The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Measurements >> Other Measurements >> Can Code Density method insure ADC monotonicity?
https://designers-guide.org/forum/YaBB.pl?num=1191997963

Message started by ywguo on Oct 9^th, 2007, 11:32pm

Title: Can Code Density method insure ADC monotonicity?
Post by ywguo on Oct 9^th, 2007, 11:32pm

Hi Guys,

Many literatures state that DNL < 1 LSB guarantee ADC monotonicity. We often use code density (histogram) method to measure the DNL/INL. Can this method insure ADC monotonicity? Furthuremore, as indicated in the tutorial http://www.analog.com/zh/content/0,2886,761%255F795%255F91286%255F0,00.html, there are code transition noise. So how do I identify if the ADC is monotonicity when there are code transition noise? Is code density method enough to detect any non-monotonicity?

Thanks
Yawei

Title: Re: Can Code Density method insure ADC monotonicit
Post by Frank Wiedmann on Oct 10^th, 2007, 1:30am

You cannot test ADC monotonicity with a histogram test. See e.g. page 362 of the Stanford EE315 Course Reader by Boris Murmann, available in the handouts section of http://eeclass.stanford.edu/ee315/.

Title: Re: Can Code Density method insure ADC monotonicit
Post by vivkr on Oct 10^th, 2007, 2:42am

Ah! That one confirms a long-standing suspicion of mine. Thanks for that link Frank.

However, I have to say that a reasonably well-designed ADC ought not to have such weird flips. After all, it is not so
easy for 2 adjacent codes to get flipped without any visible impact on the other codes.

Don't you think?

Regards
Vivek

Title: Re: Can Code Density method insure ADC monotonicit
Post by ywguo on Oct 10^th, 2007, 7:33pm

Frank,

Thank you. Your reply clear my confusion.

Normally, which test method is used to check the monotonicity? Checking the ADC output directly?

BTW, the last slide on page 362, it reads that INL looks a lot like there are 5 missing codes. I don't understand it. :(

Yawei

Title: Re: Can Code Density method insure ADC monotonicit
Post by Frank Wiedmann on Oct 11^th, 2007, 12:13am

Yes, you will have to look at the ADC output directly.

A missing code is defined as a code with DNL≤-0.9 by IEEE Standard 1241-2000. I believe that Boris Murmann is referring to the fact that these codes may be much easier to find in the INL plot, which is the integral of the DNL plot and where the missing codes appear as downward steps. This is especially true if you have very many codes like in his example so that there may not be a pixel in the plot for every single code. I cannot tell you where exactly he is seeing the 5 missing codes, however.