Thermometer / Thermocouple Calibration

Hi there,

First of all - how do you propose to calibrate the thermocouple?

It consists physically of two dissimilar metals - there is nothing you can calibrate. You can check the milli volt reading of the thermocouple at ambient temperature or in an ice bath (or in boiling water) and compare it to the relevant "type" chart - but if it is out you can only throw it away.

The only thing you can calibrate is the transmitter.

Regards,

Craig

It is NOT possible to calibrate a thermocouple,you can check it's accuracy if you want but it can not be calibrated...thanks

HI ALL,

Yoy can not calibrate the thermocouple.Only compare the reading result accourding the revelent temperature chart .One thing can do ,If you found any errors in reading so just adjust the digital disply itself.all digital temperature indicators have such kind of adjusment.

If you get a bucket of thermocouples and connect them one-by-one to your thermometer and stick them in boiling water they will all read "slightly" differently. So to get the most accurate calibration of your system you connect the relevant thermocouple to your thermometer and calibrate them as a pair. It all depends on how much accuracy you require and the normal temperature range you operate in. These days I would always like to do this if the thermocouples come from China or other possibly "dodgy" manufacturer to ensure accuracy.

Rok

Hi all,

Thanks for all your comments.

But one of the calibration lab saying that thermocouple can be calibrated as like below:

"""Using oil bath and dry temperature furnace with a set temperature [whose stability is 0.2 degC]Master thermocouple with indicator is used to calibrate other thermocouple [Comparison of the temp reading b/w master thermocouple & the thermocouple underwent for calibration give you the result]"""

Is this method ended up with accurate output.-Comment

Anticipating your quick response

Hari

Hi, haricool!

What this piece of information seems to be saying is that if you have an accurate contact (mechanical) thermocouple, you can use it to adjust the reading for another contact thermocouple being heated in the same soup by copying the new thermocouple's reading to match the reliable one. The soup you are using to calibrate the second thermocouple in will probably be its end use.

A mechanical thermocouple is usually set for one particular use because its accuracy may vary at different temperatures due to the construction/conductivity of its sensing end possibly changing as the temperature surrounding it also changes. But when it is calibrated for that one use, it may be at any temperature within the range of the instrument itself.

If you want a thermocouple-type instrument capable of performing on a sliding scale, you may wish to examine the range of infrared sensing or other "optical" temperature sensors.

Mark

The definition of calibration is the comparison of two instruments, one of which is the instrument-under-test and the other is the reference. Calibration may or may not necessarily involve adjustment of the instrument-under-test.

Therefore, calibration does not always involve adjustment. If your instrument can be adjusted, then by all means, do so. If not, you should be aware of (and record) the instrument's deviation from the reference and use this information to compensate for the deviation.

We once had a temperature indicator that couldn't be adjusted. When we calibrate it, we find out how much it deviated from our reference. For example, if it reads 35°C vs our reference of 37°C, we put a sign near the instrument that instructs the user to add 2°C to the display to get the correct temperature. Of course, I got rid of that instrument as soon as possible .

So, my advice is for you to calibrate the thermocouple and the transmitter/controller/indicator as a unit, just as several posters suggested (to whom I each gave a GA).

regards,

Vulcan

Thermocouple calibration is useful if you need better accuracy than +/-1°C, doing so may give you +/-0.1°C accuracy.

Actually you are only compensating at the temperature you are calibrating. A 1 point calibration may enable you to do offset compensation for your measurement range.

Only useful if you have a very good quality indicator or use mV measurement and table conversions.

Hi,

I have calibrated hundreds of thermometers AND thermocouples. Both for very different applications though. Is what you have a digital base that the thermocouple plugs into and then you use that to take a single temperature reading? Unless you have an extreamly large quantity of these things I suggest that you send it out for calibration to your "lab". It will be much more cost effecitve that way since the heat/cold block and the traceable probe (which will need to be sent out for cal) aren't particullary cheap. The oil bath is used for calibrating at VERY low temperatures and those are also expensive (not to mention the liquid you will need to purchase for it). Thermocouples are typically calibrated for individual temperature monitoring events, such as chamber mapping where you would calibrate over one range of temperatures for an incubator and another for a freezer. So, a tight range for each of the two separate events. A hand held thermometer like I mentioned previously would typically be calibrated over the range for the capabilities of that particular thermometer and is not so specific. But I don't really know exactly what you have there or what you are using it for. I just thought I would put my 2 cents in here.

Hi there you can produce a calibration curve for the thermocouple in question by measuring the voltage output at 5 degree intervals between 0 C and 100 C. This will give you a series of number pairs which you can put into a spreadsheet like Excel and from this produce a least squares curve fit. Having arrived at the formula you can then predict the temperature from the voltage reading within this range. You can program this into a PIC if needed to give you the correct temp display when using the thermocouple on the PICs analogue input.

Hi all,

I am now very confused - how do you physically calibrate the physical thermocouple itself? For that matter - how do you phsically calibrate a thermometer (besides changing the scale on the side)?

It consists of two dissimilar metals. It will give you a mV output based on its construction. There is no zero pot or span pot - just two terminals? In an ice bath (zero degrees C) Type x thermocouple will give you x mV, at 100 deg C type x will give you x mV. This is given and you have charts for type x,y,z therocouples.

I understand that you can influence the class therocouple (i.e. Class A/B) by the purity of the metals that you use in construction - but how do you physically calibrate the thermocouple itself (not the transmitter)?

You can however calibrate the trasnmitter that you connect to the thermocouple or the board that you make (which represents the transmitter).

Please for my sanity explain this? Do you disassemble the thermocouple and re-create it? Even then - this should have negligable affect.

Regards,

Craig

Hi, craigza!

As I understand it (I employ a calibration company when I need any calibrating), one doesn't ordinarily calibrate a thermocouple by physically adjusting its sensing parts to match another correctly adjusted thermocouple. Since it is recognized that each must in no matter how small a way be individual in nature, the thermocouple to be adjusted is hooked up to its data reader output, and together with a thermocouple that has been recognized as accurate, is subjected to the accurate thermocouple's accurate temperature range, as noted on its data readout machinery.

The adjustment is then made on the data reader of the thermocouple that has to be calibrated so that it matches the readout of the accurate thermocouple, or accurate thermometer, or accurate optical sensor measuring the same temperature. Normally, the thermocouple sensor is not adjusted, only the readout device receives a locked-in setting datum.

Mark

Calibrating a thermocouple or a glass thermometer only means to determine its output compared to that of a standard. After calibration you will have a chart of error values versus that of a standard for the temperatures of interest e.g. +0.42 deg C @ -100 deg C, -0.6 deg C @ 0 deg C, and -1.2 deg C @ +100 deg C. There is no way to calibrate a specific thermocouple to make it perfectly accurate although you can buy premium grade thermocoouples that offer higher accuracy (see Omega).

Hi there,

Thanks for that explanation - So it is not the PHYSICAL thermocouple that is calibrated as I stated earlier on - but rather the transmitter/recorder.

Thanks for that - otherwise science and laws would have indeed changed.

Regards,

Craig

Craig

When you make a thermocouple using a standard method its output voltage will to a certain extent be predictable from standard tables. However if you want better accuracy then the method I described which is used in physics labs is to plot a curve for your specific thermocouple as I described. This curve should be approximately linear or first order but by applying a second order fit you increase its accuracy ( you can do it in Excel). This calibration curve can then be used to accurately give a temp for a specicic voltage output. By applying the formula derived from the fit you can automate the process using a computer or PIC microprocessor. Obviously you need an accurate standard thermometer to do the initial calibration and this may not be available to the average person hence sending it away to be calibrated.

Hi, Pj3ns3n!

How do you manage the bumps in the curve?

Mark

You slow down, shift to first or second gear and ride the bumps smoothly. Then accelerate like an F1.

I just want to mention that calibration and verification are two different horses. Calibration is when you include adjustment in the procedure. Verification involves no adjustment, just recording.

So, in the case of a thermocouple, you just verify, not calibrate.

_{Where's my tin hat?}

regards,

Vulcan

Hi mark

Shouldn`t be any bumps but there will be errors and the least squares fit will smooth them out. For a more accurate curve take more readings at smaller intervals.

Hi, Pj3ns3n!

Thanks.

Mark