The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> DNL's equivalent noise in ADC
https://designers-guide.org/forum/YaBB.pl?num=1454606684

Message started by newroad on Feb 4^th, 2016, 9:24am

Title: DNL's equivalent noise in ADC
Post by newroad on Feb 4^th, 2016, 9:24am

Hi, when we calculate the SNDR or ENOB in ADC, we add all the noise together, i.e., quantization noise, kT/C noise, DNL noise, etc. Let's say DNL is +/- 0.5LSB and LSB=1mV, what's its equivalent noise in terms of mV? Thank you.

Title: Re: DNL's equivalent noise in ADC
Post by DanielLam on Feb 6^th, 2016, 2:11pm

First, quantization noise is a misnomer. It is a quantization error (deterministic). Designers often say quantization noise because they assume it is white (typically > 3 bits).

Second, I've never heard of DNL noise. So I don't think DNL noise exists.

In my experience, you get the noise from the FFT chart. For the INL/DNL, you take a lot of samples. A lot of samples depends on the what resolution and confidence interval you want (papers have been published). For example, I took ~3.5 million samples for an 11b ENOB ADC for a 97% (I think) confidence interval.

I don't think white noise sources affect the INL/DNL because they will average out. You just have to take enough samples. But the quantization error will remain because it is deterministic. I am assuming there are no other distortion terms besides quantization.

Regards,
Daniel Lam

Title: Re: DNL's equivalent noise in ADC
Post by sheldon on Feb 9^th, 2016, 10:23pm

Newroad,

Think what you are trying to say with DNL noise is that device
mismatch will limit the dynamic range. You don't need to calculate
the DNL to capture that effect. Usually you look for ways to avoid
statistical DNL simulations. If you have access to the Cadence
support site, there is an ADC Verification Rapid Adoption Kit
that describes how to calculate SINAD. It is not straightforward.

Sheldon