Temperature resolution with LM34CAZ and U3LV | LabJack

# Temperature resolution with LM34CAZ and U3LV

4 posts / 0 new
Kalman8
Temperature resolution with LM34CAZ and U3LV

What is the maximum accurate resolution that one can expect when using the U3LV with an LM34CAZ temperature sensor.

Assuming that we do a multi point calibration on the LM34CAZ sensor to get it's accuracy down to 0.1F, what are the limitations of the U3? The U3 literature states "using the 0-2.44 volt single-ended input range, resulting in a voltage resolution of about 2.44/4096 = 596 µV". The LM34CAZ has a 10 mV/°F output voltage with a maximum output of 2.3 volts, that almost perfectly spans the U3's input voltage of 0-2.44 volts. A change of 0.1F will result in a change of 0.001 volts (1 millivolt) on the output pin of the LM34. Since the voltage resolution of the U3 is 596 µV, it seems that it is just withing possibility to get a 0.01F resolution. Is this correct? Will this resolution be accurate as far as the U3's limitations?

LabJack Support
Consider resolution vs.

Consider resolution vs. accuracy:

https://labjack.com/support/app-notes/resolution-and-accuracy

The resolution of the U3 on the normal low-voltage 0-2.4V range is about 0.6 mV, so the resolution of your temperature readings will be about 0.06F.

For accuracy I would say the limitation here is the linearity of the U3 which is specified as +/-0.05% full-span, or about 1.2 mV, which suggests that with your own calibration you can wind up with a temperature accuracy of 0.12F.

https://labjack.com/support/datasheets/u3/appendix-a

Kalman8
OK. Now suppose that we

OK. Now suppose that we account for the non-linearity of the U3. If one was to calibrate the entire assembly, LM34CAZ while connected to the U3, against a reference thermometer, would that then give an accuracy of 0.06F?

Are the non-linearities the same for all channels or can the non-linearity vary from channel to channel?

LabJack Support
All low voltage channels have

All low voltage channels have the same errors.  They have no signal conditioning before being multiplexed.

To do a calibration that accounts for the possible non-linearity of the U3, you would need lots of points.  We don't specify how many, but I would say 1 out of 4 counts, meaning that if you were going to do this across the entire 4096 count range (~ 0-2.4V), you would need 1024 evenly spaced calibration points.