thermocouple signal amplifying

[edvardas]

Hello,

I'm trying to amplify a thermocouple signal and read it via velleman K8055 board. Im using LM258P operational amplifier. I'm connecting it like this:

**broken link removed**

I connect thermocouple parallel with multimeter, to check readings. K8055 input is 8bit.

Problem is that when i connect it the computer shows me 19bars ar 21C° and it is increasing... in 10 mins there is 44 bars and stops increasing, though multimeter shows stable 21C°.. I have no idea why is this happening? maybe its because of too much noise? or i should connect + of the thermocouple via ~10MOm resistor to ground?

any help would be very welcome.

 

[ericgibbs]

hi,
Is the Voltage output of the OPA stable over that 10 minute period.??

Whats the gain setting of the K8055 internal OPA.??

 

[edvardas]

hi,
Is the Voltage output of the OPA stable over that 10 minute period.??

Whats the gain setting of the K8055 internal OPA.??

K8055 gain is 1. I'm measuring OPA voltage output with K8055 and it shows that it is increasing. Also tested that with multimeter it rises slowly..

I've kept it online over an hour now and already from 19 bars to 80 and still keeps rising. So it looks like it is integrating signal or something..

 

[ericgibbs]

K8055 gain is 1. I'm measuring OPA voltage output with K8055 and it shows that it is increasing. Also tested that with multimeter it rises slowly..

I've kept it online over an hour now and already from 19 bars to 80 and still keeps rising. So it looks like it is integrating signal or something..

Have you checked the current from the non invert pin of your OPA to the TC, is there a current flowing via the TC which is causing self heating of the TC junction.???

Which TC type is it.?? K T, etc

 

[edvardas]

TC is K type. It's a TC From Mastech MS8222H multimeter, It's K type TC. It is marked P3400 Thermocouple

I've measured current from non inverting input to the TC and multimeter shows nothing. What else could it be?

 

[ericgibbs]

TC is K type. It's a TC From Mastech MS8222H multimeter, It's K type TC. It is marked P3400 Thermocouple

I've measured current from non inverting input to the TC and multimeter shows nothing. What else could it be?

Try the following test, disconnect the TC from the OPA, connect a 1K resistor from the Non invert to 0V [ so there should be no signal into the OPA] check the output voltage from the OPA, over a 10 minute period , starting from switch ON .
Lets know

 

[edvardas]

i tried that test. After connecting it still is raising, After connecting it was 51 bar (5V/256=0.0195V/1bar, so 0.996V) and after 10 mins it was 55 bars (1.07V) but it stopped increasing and was stable..

 

[BrownOut]

Your schematic shows no ground connection to your OPA.

 

[edvardas]

Yes. I've missed that. OPA is powered via single power supply, and another clamp is connected to the ground.

 

[squishy36]

I'm no expert, but I don't like the fact that you have one side of the thermocouple grounded. That is probably done using some type of metal-to-metal contact and you have the opportunity for stray thermoelectric voltages. It's better to connect the thermocouple's leads to an isothermal block to avoid such problems. Then you could run a wire from each thermocouple leg to an input of the op amp -- and use it as a true differential amplifier. Hopefully, the CMRR would take care of any noise problems on the lines (if not, the EEs will have some good suggestions).

I also don't see a zero point reference, so your temperature measurements will be floating in a temperature sense. This is fine if you want to make temperature difference measurements, but make sure the isothermal block doesn't change temperature between readings.

Note you can check the thermocouple's operation by running the thermocouple's conductors directly to your DMM. You'll only get e.g. 3 or 4 mV for dipping it into boiling water, but that can be enough to see that the thermocouple is behaving as it should.

 

[Reloadron]

If you are doing this with any hope of actually measuring temperature the operational amplifier of choice is not a very good choice. There are operational amplifiers designed for use with thermocouples like the Analog Devices AD594 and AD595 family for example. Operational amplifiers designed for use with thermocouples take into consideration CJC (Cold Junction Compensation) which once you understand and become familiar with it is very important when working with thermocouples. Beyond CJC there are other considerations. You mention using a Type K thermocouple for example.

At a temperature of 0 Deg C the output of a type K T/C is 0.00 mV and at 100 Deg C the output is 4.096 mV followed by 8.138 mV for 200 Deg C and at 300 Deg C we get 12.209 mV. Thermocouple outputs are non-linear. Therefore you have two problems to overcome before you worry about the A/D conversion. You have CJC and linearization.

The A/D you are using is an 8 bit A/D converter. Depending on the range you want you won't get very good resolution with 8 bits.

Depending on the application and if this is important as to accurate temperature measurement you may want to think about using a temperature transmitter designed for a thermocouple input providing a signal conditioned 4 to 20 mA or 0 to 5 volt output.

<EDIT> Years ago I did some temperature measurements using the type temperature transmitters I mentioned. I used Type J T/Cs and wrote some software to measure some computer case temperatures. Some of the results can be found here.
I used a device similar to the device you have for data acquisition. Running 4 channels for trmperature it worked out well. </EDIT>
Just My Take
Ron

 

[edvardas]

Thanks for answers.

Could you tell me an example of a isothermal block in practise? i found some info saying - isothermic connector made of copper. What is it? Big piece of copper? I think this principle is used in multimeter, because my TC is joined in one end.

I'm afraid that these specially designed amplifiers are way too expensive for me.

I'm interested in temperatures from 20-70C°, i am doing some experiments with the shape memory alloy springs and i need to know and keep their transformation temperature. So 8 bit's should be enough for these temperatures. I know that TC is non - linear, and i was wondering how it is compensated in the multimeter. But i was going to test TC in different temperatures and make a table for it's output's comparing to the temperature, so i make coeficients at 25C° to temperatures from 20 to 30 degrees, and so on, 2C° accuracy is ok for me now.

If i understand CJC right - we need to know the ambient temperature, and according to it we choose value for TC electric output from lookup table. But i've found a K type lookup table with data that seems to be absolute values from -200C to 1200C°. And i was thinking of using this, deciding the temperature directly from the lookup table, e.g. i have 1mV output, i look in table, and i see that it's 25C°. So i guess now this CJC effect is doing this for me?

 

[ericgibbs]

hi ed,
I would suggest for a temperature of only 20-70C a K type T/C is not the best sensor for the project.

What sort of temperature accuracy and response time are you designing for.???

 

[edvardas]

Im hoping of accuracy 2-3C°, response time the faster the better, i think 2Hz at least. The connection scheme will be like this:
**broken link removed**

sorry for horible sketch so i will heat the shape memory alloy (SMA in short) spring with electric current (5V 3-4A) and i want to keep an eye to it's temperature via TC. And acording to this i will be able to control it, e.g. keep temperature constant, prevent overheating, etc..

 

[Reloadron]

I would agree with Eric in that a Type K T/C is really not the best choice for a range of 20 to 70 Deg C. With a range of 20 to 70 C (68 to 158 F.) and you want an uncertainty of +/- 2 Deg C. you may want to consider a sensor like the LM35 Precision Centigrade Temperature Sensor. The nice thing about using a device like this would be the Linear + 10.0 mV/°C scale factor output. Using a simple and very inexpensive device like this for example would give you the range and uncertainty you want. I am not sure about the response time but unless you need very fast response and plan to sample at a very fast rate you may want to consider the LM35. There are other sensor options including the use of a thermistor, however, the old tried and true LM35 is likely a good choice for you. They cost less than $2.00 USD and are readily available globally. The LM35D for example would give you 0 to 100 Deg C range with 10mV/Deg C output.

The small TO92 case can easily be epoxied or bonded to the product surface you are testing (alloy springs) and provide an accurate indication of the actual surface temperature of the part. I believe part temperature is what you are after. Beyond the linked data sheet a Google of LM35 Circuits will bring up countless examples of using the sensor.

If you still have questions on thermal blocks (CJC) and thermocouples in general I will be happy to answer them but again I really agree with Eric in that a T/C would not be a good choice for your application.

<EDIT> I see you were posting as I replied. Just as a foot note a sample rate of 2 Hz, a 500 mSec sample rate or two samples per second. Don't confuse the sample rate for your A/D with the response time of the thermal sensor. They are different animals. </EDIT>

Ron

 

[edvardas]

Im very grateful to all of you for your help. I've found LM35DZ in one of my local shops, and it is exactly 2USD i will try that. That's a great discovery for me

 

[ericgibbs]

Im very grateful to all of you for your help. I've found LM35DZ in one of my local shops, and it is exactly 2USD i will try that. That's a great discovery for me

hi,
I would also suggest that you use the K8055 +5V line for the LM35 supply. An external 5V is not required, the USB can easily supply the LM35.

Use one of the K8055 internal ADC , OPA's [ increase the Gain to suit]

I have modified my K8055 to 20MHz and a PIC18F2550 which I have programmed.

 

[edvardas]

Yea, i was hoping of using that board's supply. And how does that upgrade improved K8055 board? In my eyes, the biggest dissadvantage of my K8055 is the slow connection speed to pc and low-bit ADC. Pre program new pic is hard task, and you're a true expert.