I am currently using the AD7685 in my design with the following circuit layout (apologies for sloppy presentation!).
The problem i am having is that when a voltage of 5V is applied to the input i am getting an output of 61497, rather than the expected 65535 (or thereabouts). So a full scale input is not producing a full scale output. After this i noticed that all other inputs between 0-5V contained an "offset" error. The smaller the input voltage, the smaller the error.
Has anyone else had this problem or possibly know what my problem is or what stupid mistake i have made?
Did you follow the AD7685 layout guidelines for your circuit (placing reference decoupling capacitor close to the REF and GND pins, etc)? Please also note that as per the REF02 datasheet, "While the REF01/REF02/REF03 series of references are designed to function stably without any external components, connecting a 0.1 μF ceramic capacitor to the output is highly recommended to improve stability and filter out low level voltage noise."
What are you using for your input signal? What are the tolerances of your resistors? It may also introduce error to your reading.
Hi Karen thanks for the reply,
Yes i have followed both datasheets carefully and have applied all necessary decoupling. The input is provided by a simple voltage divider with a variable pot to adjust the voltage. The tolerance on each is 5%, however i have been probing the input to make sure the voltage is what i expect it to be and also check for any excessive noise.
Probing around other parts of the circuit i can't seem to see a significant amount of noise which could be causing this.
With noise however i would expect random changes in the read value, however my read value is consistently wrong (1000 readings resulted in 98%+ of them as the value i stated above).
Why did you choose the REF02 as your reference? Did you probe at its output as well? If it's output is off, your LSB size would be different and would result in a gain error.
Note that the AD7685 datasheet suggests buffering the reference output with AD8031 or AD8605. We also recommend using the ADR435.
I chose the REF02 quite simply because of its availability at a local distributor in a DIP package for prototyping. I have probed all possible outputs and all signals looks clean and correct.
I understand maybe a different reference setup would help, however with the current setup i have, it really does not warrant the amount of error i am seeing (as far as i can see).
Can you tell me what you're using to provide the 5V rail? I noticed from your schematic that you're not using an ADC driver for this prototype. The input of the AD7685 is a dynamic, capacitive load, and a typical bench-top supply may have trouble dealing with this.
We recommend using a low-noise, high-bandwidth amplifier in combination with a single-pole RC filter to drive the input of the AD7685. The amplifier provides a low output impedance source for the input, while the RC filter acts to stabilize the switching capacitive load.
For a more in depth description of how to drive the inputs of the AD7685, see the Typical Connection Diagram, Analog Inputs and Driver Amplifier Choice sections of its datasheet. These sections describe how the AD7685 input behaves during converter operation and which components to use to condition the input signal.