The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> Getting the INL from the DNL of an ADC
https://designers-guide.org/forum/YaBB.pl?num=1300415894

Message started by Mischivis on Mar 17^th, 2011, 7:38pm

Title: Getting the INL from the DNL of an ADC
Post by Mischivis on Mar 17^th, 2011, 7:38pm

Can I get the INL if I just know the DNL information of an ADC?
For example, if I have a 9 bit ADC and I know the DNL information
of the entire code (0.5LSB, 0.8LSB, etc.), can I get the INL from this?

Title: Re: Getting the INL from the DNL of an ADC
Post by boe on Mar 18^th, 2011, 3:06am

Mischivis,
Yes. The 511 DNL values of your 9bit ADC describe the step size between neighbouring codes, so you can determine the characteristic and the INL from it.
BOE

Title: Re: Getting the INL from the DNL of an ADC
Post by loose-electron on Mar 18^th, 2011, 11:44am

I know that this answer is in a dozen textbooks and on the internet as well.

Of course you can! the entire static response (Not dynamic) can be derived from the DNL, after all the voltage of each and every LSB step is in the data.

Title: Re: Getting the INL from the DNL of an ADC
Post by carlgrace on Apr 1^st, 2011, 9:00am

Just add up the different DNL values to get your INL. It is just the running sum (hence the term "integral").

Carl

Title: Re: Getting the INL from the DNL of an ADC
Post by Vladislav D on Apr 16^th, 2011, 5:34am

I have a doubt. INL is an absolute shift of a transfer from the ideal one. DNL measures the ratio between an ideal and real step. So, if the whole transfer function is shifted, INL will change but DNL remains the same. I don't see correlation between INL and DNL in this case.

Title: Re: Getting the INL from the DNL of an ADC
Post by boe on Apr 18^th, 2011, 10:31am

Vladislav D wrote on Apr 16^th, 2011, 5:34am:

I have a doubt. INL is an absolute shift of a transfer from the ideal one.

According to common convention INL (and DNL as well) is measured against the best-fit line, so offset and gain-error do not affect INL.

Offset is not a non-linearity, so it should not be part of INL, whatever definition you use.

BOE

Title: Re: Getting the INL from the DNL of an ADC
Post by Vladislav D on Apr 18^th, 2011, 1:05pm

Hi Boe,

boe wrote on Apr 18^th, 2011, 10:31am:

According to common convention INL (and DNL as well) is measured against the best-fit line, so offset and gain-error do not affect INL.

I don't know about common convention but this is written in the book of ADI about the end point method of measurement INL:
"This is the most useful integral
linearity measurement for measurement and control applications of data converters (since
error budgets depend on deviation from the ideal transfer characteristic, not from some
arbitrary "best fit"), and is the one normally adopted by Analog Devices, Inc"

Quote:

Offset is not a non-linearity, so it should not be part of INL, whatever definition you use.

Offset is not non-linearity. However, offset causes non-linearity as well as second harmonic causes DC offset. Isn't it?

Title: Re: Getting the INL from the DNL of an ADC
Post by boe on May 9^th, 2011, 10:48am

Vladislav_D,
Sorry for the late reply.

Vladislav D wrote on Apr 18^th, 2011, 1:05pm:

Offset is not non-linearity. However, offset causes non-linearity ...

No, but if you have a non-linear system, offset makes your non-linearity issues worse.

Quote:

... as well as second harmonic causes DC offset.

if an AC signal is present (in a single-ended circuit).
B O E

Title: Re: Getting the INL from the DNL of an ADC
Post by carlgrace on May 14^th, 2011, 2:53pm

This is a situation where the application matters for the definition. A lot of ADCs are used in situations (e.g. imaging, communications) where the absolute deviation from ideal is not important. In those cases best-fit is the way to go.

To answer this question definitively, we need to specify the application. Otherwise, we are just chasing our tails.

Title: Re: Getting the INL from the DNL of an ADC
Post by boe on May 17^th, 2011, 9:25am

Carlgrace,
I agree.

Vladislav D,
I found http://www.analog.com/library/analogdialogue/archives/43-09/EDCh%205%20sampling%20theory.pdf. Do you mean 1st paragraph on p. 5.15 (p. 17 of PDF)?
An offset shift will not affect the INL. Cf. Fig. 5.16 on previous page. If you shift transfer characteristics, you also shift dotted line, keeping the linearity error constant.

B O E

Title: Re: Getting the INL from the DNL of an ADC
Post by loose-electron on May 18^th, 2011, 6:05pm

carlgrace wrote on May 14^th, 2011, 2:53pm:

Well stated!