In Reply to: to Dean: What's "watching short-term trends",for example? posted by sarcophile on 06/04/03 at 10:55 PM:
"Actual" measurement is the value of the measured quantity. Suppose that you have a voltage standard with an output of exactly 5.000000 volts. If I have a 5-1/2 digit digital multimeter that has an accuracy of ±1% (a ridiculous set of numbers that I made up to illustrate a point) to measure this voltage and the result will display as 5.0302 volts. In this case, the meter with its 1% accuracy can only give me a guaranteed true reading within ±0.05 volts. That means that the circuitry cannot support the last two digits of the display, for the display can resolve down to 0.0001 volt while the measuring circuits within the meter can only provide an accuracy of 0.05 volts. Those last two digits may change quite a bit over several minutes of measurment even though the original voltage being measured doesn't change at all.
However, for a minute at a time, those last two digits may not change. Although I can't use them to describe the actual voltage being measured, I can use them to see if a less-accurate voltage is changing. I could instead connect the meter to a simple LM340T-5 regulator output (±0.1 volt) and monitor the voltage. My reading of 5.0824 volts tells me that the output voltage of the regulator is correct, although I can only insure that the votlage is 5.08 volts and not the full precision of the 5.0824 volt reading. However, if I know that my meter is stable enough that the display and system will keep the same reading over a full minute and I see that the display is changing, slowly going down, I can assume that my regulator output is slowly decreasing.
So in this way, I can use the meter to make less-precise readings but more precise indicators of small changes, the "short term trends".
"Accuracy" is how close to the actual voltage my meter is reading.
"Precision" is to what degree I can read this voltage.
My meter can be very precise, but if it hasn't been recently calibrated, it may not be very accurate.
So, you don't need an A/D converter of that precision for a less-precise temperature converter. However, you may have to use that A/D converter because it's also being used for other more precise jobs. You just have to realize that you can't use the full precision of the converter for the temperature readings. The output of the converter is no more accurate than the input.