Increase A/D resolution?

Georg Prinz · Joined: 07 Jan 2004 Posts: 22 Location: Frankfurt, Germany

Hi folks,

is it possible to increase the 10-bit resolution of the pic A/D-converter by oversampling the signal? If yes, what is the mathem. background or a hint where to find it and does anybody have a code snippet for CCS-compiler for 18F4620.
What are the limits?

Thanks in advance

George
_________________
dl 2 kp

Humberto · Posted: Tue Sep 13, 2005 5:50 am

The built in A to D converter resolution in PIC microcontrollers is impossed by the
hardware architecture, not for the sample rate. Resolution indicates the number of
discrete values that an A to D Converter can produce.
Using microcontrollers with built-in AD Converters, the user only determine the
Start of Conversion. If enabled, the End of Conversion can set an interrupt flag,
meanwhile the user can�t do anything to modify this process.

Some external AD converter devices that use Slope-Integrating Type, perform the
conversion process by feeding the unknow signal to an analog integrator for a fixed
period of time. In this type of converters an external counter is used to measure
the time of a know signal against the time of the unknow signal. Increasing the
clock frequency of the external counter increase the resolution.

Fortunatelly, ranging from 6 to 24 Bits actually you get a lot of very good external AD
converters that can suit your needs without being involved in such obscure tasks.

Humberto

Georg Prinz · Joined: 07 Jan 2004 Posts: 22 Location: Frankfurt, Germany

Ok Humberto,

thank you for your reply. This is what I suggested, too. The reason was that I found a software at the microchip forum (18ADOver), which says "Oversampling and averaging technique used to improve the S/N and number of useful bits in an A/D convertrer". It sounds that they pretend to increase the resolution by simple oversampling. Averaging is clear and accepted, but increasing the resolution is nonsense. Or they mean that they increase the useful bits within the range of 10bit resolution of the A/D converter.

Ciao

George

_________________
dl 2 kp

Mark · Joined: 07 Sep 2003 Posts: 2838 Location: Atlanta, GA

Ttelmah · Guest

How effective it is, depends on the nature of 'errors' in the ADC. If (for instance), your accuracy is not really 10bit, because of noise in the system, and this noise is random, then oversampling allows the full accuracy to be achieved. Similarly, if the ADC is deterministic, and for a signal sitting half way between two values, returns the two possible readings in a random pattern, that is symmetrical in terms of how often the two values are given, the results can be averaged to give a genuine extension of the accuracy. However you have to remember that linearity errors in the ADC itself will still remain at their original values. Hence though you may get legitimate readings, corresponding to an extra bit, the linearity of the converter, and accuracy of the reference source will remain unchanged. Also the time that needs to be allowed for the sample capacitor to be charged needs to be extended, to handle the potential extra resolution, and response speed will be proportionately lower.
This approach will not give you a 14bit AD, from the PIC hardware, but in some cases can be used to maximise the resolution available.
The standard averaging algorithm publshed in the past can be adapted to test this:

Humberto · Posted: Tue Sep 13, 2005 10:03 am

___________________________________________________________
Mark,

I don�t understand why you "remark" all my post. Did I wrote something that is not true nor real or senseless?
After that you didn�t write anything and redirect to an inexistent link and to an example coded in Forth (???)

Kindly, pls post your opinion and keep my posts.
___________________________________________________________

Regarding the oversampling method to improve the resolution in AD converters, it is a technique that can be used only in special cases, it is not applicable to all kind of signals.

Following I pasted some conclusions about this subject in chipcenter tech app note:
http://archive.chipcenter.com/TestandMeasurement/tn035.html
by Leonard Staller, Applications Engineer,
Cygnal Integrated Products, Inc., Austin, Texas

"Oversampling and averaging will work only if the A/D converter noise can be approximated as white noise. If the input signal changes randomly from sample to sample by amounts (amplitude) comparable to the code size (1 LSB), and the input signal has an equal probability of being anywhere between two adjacent codes, then the noise can be modeled as approximating white noise.

What is white noise? It's noise that's characterized as having a uniform power spectral density over a frequency band of interest. When noise can be approximated as white noise, then oversampling and averaging can improve the SNR and increase the effective resolution of your data.

If the overall noise isn't stationary (some systems may have some correlation due to feedback), then oversampling and averaging may not be effective. Additionally, if the quantization noise is comparable to sources of white noise (thermal and shot noise are small compared to quantization noise), then oversampling and averaging may not be effective."

Humberto

rwyoung · Joined: 12 Nov 2003 Posts: 563 Location: Lawrence, KS USA

RJ is right. The key is that the errors be uniformly distributed (white noise). Then the averaging of multiple samples will result is an apparent increase in resolution. If the ADC is of such poor quality that its quantiziation noise overwhelms external noise sources then you will not get the sqrt(num_samples_averaged) improvement factor. The resulting data will be colored by the distribution of the quantization errors.

The terms "dither" and "dithering" apply and might make good Google search terms. Or you might end up with something more related adult entertainment Shocked

, Google is funny that way.
_________________
Rob Young
The Screw-Up Fairy may just visit you but he has crashed on my couch for the last month!

Mark · Joined: 07 Sep 2003 Posts: 2838 Location: Atlanta, GA

Neutone · Joined: 08 Sep 2003 Posts: 839 Location: Houston

Georg Prinz · Joined: 07 Jan 2004 Posts: 22 Location: Frankfurt, Germany

Thank you all for the valuable explanation and references, especially from Humberto and Neutone. It helped me a lot to revise my opinion and to make the right decision for my design.
Very Happy
Unfortunately I am not allowed to access the url //home.houston.... because members only are accepted.

Best Regards
Georg
_________________
dl 2 kp