not say there is some offset per code. When we combine all error sources, you can get a spec called 'total unadjusted error' or TUE, that we show on the typical curves. Consider this curve from the DAC80508 datasheet:
The 'fuzziness' of the line is the code-to-code error contribution of the INL/DNL type error. You can see that most of these lines have a positive offset and a gain error. DAC1 (red line) has almost exclusively offset error as the line is very flat. While DAC6 (purple) has a smaller offset error, but it does have a gain error. Somewhere around code 20000 the line crosses 0. This does not mean that we no longer have an offset error, only that the gain error has canceled out the offset error at that point. If this channel had no offset error, then the total error would be greater as the gain error would dominate.
Thanks!
Paul
