View Full Version : A/D advice needed
01-25-2008, 12:10 AM
What benefits would a 16 bit a/d converter have over an 8 bit a/d converter that can be adjusted; i.e. ADC0831 with variable Vref and Vin(-).
I need to use an a/d with my propeller and I am leaning on the ADC0831 and by PWM Vin(-) and Vref I can measure a small voltage. I do need very precise measurement though so this is why I am asking. Otherwise i wanted to implement a BB8341E
You'll need to be very careful with your PCB layout and make sure that the supply is very clean to get the most out of the 16-bit device.
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
01-25-2008, 04:27 AM
... and I should add that more than TRUE 12 bits can be accomplished by the professional only
This could be a very interesting thread if you would be specific about the accuracy issue. You have deSilva responding ... that's at least one half of a good engineering debate.
As a non-engineer, my suspicion is that whether you can use the 0831 depends upon the actual range of values and the precision you need... these are both hard numbers...without these numbers, your question can't be answered.
12 bits is a lot of precision...
8 bits is an itty bit of precision..
but if you need precise numbers in a narrow range an itty bit of precision can be sometimes be made to work just fine.
01-25-2008, 04:31 PM
The most important thing is to understand noise. Where does it come from? What is its spectrum? It is time-variable? Do you know specific causes?
Noise is considered to be random, but in many cases is not...
Much of "noise" is generated by your own mis-choice of components: Carbon resistors! Use of high gain amplifiers that are nor "low noise" rated,...
Good ADC needs the full understanding of physics, not only of abstract electronic components http://forums.parallax.com/images/smilies/smile.gif
01-25-2008, 10:48 PM
Well my application is such that I need to read a temperature sensor that varies from-40°F to 302°F spanning from 2volts to 4.5volts.
It must be accurate and has been so far. The noise has not been an issue at the moment. I am not employing an op-amp and the resistors are 1% SMD1206 Ruthenium Oxide. I have put into service a 16bit a/d converter that is working well. I just don’t like the price and availability “even in large quantities i.e. 1000pcs it is costly” and thought I could get opinions on using an ADC0831.
My idea was to replace the 16bit a/d and use a 8bit a/d changing my program to zoom in to a particular voltage.
Post Edited (grasshopper) : 1/25/2008 3:58:27 PM GMT
01-26-2008, 02:52 AM
I think the main limitation is the accuracy of the temperature sensor as such.
(a) They must be calibrated in the first case
(b) even after that I am not sure they have readings better +/- 0.5 degrees wicht will be 1/600 = around 9 bits.
(c) When you now sample 16 bits = 60 µV), there MUST have been noise; the question is to what amount?
I don't see why it wouldn't work... and as long as you have enough pins available, you could put together a very direct implementation using RC logic to segment your dynamic range. Basically you'd be saying...
in psuedo-psuedo code: "is it above (below) x, if so then is it above(below) x1, if not check in this range."
With 8 bits you get 2.5/256 V maximum resolution... adjust that by some number (?2) to allow for range noise and you then know how many "decades" you need to define... how many pins you need without anything but RC logic.
Post Edited (rjo_) : 2/1/2008 1:49:06 PM GMT