LM35: Using the LM35CAZ as a cold junction sensor for a thermocouple

Greg,

How accurate is the temperature setpoint in your calibrated oven? Are you comparing against any other calibrated thermometer? I suspect the drift could be that the oven is not always exactly at the desired setpoint, but just slightly deviating from that as temperature changes.

Thanks

-Alex Thompson

We see ~0.43C variation when the oven is holding temperature but we pick up this variation in the LM35. We ran one test with a calibrated witness thermometer, attaching a screenshot of the data. But the witness thermometer was not next to the oven temperature thermometer or next to the lm35. 
we will run a more controlled test in the next few days with a more controlled setup.
  

Greg,

Thanks for running these tests. I also wanted to ask, are you monitoring for any noise on the LM35 supply line? If large voltage swings/spikes/noise is present on the voltage supply line, the voltage output will be affected. This is described in the Line Regulation spec:

Looking forward to your additional results.

Thank you

-Alex Thompson

we ran the oven test with a RTD connected to a keysight digital multimeter as our temperature witness. On Our assembled prototype units we measured the LM35 output with a digital multimeter. We saw a little over 1C temperature drift. 

We had a lm35 witness connected to our rtd thermometer outside of our prototype unit and measured the output and we saw less than 0.2C drift. 

can you think of anything that could cause the LM35 drift to be out of spec? 

Greg,

Please assist me in understanding these two graphs. 

From what I understand, the first graph is showing two different assemblies with LM35 in blue and orange compared to an RTD which is the black line. And the temperature is being stepped up? So as temperature is increasing, the temperature error (y-axis) of the LM35s is getting more and more negative. Or is the chamber being held at one temperature, and the LM35s are drifting up in their temp result. But if that is the case, why is the RTD line going up and why are the LM35 lines having those negative spikes? Are those from something other than temp changes?

For the 2nd graph, I see that temp is changing and the y-axis is temp error. But the LM35 compared to the RTD is changing less than 0.25C across temp, while the chamber is changing more dramatically. What this tells me is the source of the drift is the temperature chamber moving around, not the LM35 having a real drift problem.

Am I misreading these graphs?

Thanks

-Alex Thompson

for the first graph the Temperature is being stepped up in 5 degree increments then dwell for 2 hours. The black line is the RTD temperature. the orange and blue is the error between the RTD and the LM35 so (LM35 - RTD). As the temperature increases the LM35 has an increasing error compared to the RTD close to 1.2C. 

so yea on the second graph there is no real drift problem but this is just a lm35 and RTD on the same piece of metal. 

but in our system we have an LM35 assembled on a PCB and we are seeing drift that is out of the LM35 specification on the data sheet