Hi,

I need to connect the ZTP-135SR temperature sensor to a microprocessor

to measure the temperature of a surface without contact, holding the sensor at a fixed distance ...

attached is a diagram I found ...







could you tell me the values of the components considering Vin = 5V?

Could the mass of C2 be that of the rest of the circuit?

thank you.

 

Interesting device, but precious little application info on the web. Datasheet and 'training guide' attached.

Now to your problem... the thermopile outputs approximately -3mV at 0°C and +10.5mV at 100°C. Assuming a single ended ADC at 3.3v you need to amplify the voltage of the pile by about 200 - 240 - this is probably done with a small fixed amplification in one stage (say x10) and a following stage (x24) To get a resolution of around 2°C you need to reliably detect a 75uV change allowing for noise and opamp offset. That means you need to start by finding an opamp - or more likely an instrumentation amp - for the first stage with a low noise figure and sub-10uV offset. The device has a noise figure of 0.5 √nW/Hz so your amp will need to be in the same or near order of magnitude. This isn't going to be a common hobby op-amp on a breadboard if you expect any degree of accuracy or repeatability - PCB layout will matter too.

I don't have time right now to run the numbers...

 

Thank you...

let's say I'm a programmer (atmel, microchip, etc ...)

about electronics I am thirsty to learn continuously ... and therefore I ask for advice from experts ...

as far as my application is concerned, as previously mentioned,

I have to read the temperature variation of a non-contact ceramic surface by placing the sensor at a distance of a few millimeters (type 5mm) ...

important thing: read the temperature change quickly and accurately ...

concerns the first scheme ...

are the values of the components ok like this?



as for C2 I will make a 2,5V divider ... what value should the capacitor have?

 

Here's my take on the circuit. I don't like resistive dividers for references. I know it makes it ratiometric to the supply voltage but I prefer a proper reference. Its pretty linear from 0 to 100C though the thermistor isn't a great match for the thermopile curve. By limiting it to 1.25v though its not too bad. Obviously you'd remove any non-linearity by calibrating it and providing a calibration table in the microcontroller. Note this circuit will not drive a very long cable so the microcontroller needs to be fairly close.

I took out the capacitor across the thermopile; not convinced its needed and it'll slow the response down - not that thermopiles are very fast. You'll need decoupling capacitors on the opamp +ve supply pin(s), something like 10uF in parallel with 100nF will suffice. U2 is there to buffer the thermistor output - if you dont need that output you can take it out.

 

Here's my take on the circuit. I don't like resistive dividers for references. I know it makes it ratiometric to the supply voltage but I prefer a proper reference. Its pretty linear from 0 to 100C though the thermistor isn't a great match for the thermopile curve. By limiting it to 1.25v though its not too bad. Obviously you'd remove any non-linearity by calibrating it and providing a calibration table in the microcontroller. Note this circuit will not drive a very long cable so the microcontroller needs to be fairly close.

View attachment 273700

Nice, could you provide the .asc file for that? I want to look at how you implemented a changing Resistor value/Output voltage graph using LTSpice. The only thing I have used thus far on that tool is voltage/current vs time.

 

You might want to change C2 to 10n else the step response is measured in eons... with 10n theres a 37mS response time excluding the thermopiles own response time which seems to be 25mS.

 

Here you go...

 

Nice, could you provide the .asc file for that? I want to look at how you implemented a changing Resistor value/Output voltage graph using LTSpice. The only thing I have used thus far on that tool is voltage/current vs time.

I used an LTSpice table construct. The form is:

.param <param> = table(index, key1, value1, key2, value2...)

Index is your input, in this case 'temp' and then a series of keys and values here temp/resistance tuples, arranged in increasing key order. As I understand it an index between two keys will get an interpolated value.

So here I've just replicated the thermistor value sequence from the data sheet for temp values 0, 10, 20, etc.

 

The voltage must be 200 - 240 it cannot have more, and it must not be less as well as it is 220

@lewis40 what effect would exceeding these limits for a short period of time have?

 

I used an LTSpice table construct. The form is:

.param <param> = table(index, key1, value1, key2, value2...)

Index is your input, in this case 'temp' and then a series of keys and values here temp/resistance tuples, arranged in increasing key order. As I understand it an index between two keys will get an interpolated value.

So here I've just replicated the thermistor value sequence from the data sheet for temp values 0, 10, 20, etc.

Thanks! When I get some time, I will play around with .param/.step commands. I did find this that talks about it more:

https://www.analog.com/en/technical...tep-command-to-perform-repeated-analysis.html