Hi Folks,

I am operating a furnace having temperature reaching approx. 1900 °C and use a Thermocouple to get the temperature. But I am having a serious problem with my Thermocouple since it can only read till 1650 °C.

I have not put the Thermocouple in contact with the heating point because of this. I roughly estimate the temperature based on previous calibration results. But I feel what I read is wrong due to unexpected results I get from the heating process.

I am planning to use a Photo-transistor (View hole can be used to place the Photo-transistor) to fix this and I need some help from you Guys! I have enough knowledge in PICs.

Basically what I need to know is an interfacing circuit to connect the Photo-transistor to PIC.

Please let me know how can I build an interface circuit. Any help from you os greatly appreciated!

Regards,
/Lakmal

 

Hello,

What type of thermocouple are you using?
There are special types for high temperatures upto 2300 °C
http://www.omega.com/techref/thermcolorcodes.html
(types G C abd D )

How do you want to measure the temperature with a photo transistor?

Bertus

 

Hi,

Yes, there are. I googled for those thermocouples and it seems they are comparatively high cost.

I want to have a quite low cost solution. Deviation of ± 3 °C from actual temperature is acceptable.

My idea is to detect emitted light from glowing object in the furnace and make it incident on the photo-transistor and measure the output and express it as Temperature.

I might be heading in the wrong direction. Please help...

Regards,
/Lakmal

 

I want to have a quite low cost solution. Deviation of ± 3 °C from actual temperature is acceptable.

That is quite a contradiction, you can never have cheap and accurate at the same time. Besides, ±3°C at 2000°C is 0.3% accuracy, which I think even the right thermocouple will struggle with.

The phototransistor might work, but there is a lot of problems with that.
First, I have no idea if whatever you place inside the furnace and its shape will change the amount of light registered.
Second, I am not sure how much is the amount of emmited light proportional to the tempereature, and if it even is possible to detect such small changes.

I would definitely go with the thermocouple, more so if it is a one off project.

 

Hello,

A high resolution will not be cheap.
A pyrometer will probably work.
Here are two examples of commercial types:
http://www.pyrometer.com/Pyro_Optical.html
http://www.pyrometer.com/PyroLaser.html

Bertus

 

Here's another one that will go to almost 4000C:

http://content.honeywell.com/sensing/control/products/23750303.pdf

All of the high temperature measuring stuff at the steel mill was either infrared or T/C. Mostly Rayotube and Leco. Leco vanished.

 

Hi,

Thank you all for your time in replying to the thread. Having read the articles attached to the thread thought of going for "http://www.pyrometer.com/Pyrptical.html" one.

Thank you again!

/Lakmal

 

Check out stores serving pottery makers. The ovens used for firing reach those temps and higher, and they have something called 'cones'. It might be more what you want for setting equipment calibration points than for everyday use though.

 

The Pyro website doesn't say much about emissivity except for molten metal. If you are cooking different things, they will have different emissivities which will affect your readings.

 

Hi,

Thanks!

Pyro website doesn't say much about emissivity except for molten metal. If you are cooking different things

This is for a Geuda heat treatment process. So I am cooking only one type of thing. Once it is setup and calibrated, I think I don't have to worry about it.

...they have something called 'cones'

I think you mean "Seger cones". I used them. But my heat treatment process takes 3-4 hrs for a single firing. So it is difficult to keep an eye on the Seger cone till it gets melted. Thats why I am looking for an automated process.

Though I was planning to go for Pyrometer's one, still I am thinking of a low cost one compromising the accuracy.

Anyway I am planning to experiment with a photo-transistor too.

Can I use the photo-transistor (with a resister) output directly to input to PIC, or do I have to build any extra circuitry? Please advice.

Regards,
/Lakmal

 

I just looked up the seger cone. Perhaps you could use them in some automated way with a break beam detection system.
Using an op amp buffer is usually recommended with a resistive divider type sensor and PIC because of the characteristics of the analog input circuit in the PIC.

 

A PIC ADC needs a low resistance source driving it- like 2.5k or less. It is unlikely that a simple phototransistor + resistor circuit will meet that requirement with out an opamp in between.

 

Thank you for giving useful information. I will look in to the OP-Amp. I thought I can directly feed PIC with the transistor-resistor network.

 

I searched the internet and couldn't think of which op-amp configuration to be used. Please give me some hints regarding the op-amp configuration.

 

This is the one:
http://www.allaboutcircuits.com/vol_3/chpt_8/4.html
Note, if you are using something like the LM324 and a 5V supply it will only work for inputs 0-3.5V so plan the divider so that voltages are at the lower end.

 

We used tungsten-rhenium thermocouples for high temperature measurements (over 2000 °C) over 30 years ago and it was fairly routine. But our environment was in ultrahigh vacuum systems (molecular beam epitaxy systems), which made things easier. The problem with high temperatures and most metals is the resulting oxidation of the thermocouple material.

As you've probably discerned, optical/IR pyrometry is probably your best choice. An inexpensive IR photodiode could be a good choice. However, unless you have a calibrated spectroradiometer, you're not going to be able to calibrate your sensor. The spectroradiometer's output can be fit to a black body spectrum and the absolute temperature inferred. A hole in the oven makes a good simulation of a black body, as is well-known. Otherwise, you also have the problem of estimating the emissivity if you're trying to estimate temperature from a particular heated material.

If I was doing this, I'd want a spectroradiometer to also measure the response of the photodiodes as a function of temperature. Otherwise, if you want to estimate temperature over a range, you'll have to know the bandwidth of the photodiode and do some integrals of the blackbody curve.

My guess is, since you want cheap, you'll be satisfied with a relative calibration rather than an absolute one. You can use the slumping cones to provide the calibration point and thereafter use the output of the photodiodes. Note you can get Tempilaq materials that measure up to around 1000 deg C, so that could provide another data point.

 

I searched the internet and couldn't think of which op-amp configuration to be used. Please give me some hints regarding the op-amp configuration.

This is one of dozens of rail to rail i/o single supply op amps that will work. Its about $2.50:

http://cds.linear.com/docs/Datasheet/16389ff.pdf

Wire it up as a voltage follower as previously mentioned.