- Ask a related questionWhat is a related question?A related question is a question created from another question. When the related question is created, it will be automatically linked to the original question.
Hello,
For an application, I am looking for an adequate ADC and I would like to know which parameters I should consider and how I should interpret them to calculate the final error. I am measuring an NTC which full range voltage ranges from 65mV to 85mV and I have to detect changes of 20uV. This means that the final error in the ADC, after offset, gain error, noise etc cannot be higher than 20uV*Gain. I was thiniking about using a Wheatstone bridge (to "extract" the 20mV difference), and amplify and read this difference by using a sigma-delta ADC with gain (at least of 8 or higher) and 18 to 24bits resolution. Could you give me some guidance for the correct selection of the ADC or a model suggestion? I'm also a bit confused in the part of calculating the overall error. For example, if the offset of an ADC model is 30uV (higher than the 20uV maximum error), then should I discard this model? For instance, ADS8887IDRCT offset, according to datasheet, is up to 4mV.
I hope you can help me in that.
Best regards,
Hello Cyntia, thank you for all the info, it is very helpful, above all the Total Unadjusted Error Explained ~click here~. I would also like to read the part 1 of this article https://e2e.ti.com/blogs_/b/precisionhub/archive/2014/10/07/adc-accuracy-part-1-is-accuracy-different-from-resolution.aspx but for some reason the link doesn't work to me.
Regarding this article: https://www.ti.com/lit/an/sboa323a/sboa323a.pdf?ts=1652384359900&ref_url=https%253A%252F%252Fwww.ti.com%252Ftool%252FCIRCUIT060002&_ticdt=MTY1MjkwNTY2NHxsaXRhOTY4YzNkMGE5NjhjM2QwYTk2OGMzZDA2MGFiYjgzZjEyNzYwY2Q2fEdBMS4yLjgzMjU1MjU1MS4xNjIxOTYxODY3 is the current design that we are using, but I find a bit difficult to calculate the error. For an ADC I know that I have to take into account Voffset, gain error and INL, but for the operational amplifier I get a bit more lost.
Regarding giving more details about the sensor circuit, here I go:
We have an NTC, which bias current has to be around 50uA (to avoid selfheating), and also it is enabled just during 100ms every second to do a reading. It is enough with getting just one trustable reading every second (so I suppose that this would answer the question about sampling rate)
My intention is to create a Wheatstone bridge with that,
so that Vdd is REF7012QFKHT, with 1.25V output, 0.025% accuracy and 2ppm/ºC. R1 and R3, 20K, 0.01%, 10ppm/ºC, RX would be the NTC and R4 would be 1K 0.01% resistor.
I plan to use Vdd also as the reference voltage for the ADC
For the temperature accuracy that we need, we should detect, in the worst case, 0.9Ohm changes in the NTC (where NTC resistance is about 1107Ohm), which with a bias current of 50uA would be around a 45uV change, so that should be the final accuracy of the system.
I know that because of Vdd and R1 accuracy and tolerance, I also have a deviation, this is why I was targeting a final accuracy of 20uV instead of 45uV ini the ADC selected..
Because the total difference in the Wheatstone bridge would be 20mV (NTC value would change between 65mV and 85mV), I would need to amplify it, if possible in a factor of Vref/wheatstone range=0.9*1.024V/20mV=46 (0.9 for not using the full range of the ADC and using the most linear part of it).
For the gain, I would like to get an ADC with gain included instead of using an instrumentation amplifier.
Calibration could be an option, although it would be preferable not to do it. In case to do it, do you have a reference document for that? Do you think that in this case it would be possible to avoid the calibration, having a final accuracy so strict?
Thank you very much for all the assistance,
Best regards,
Esteban
