how do I connect an infrared temperature sensor to a pic?

[Kristian]

infrared temperature sensor

Hi.
Im working on a project where I need to incorporate an infrared temperature sensor (ZTP-135SR). The datasheet from general electrics is very spartane in information. On the sensor there are four wires - A thermistor, a thermopile and a ground wire for both the thermistor and the thermopile.
I have been searching everywhere I can think of to get information on how to connect the device. Im using a pic 16f873a with an ina128p instrumental amplifier.

As far as I can understand the thermistor is there to compensate for the ambient temperature, but how do I connect it?`

If i just connect the thermopile to the instrumental amplifier I get readings, but they vary too much over time to be ok.

If anyone out there has had experience with hooking up an infrared sensor to a pic processor I would very much apriciate any guidance you can give me.

Thanks in advance

Kristian

 

[FvM]

connecting ir sensor to amplifier

If i just connect the thermopile to the instrumental amplifier I get readings, but they vary too much over time to be ok.

Did you calculate the expected noise from sensor and amplifier noise?

The measurement bandwidth has to be carefully adjusted to your application,otherwise you see mostly noise. Part of the filtering can be achieved in software but some analog low-pass filtering should be provided. INA128 has basically a too high noise current for the said sensor. At 10 Hz, the amplifier noise current contribution is about double the sensor thermal noise. A very low noise FET amplifier or a lower noise current bipolar instrumentation amp, e.g. INA114 would be better suited.

But most likely, a large amount of the observed noise is caused by enviromental temperature fluctuations. A thermopile sensor must be shielded carefully against enviromental temperature changes.

 

[Kristian]

infrared temperature sensor +project

Hi FvM

Thanks a lot for your reply. I have not mesured the noise level from the amplifier or the sensor yet, I will try to do this as soon as possible. Also I will try to get hold of the ina 114 amplifier.

I have conducted a series of tests (the most succesfull one i have atatched for you to see). I have been able to get these readings if i do the folowing.

1. 1000 cycles where i add the results together and devide by 1000 in the end.
2. between each cycle i put a delay of 4 milli seconds

This means that it takes 8 seconds for the setup to produce a single printable value.
This is a bit on the high end i think.

The problem is (as you can see on the graph) that it takes about 19 cycles before the system reaches a steady state...... so about 2,5 minutes before I can actually use the results it produces.

You mentioned that the sensor has to be shielded from its environmen, this migth be the problem. I am using a light bulb that emits about 65 dgr celcius of heat. I am figuring that the sensor itself is slowly getting warmed up by this heat. ... this may account for the slope of the graph until the sensors own material is equilibrium (temperature wise)....

I simply dont know - am only speculating...

If you have any further comments on what information i have added here please let me know - its much apriciated

Kristian

 

[FvM]

circuit connect with temperature sensor

Depending on the dominant noise source in your setup (sensor and amplifier or "enviromental"), a better amplifier must not necessarily bring a significant improvement.

When using a sampling uP-ADC with e.g. 250 Hz sampling rate, an analog low-pass should be present between amplifier and ADC. You didn't mention it, also the gain (respectively the ADC resolution referred to amplifier input, how much uV is an ADC step) would be interesting for noise calculation.

As another remark, a digital low-pass filter has a somewhat lower output noise, if you use a more smooth window function, than your boxcar averager.

Noise and initial drift or settling time are different problems, basically. They should be measured separately and may be handled by different means.

 

[Kristian]

ir temperature sensor project

Hi again

I am fairly new to electronics and programming so please excuse me if i am slow on the terms

I am using an analog low pass filter at the moment ("10muF, 16V", with 46kOhm resistance). Regarding the ina128p, I am using 50 ohm resistance on the gain which gives about 1000 times amplification. If the latter is not what you refer to as an uv-adc step pleas let me know. I was under the impression that you need an oscilloscope to se the noise levels - can you calculate it instead ?

As you mentioned noise and setteling time are different problems. I would very much like to deal with both of them - have you got any suggestions for how to get the setteling time down?

 

[FvM]

op07 temperature sensor

10µF/46k (0.46 sec) low-pass should be effective. A gain of 1000 results in a 5 µV/LSB input referred resolution with a Vref of 5V. The INA128 noise, including the sensors resitive noise (60k) and noise current effect, should not be higher than one or a few bits, if low-pass filtered with the said RC combination. You can check, by replacing the sensor with an (about) 60 k resistor.

If so, the considerably higher observed noise would be completely generated by the said temperature fluctuations. Then there is nothing to improve on the amplifier side.

 

[Kristian]

pic thermistors temperature

Hi Fvm

Than you for your reply.

I Have tested as you said by replacing the ir-sensor with a 60 kOhm resistor. I get fine readings that vary one or two bit as you said.

I know that your respond is regarding the noise level on the amplification but maybe you could give the setup a glance as well. I have effectively brought the noise level down but I would like to be sure that I have connected the Ir- sensor corectly. I get the best readings by setting up the wires as seen below in the picture, but I have not been able to find any information in the data sheet or on the internet on whether the setup is correct or way off . . A good guess would be more than I have now

 

[Kristian]

ir temperature sensor

I just see now that I got one of the labels wrong. The top label that sais Thermopile GND is supposed to say thermistor...
The rest of the information is what was available from the producers datasheet - if you can get some information out of that regarding the correct configuration it would be. I have not.

Danke

 

[FvM]

temperature sensor cct conduct by pic

To my opinion, at least two details are bad in your circuit. I don't see a hint, that the manufacturer intended to connect the thermistor in parallel to the thermopile. It's intended for an independent temperature measurement, e.g. with a second AD channel, using a simple resistive divider with the NTC at the bottom. The purpose of the thermistor is mainly as an absolute temperature reference, because the thermopile signal is representing a temperature difference. It's also meaningful to connect the case (Thermistor GND) to circuit GND to reduce interferences.

The thermopile is apparently isolated. If you connect it to a differential amplifier, a bias voltage has to be provided, too. The INA128 datasheet is discussing this point under "Providing an Input Common-Mode Current Path". If you have a dual supply, it can be simply biased to ground.

Actually, a single OP, e.g. a good-old OP07 or a more recent medium input impedance, low noise type could be also used in a simple non inverting V=+1000 circuit. There's no particular advantage in using a differential amplifier to my opinion.

 

[Kristian]

pic cpu temperature sensor

OK - thank you . I will try to connect a second amplifier and see if I can counteract the input from this to give me a faster steady state for the system.

Your help has been much apriciated.
Kristian

 

[raoufthelionheart]

Re: how do I connect an infrared temperature sensor to a pic

hey guys , actualy am workin for a distance temperature measurement for humain ,
its too hard usin 1 m dis without laser orientation ,
so i choose 30 mms for ma project ...and i get : ZTP-135 SR for Ge and am buildin its compensated circuit , use TL084 amplifier , in any why i will try to fix ; 50 mv/ deg c ...its hard for its calibration ....
please if there is any one understand good whot is the relation beween the thermistor and the thermopile and exactly ; why they do " defferentiel amplifier "
i mean for whot reason ! , i read many PDFs and i see that it used to get a clear humain radiation , but idono why they do (Uth - Uthmistor).. thanks