Pages: [1]   Go Down
Author Topic: Question re: Accuracy of Reference re: Accuracy of ADC  (Read 242 times)
0 Members and 1 Guest are viewing this topic.
Peoples Republic of Cantabrigia
Offline Offline
God Member
*****
Karma: 6
Posts: 722
Arduino happiness
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Hi guys,

I have a general question regarding the accuracy of voltage references as they pertain to ADCs. Specifically, I was under the impression that the accuracy of the reference was a key component in providing accurate and consistent ADC results.

But, a specification for a analog front end from MCP, the MCP3911 or MCP3910 has me a bit confused. On the one hand, MCP claims that the MCP3910 can be used for power measurements that exceed 0.1% in accuracy and that the references that are built into these chips have very low drift (<10PPM/*C) but the actual accuracy of the reference is about +/- 2%.

Is the explanation that while the reference is "only" 2% accurate, it is very stable in terms of its output? It certainly seems very stable re: ambient temperature, for example. Thus, the error of the reference is so consistent that it can be calibrated out of existence?
Logged

Left Coast, CA (USA)
Offline Offline
Brattain Member
*****
Karma: 361
Posts: 17294
Measurement changes behavior
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
Thus, the error of the reference is so consistent that it can be calibrated out of existence?

 Yes. If the ADC exibits good 'repeatablity' then any constant variation of the reference can be compensated in software, either initially or as part of a automatic or manual calibration function.

Logged

Peoples Republic of Cantabrigia
Offline Offline
God Member
*****
Karma: 6
Posts: 722
Arduino happiness
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Hi Retrolefty and thanks for the reply!

In other words, unless a reference was more stable (i.e. better than 9PPM/C) or so, then there would be little benefit in employing a high-accuracy external reference like the ADR3412?

That is, the high-accuracy reference would be mostly useful to reduce calibration requirements since its initial accuracy would be much better than that of the voltage reference in the MCP3910?
Logged

Left Coast, CA (USA)
Offline Offline
Brattain Member
*****
Karma: 361
Posts: 17294
Measurement changes behavior
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
That is, the high-accuracy reference would be mostly useful to reduce calibration requirements since its initial accuracy would be much better than that of the voltage reference in the MCP3910?

I would use the word more-accuracy rather the high-accuracy, as high is subjective and relative to some unstated standard. In a typical ADC converter there are two most important attributes that dominate it's 'accuracy', bits of resolution (step size) and reference voltage accuracy. And lastly even with 'high-accuracy' ADC one may very well still require (or want) a automatic or manual calibration function, as the signal being measured from some sensor may not have a 'perfect zero' and require at least an offset correction. And if you have a calibration function anyway, then absolute voltage reference is not as important as long as it has good stability specs.

 True accuracy is a very complex subject if and when wants to dive down into it, in fact there is a whole field called Metrology that deals with the subject.  http://en.wikipedia.org/wiki/Metrology
 

« Last Edit: August 13, 2014, 09:16:17 pm by retrolefty » Logged

Pages: [1]   Go Up
Jump to: