Skin-Temperature-Sensor: sensor-type & circuit design?

Thread Starter


Joined Oct 20, 2008
Hello everybody,

I am quite new in all this electronics and circuits stuff, but I have to deal with it somehow, as I need some basics for a studies related project. Unfortunately I do not have a clue, so hopefully I can bother you with some of my questions ...

My current task is to find a circuit design for measuring skin temperature. I already developed some ideas on that, but I am not sure, if I am doing right.

So this is, what I want to achieve:

What temp range?
"normal" skin-temperature is 32/33°C; Temperature Range in certain conditions can be between 31°C and 37°C
So range 31°C - 37°C is required.

Different medical devices use different values here. I think for me ±0.1°C is ok.

Response time?
Seconds should be ok.

direct skin contact - some mms

Other requirements?
- Sensor should be as small as possible (plan is to built the sensor into a mobile phone)
- Sensor should work with +5V

Do you recommend any sensor-device for this application? I thought about the LM35 ( Is this a good choice? Or can you recommend anything better, smaller, ...?

Further, I have this circuit diagram - which I am not sure is a good one (see attached file skintemperature.jpg) Would be happy for any help with this circuit diagram (as I do not know, if this could work for me): does this circuit diagram make any sense for what I want to achieve?
Sorry, this sounds like a stupid question, but this is the first time that I try to design such stuff ...

Thank you very very much for all of your efforts in advance.

Hope to get some answers here, that might help me.



Joined Oct 20, 2008
The LM35CAZ is a very good device with a nice 10mV/degC output. Your requirement should give an output range of 310mV to 370mV.
If you use a stable 10x amplifier then you will get 3.1 to 3.7V without too much noise. Although, if you are inputting to an ADC, you could keep the component count down by a direct feed to the ADC.

Thread Starter


Joined Oct 20, 2008
Thank you very much for your fast reply and your recommendation to go with the LM35CAZ. Good to hear, that my initial choice of the LM35 is not too bad.

Can you recommend any sable 10x amplifier?

Sorry, but I did not understand much about the ADC-part. I planned to connect the sensor to an ARDUINO-board, which than would do the analog-digital conversion. Or am I wrong?

Thank you again for all your time and efforts!

Thread Starter


Joined Oct 20, 2008
Oh. I just forgot. Would this work with the circuit diagram that I posted earlier? Sorry, to ask, but this is the first time a do such a circuit design.


Joined Apr 20, 2004
Your op amp has a gain of 5 as drawn. Reduce the 500K to 100K, and the 100K to 10K and you have g gain of 10. You could go as low as 20K and 2K.

Thread Starter


Joined Oct 20, 2008
Thank you very much for the idea with the IR sensors. Actually they seem to fit my needs even better than the LM35, as the LM35 has "only" a 0.25°C accuracy.
Unfortunately I do not understand much about the IR-sensor you suggested. Any ideas how to measure absolute temperature?
Maybe you can help me out with some schematics?

I really appreciate all your help on my (long) way in understanding electronics ;)



Joined Apr 20, 2004
The voltage out of the thermopile is relative to the difference between that of the thermopile itself and the isolated wafer. The buried thermistor has to be read to know the thermopile temp before the incident temperature difference can be converted to an absolute temperature.

The thermistor is non-linear in its response, so compensating to produce an absolutely correct result is challenging with analog circuitry. Reading both with a microprocessor and doing the calculations is pretty simple. Notice that the Metaxis suport chip needs a serial clock to read out its values, so it is also predicated on some microcontroller.

The ARDUINO should be able to handle the thermopile, assuming the A to D converter has enough resolution.


Joined Dec 27, 2007
No problem,

I would use their interface IC with the sensor, then read it into your ADC as beenthere suggests. You would apply the equation given in the datasheet to find the temperature.

I don't really know what kind of precision you can get with this sensor. The LM35's +/-0.25C is actually pretty good! You would be hard pressed to get better than this with the thermopile. The advantages, as I mentioned, is that you can creatively seal this into a cell-phone through an IR window. Also, with less thermal mass, you can measure faster. Also, it is less prone to 'parasitic' heat, or heat that you do not want to measure, since it is directional.



Joined Aug 30, 2009

I've found this thread to be helpful for my own project involving a thermopile and my Arduino. However, after looking at the PDF for the thermopile support chip you described, it makes incomplete reference to external components and circuitry. Without understanding this part, I'm back to an impasse.

The PDF does show a simple Op Amp circuit, using Melexis' AD8628, which is similar to the one I use but omits the temperature compensation circuit, it seems. It labels "Vref" for setting an offset voltage of some sort, but does not give enough detail for this either.

Right now, I've got a seemingly working circuit using an op amp, increasing the tiny voltage output of my thermopile, and fed into one of my arduino's ADC ports.

I've found and adapted an Arduino sketch that sends the data to the serial port; my monitor shows rapidly scrolling C and F readings.

My eventual challenge is to find a Processing sketch that will convert this data to a dynamic bar, or some other graph type. I've spent a lot of time researching that part already but have not found one suitable for adaptation to my project.