Thermocouple to measure the temperature of glass

[c1a3m1]

Is it wrong to use a thermocouple to mesure the temperature of glass that will range from 20-300oC? as the range of the thermocouple is much higher, does this make it inaccurate?

 

[Dr Transport]

no, staying well within the upper and lower bounds of the devices range alows you to be more comfortable with the data. I.e. there usually is a linear portion of the calibration curve...

 

[sam024]

Depending on what type thermocouple you are talking about as there are various types (k,j,e,n,t,r,s,b) they all have different ranges, type k being typically around -200 deg.c to +1200 deg.c and yes, you're right , the greater the range of the measuring instrument, the less accuracy you will have. That's one of the reasons why that in engineering, the calculations need to be accurate.
Think of it like you need to put 10 PSI in your bike tyres but the pressure gauge your using to measure it is 0 - 1000 PSI. Wouldn't it be wiser to use a gauge with a range of about 0 - 20 PSI? Then you have the resolution.

Hope this helps

 

[Integral]

Sam,
I do not think that a thermocouple is analogous to a tire gage. The thermocouple produces a voltage which is proportional to the temperature, you then use a meter to measure that voltage. Your gage is like the meter, the thermocouple is the device being measured. The key issue is, as mentioned above, that you use a thermocouple which has linear response in your region of interest.

http://www.omega.com/prodinfo/thermocouple.html is an excellent resource for temperature measurement questions. They have very good technical reference available and sell (at a price!) everything you need to measure about anything.

 

[sam024]

Integral, you missed my point. What i was trying to explain was that if the measuring instrument is of a much higher range (say, +200%) then you start to lose accuracy.The tyre gauge example was just a way of explaining.
And yes, you're right about measuring the millivolt output with a multimeter but you will only get an output if the 2 junctions are at different temperatures.Also, if the hot junction is in the process and the cj is at ambient, then the corrected output is the difference between the two.
Hope this clears things up.

 

[mechie]

Sam,
...you use a thermocouple which has linear response in your region of interest.

on an exponential curve? surely you refer to the indication system (thermocouple and indicator), the pair may have linearity limits?

the greater the range of the measuring instrument, the less accuracy you will have.

Are we confusing accuracy with resolution?

...range from 20-300oC? as the range of the thermocouple is much higher, does this make it inaccurate?

Every instrument is inaccurate ? only the magnitude of the error varies.
If you want the best accuracy then you could consider a resistance thermometer as these are generally considered to be better for this kind of temperature range.
Thermocouples are fine if you have a decent indicator - the question now becomes ... "what kind of accuracy do you want?"

 

[Integral]

on an exponential curve? surely you refer to the indication system (thermocouple and indicator), the pair may have linearity limits?


Are we confusing accuracy with resolution?


Every instrument is inaccurate ? only the magnitude of the error varies.
If you want the best accuracy then you could consider a resistance thermometer as these are generally considered to be better for this kind of temperature range.
Thermocouples are fine if you have a decent indicator - the question now becomes ... "what kind of accuracy do you want?"

The response charts I see for T/C is very nearly linear for large portions of the temperature range. None look exponential. Could you please provide a reference? (See my Omega link.) a T/C requires a very sensitive and stable voltage source for a reference (this can be used in place of a second junction and good millivolt meter for the measurement. In this respect they can be a bit hard for a home hobbyist to use. Though you can buy (for a price) temperature compensated temperature controllers which very easy to use.

 

[mechie]

I followed the Omega link but it appears to me that the manufacturers are trying to bamboozle me with science! I find little useable information here, only 'screen clutter'. However...
On the first page of their "introduction to practical temperature measurements" is a table listing merits of various temp. measurement techniques; thermocouples are listed as non-linear

On page z-26 of their "using thermocouples" pdf is a graph of various thermocouple response curves.

This (attached) formula is that for 'K' type thermocouples and for positive temperatures ONLY.
This is as specified by the UK National Physics Laboratory as a simplification suitable for computer calculations to about 0.1 degree C.

The full calculation (as per BS EN 60584.1) gives polynomial functions with between four and fourteen terms, with constants given to eleven significant figures.
My reference is www.tc.co.uk, they do an excellent (and free ) wall chart of tc look-up tables.

I agree that thermocouples aren't the easiest things to use but given the gear (and there is no clear statement in c1a3m1's origional posting as to what is available or where the measurement is taking place) I wouldn't rule them out - I use 'K' type thermocouples at home with a commercial 3.5 digit temperature meter.

To give an answer to the origional posting I would still say
"Every instrument is inaccurate ? only the magnitude of the error varies".

ps. I gave up trying to use LaTex to display the formula, I couldn't get it to show anythig but my source text

 

[techvineet]

surely depending upon the type of thermocouple.