I would like to highlight a problem in Q2687RD.
We were using ADCs and we noted that for zero input voltage the measured value was not zero.
We asked distributors to check with Sierra if this could be a calibration problem.
After the factory finished the tests, they didn’t found any issue, because the value was inside the tolerance.
Infact reading the ADC specs in section 5.9.2 of the
manual it is stated that the 10bit ADC has
This means that we are using a 6bit ADC not a 10bit!!
I think that Sierra should change its manuals, brochures, datasheets,ecc… to inform correctly on the real performance of the ADC.
Please look at this link to be clear what is the difference between resolution and accuracy in ADC.
However, if you have a 50mV accuracy, this means that the ADC can give you a “real” LSB of 50mV.
Thus it’s like measuring the signal with a 6bits equivalent ADC.
50mV = 2V / (2^n) where n is the equivalent number of bits.
solving for n you get n= 5.3 bits…
You can have a 10bit resoution but the latest 4 bits are useless because you don’t have the accuracy to read it.
And a major contribution to the overall measurement accuracy is the external analogue circuitry - if that is noisy and/or “inaccurate”, then you won’t get good results from the ADC
That’s obvious, but I would like also to mention that the module doesn’t have an analog GND pin, which usually is provided to improve analog performances.
We have an external resistor divider with 1% resistors, but the tests are made without the resistors, just measuring the plain 0-2V range.
Please also consider that a 70mV accuracy on a 2V range correspond to a 3.5% error (thus external resistors don’t impact)