AVR-Chat

Thanks to all for your help.

 

Based on the information below from David about the source impedance, I
changed the resistors for a 1K and 5K6 to lower the input impedance seen by
the ADC and now I get accurate readings from the system.

 

I do quite well with a lot of my projects but sometimes it just needs that
second or third eye to look over the design.

 

Cheers guys and have a good weekend.

 

Dave.

 

From: AVR-Chat@yahoogroups.com [mailto:AVR-Chat@yahoogroups.com] On Behalf
Of David VanHorn
Sent: 27 October 2012 01:12
To: AVR-Chat@yahoogroups.com
Subject: Re: [AVR-Chat] Apparent non-linear output from ADC

 

  

There are a number of possibilities.
10k is not the ideal source impedance, it is the maximum recommended
source impedance. Lower is better.
The I/O pins on most AVRs can have up to 1uA leakage to VCC or GND. (see
above)
The internal AREF voltages on the AVRs are not very accurate. See
device spec sheet, +/- 0.1V isn't unusual.
Clocking the converter too fast or too slowly can cause problems.
Noise on the analog input is a problem. Low pass filters are your
friend, as this is a sampling ADC. It takes a snapshot of the voltage
at a particular moment. You may also be aliasing in some ripple or
other noise by using a sampling rate that is very close to some noise
component.
Changing references and then converting before the ref has time to
stabilize can be a problem.
Not having bypassing on AREF, or any/all of the VCC/AVCC pins.
Not connecting ALL GND pins.
Clocking the chip too fast.
Using the low power crystal oscillator.

Measure your reference voltage and the input voltage to the chip at
the chip pin, and make sure they are what you think they should be.
It may be that the chip is telling you exactly what is going on. :)



[Non-text portions of this message have been removed]