Just to discuss on the subject: There's a guy on other thread that takes for granted that you can do this (having a 16 bits ADC with two 8 bits ADC) by using "potentiometers" and then, mixing the two octets by means of the program to get a 16 bits ADC.
Assuming that he is aware of the problems on managing an analog signal that needs 16 bit (this is another discussion): I've trying to guess how is he going to connect the "potentiometers" to both the signal and the analog inputs (I've asked him but he doesn't answer: he bypasses me and preferes to discuss about how to mix -shifting bits and so on- the famous two octets to get the 16 bits one with another forum guy). I don't think you (he) can do it. Can he?
Accuracy specification for 16 bit ADC: +/-0.0007%
Accuracy specification for 8-bit ADC: +/- 0.2%
Any attempt to use 2 8 bit ADCs will lead to 8 bits of meaningful value
and 8 bits of noise (or perhaps a little better than that if the 8 bit ADC
is really a 10 bit ADC with 2 bits omitted).
Of course all this is moot because for successive-approximation you use
a DAC + comparator.
So you can build a 16 bit ADC using a 16 bit DAC and comparator, but again
2 8 bit DACs don't make a 16 bit DAC...
Let's be clear here. In the other thread, I believe we were discussing a "coarse" pot + "fine" pot setting. Both pots are measured with an 8 bit DAC, and the "coarse" pot value used for the upper 8 bits and the "fine" pot used for the lower 8 bits to create a 16 bit number. This works fine.
What you can't do is read an unknown voltage with 2 8 bit pots and combine the results.
Let's be clear here. In the other thread, I believe we were discussing a "coarse" pot + "fine" pot setting. Both pots are measured with an 8 bit DAC, and the "coarse" pot value used for the upper 8 bits and the "fine" pot used for the lower 8 bits to create a 16 bit number. This works fine.
You don't, they are totally independent. Each is hooked up from Vdd to ground with the wiper going to the analog input.
If the coarse pot is set to generate a value of 0xF0, then you can get the fine pot to generate numbers from 0x00 to 0xFF so you can get the composite 16 bit number of 0xF000 to 0xF0FF.
You're right. The guy wants to have a 16 bit word that generates by means of two pots (not a 16 bit resolution reading of a external analogue signal). Now it all makes sense.
Thanks.
(Any case, I think he still needs to rethink the whole thing: a step -mechanical uncertainty, temperature changes, etcetera- in the coarse pot will make a change of 1/256 th of the whole range. Am I right?)
("See it!. That's what I said") 