Measuring incorrect voltage by multi meter. What could be reason

Thread Starter

Vindhyachal Takniki

Joined Nov 3, 2014
535
1. I was measuring voltage of thermopile with Fluke multimeter 15B+. Attached is my circuit. Its party incomplete shown but in actual its complete in hardware.

2. When I measure voltage by only thermopile in hand and not noumted on board, only connect probes of multimeter, I can measure voltage as per its mV vs Celsius graph.

3. When I connect it to board, and don't give 3.3V supply to board and connect probes across J1 connector, then also correct voltage read.

4.But when I give 3.3V to circuit, and connect probes across J1 connector, ideally it should be the same, because I am still measuring across the thermopile only,
but its very off than actual values, which I was reading without giving power. Why is that so?

5. is it high values 1M resistor connected across? But still y probes are connected across thermopile NO?

Earlier when I was reading 10mV, not I am reading around 200mV.


1592406569192.png
 

Irving

Joined Jan 30, 2016
820
The purpose of the 1Mohm resistors is to bias the ADC inputs so that the differential voltage between Vin1N and Vin1P is measured in the middle of the common mode voltage range of the ADC to minimise errors.

The current flowing through the thermocouple due to the 1Mohm resistors is about 1.6uA, so there will be a voltage across the thermocouple dependent on the length of the wires to it. The voltage you read in-situ is no longer just the thermocouple voltage.

What type of thermocouple is it? (link to datasheet?)
 

Thread Starter

Vindhyachal Takniki

Joined Nov 3, 2014
535
I am using this thermopile:
Now i am able to measure the thermopile voltage(verified by multimeter also, same reading) & thermistor votage and then able to convert thermistor voltage into celsisus also.

Now I need to convert it into the object temperature. Any mathematical equatin where I can direclty put these values and get the object temperature? Processing not an issue, since I have send both thermopile and thermistor values to PC, there made a small software. so rather than going by array search better to go with equation.



1592657692641.png



I downlaoded TE connectivity sheet, they have this one equation(attached also)
1592657709146.png

but emisstivity of object also varies, how to keep that into accoount?

Most of places I have searched they all do by array search, is there equation method also?
 

Attachments

Thread Starter

Vindhyachal Takniki

Joined Nov 3, 2014
535
I am able to measure thermopile(Vtp) mV & NTC(Celsius) temperature correctly as black body reference temperature when measured display correct results.
Took thermopile reading and NTC and did curve fit value.


Now when I shifted to human body measurement, results are wrong, due to emmisivity. I took e= 0.95 & 0.97 values and divided Vtp by it & then tried curve fit, but values are all wrong. Can someone suggest what is right way to shift to human body?

1. Currently only dividing Vtp by e. Is it ok?
2. Or need to divide NTC also by e?
3. Or divide final temperature by e?
4.What is right method?
 

MrChips

Joined Oct 2, 2009
21,374
Here are my comments and suggestions.

#1 - Put the 1MΩ bias resistors on one leg of the thermopile, not both.
#2 - You should be using a differential instrument amplifier instead.
#3 - Where is your thermistor circuit?
 

Thread Starter

Vindhyachal Takniki

Joined Nov 3, 2014
535
the measurement have been sorted and circuit is working. Able to measure the surface temperture correctly to 0.1C after calibration no issues

When shift to human body then issues happens, dont know what changes have to done mathematciallyin code, this is what i am aksing
 

jpanhalt

Joined Jan 18, 2008
9,784
this has been sorted, we measure very accuratelty now. issue comes when we shift from black body to humna body,
Emission from your black body is relatively uniform; the human body is not:
1597143349979.png

Your thermopile measures temperature across its surface. With a uniform field (like your black body), I would expect the temperature observed to be pretty accurate. For a human, it could depend on a lot of variables.
 
Are you aware that one end of the thermistor is connected to the metal case of the thermopile? Touching the case could affect your readings, depending on how it is connected in your circuit.
Keith
 
Top