In one of our designs we are measuring the RTD temperature using similar design architecture as that of CN-0287.
We have used ADG704 in place of ADG738 since we had only GPIOs to control the switch. (ADG738 requires 3 wire serial interface).
The issue we are facing is that the we see variations in the ADC readings over a long run when there is switching involved.
i.e. If multiple channels are enabled at a time, then the switch has to change the position so as to connect the current source to the respective input RTD channel.
If we keep this setup for a long run, we observe a variation of around 0.7C in the measured temperature. I have attached a plot of temperature readings with the above setup.
If we enable only one channel, then there is no requirement of switching the current source. We have observed a very good accuracy with this setup.
Further to isolate the issue between analog switch & filters present on the input line, we turned OFF/ON the current source with the switch position being constant. We didn't face any long term accuracy issues with this setup also.
Above tests confirm that the issue is with the analog switch ADG704. We are unable to find the reason behind this with the datasheet.
We have a delay of around 100mS, between switching current source & reading the ADC input. Our system does't allow us to increase this delay further.
Can you please suggest the reason behind this issue ? Also, is there any alternate 1:4 switch which gives us better performance?
Looking forward to your response.
Suresha N S