Thermocouples & RTDs

Rev. ... 2002-03-11; 2003-02-26, 2004-01-08, 2005-03-10, -04-20, 2008-12-09, 2009-05-27, 2010-12-05

Return to Sitemap

Related Pages
 Control  Annealer  

The three most common methods of measuring temperatures involved in furnace glass work are infrared radiation (IR), thermocouple (TC) and resistance temperature device (RTD). IR is very nice but, especially in the temperature ranges of molten glass, very expensive at $1100-1200 (Omega). RTD's are limited to annealing temperatures (1200F).  Therefore thermocouples, in ordinary and platinum metals are used.

[Moved in from annealer.htm and modified]

A thermocouple is two different metals, usually wire, welded together to produce a junction that works in reaction to the other end of the wires. When the junction is heated (or cooled), a voltage is produced along the wires because of the difference in temperature.* Any pair of metals will do this (the reaction is part of what causes corrosion), but some do it much better (higher voltage, better linearity.) Because the effect will occur with any metals, care must be taken in wiring and switching thermocouples to avoid introducing additional voltages by creating other junctions. [This site, Thermocouple Calc, will calculate voltage vs. temp and vice versa for a number of different kinds. This site, Thermocouple Technical Reference, has tables of information including millivolt outputs for a wide range of temps. 2005-04-20]

Thermocouples are given letter identifications: J, K, etc. For the highest temperatures, platinum (Type R, S or B) is needed ($125-400 each.) For annealers, kilns, and glass furnaces, K type are used, although they need protection at furnace temps or get eaten up over a couple of months and lose accuracy, although the amount of loss seems small to me. (At $16-24 vs $125-400, a lot can get eaten.) Higher temperature ratings go to thermocouples with more material (thicker wire) because the failure mode is oxidation of the iron in the type K. Type K is made of Chromel and Alumel. A type S is Platinum and Platinum with 10% Rhodium. I have not used type R.

Thermocouples with porcelain insulators, mounted.I use heavy duty thermocouples with dual hole porcelain insulators and a terminal block on the end with set screws for connections.  As shown, mine are abused.  The upper left inset is on my glory hole, the lower right on my annealer.  In both cases, a 2" metal corner bracket is screwed to the case and a stainless hose clamp holds the thermocouple insulators.  A short length of wire leads to a connector like below and a long length is used to the controller or meter away from the heat. The bent unit had something fall on the end. 2009-05-27

K type is connected with specific yellow (+) and red (-) insulated wire with the metals in the wire matched to metals in the thermocouple and I use a standard mini-connector (right) on my devices that lets me also use commercial thermocouple probes and measuring devices. K type thermocouples are typically $16-25, connectors $5.50 a pair. (Each connector has two flat blades, one wider than the other. In the image a blue coated small wire thermocouple is attached to the plug at the top and the yellow cased red and yellow lead wires are attached to the socket at the bottom.)
Standard mini-connector for K type thermocouples


Graph of thermocouple voltages - S scaledVOLTAGE OUTPUT - Thermocouples put out varying voltages with temperature and the change over each ten degrees also changes.  For this reason, some kind of conversion must be done or the temperature reading will only be approximate.  Also, wiring must be carefully done because the voltages are small.

K & S type thermocouples are defined above.

The upper graph shows the line of millivolts vs. temperature in Celsius.  The red line is a K type.  The green line is S type. S type output is about 1/5th K type.  The blue line is S multiplied by 5 to show the different shape of the curves.

Scaled S type volt change graphed with K type changeThe lower graph shows the difference in mv at each 10 degrees with the S type multiplied by 5.  The K type is much closer to level, but has a funny little dip at roughly 190C.  The S type has constantly increasing millivolt output for each 10 degrees. 2005-04-20


METERS - In order to measure temperatures, I looked at analog and digital pyrometers. Analog uses a dial with a moving needle, while digital displays digits (numbers.) Less expensive analog meters with little adjustment can be used with K-type at the temps of annealing, sagging and fusing because of the levelness of the curve above.  I chose digital for convenience and accuracy and durability, since reading most analog displays to closer than ten or twenty degrees is guessing after squinting at the needle if it hasn't broken when dropped. I looked at dedicated temperature measurement units as well as adapters. I decided to buy an adapter that uses a digital voltmeter for the display. [This is early.  I have also bought volt meters with K type temp reading, which have come down considerably in price, to $39.99, as time has passed. And a K-type C only for $29.99 here 2004-01-08]
 In 2010, looking around, a number of voltmeters have added thermocouple reading capability either with K-Type mini-jacks on the meter or through the banana jacks.  To allow the common mini-plugs to use the latter, adaptors have become available such as the Fluke 80-AK-A and other brands without the shrouded plugs. 2010-12-05

Costing less than the cheapest digital thermometers, but more than analog, the adapter includes amplification and compensation so the reading in millivolts is equal to temperature and the voltage is read with a cheap voltmeter. A single purpose temperature meter is about $110.  I use the adapter with the lowest cost Radio Shack Digital Voltmeter ($17 on sale.) One adaptor still available is the Fluke Thermocouple Converter 80TK  
[2001-08-16 Omega has come out with an $88 analog to voltage adaptor SMCJ-K  which connects to some meters by directly plugging into the meter's banana jacks, but needs an adaptor for cheaper ones. Omega also has a neat, but $249, talking DMM with 2 K-type inputs.] another product]  [2005-03-10 More recently, meters costing as little as $39.99 have come on the market - on sale - with a K-type jack. 2005-03-10]

Note that it is also possible to build your own converter using a chip sold by Analog Devices which is discussed here.

The TA-1A adapter [no longer available] comes with a thin bead K-type probe rated to 1400F, except that the Teflon insulation is only rated to 500F. I started using it for testing (and eventually ruined the insulation) while waiting for more serious thermocouples. If only the tip is touched to 1400F presumably it would work. For higher temperature work, a TC made of 16 or 14 gauge wire has a higher rating. Since the thermocouple voltage is created along the temperature gradient, the connection at the tip only has to be okay, not perfect and broken thermocouples can be welded.

Following a suggestion from someone at Spruce Pine Batch, I ordered $8.50 [now $16] K-type probes from A.R.T. Studio Clay Co. [which is also a source for the large wire porcelain terminal blocks that mount on the thermocouple wires, that Omega does not sell with their heavy TC. 2004--01-08]  Neither Grainger nor A.R.T. carry K thermocouple wire separately, so I ordered that for $8.50 for 6' from Seattle Pottery Supply. The adapter uses (as do as most digital thermometers) a standard mini-connector, $7 for 2 from Grainger. For a thermocouple to work most accurately, all the wiring and all the connections should match the metals in the thermocouple. That is why special wire and connectors are needed. Wire and connectors are also available from Omega. [a much better source, 1-800-826-6342]


WIRING NOTE - Many places mention the need for care in wiring and switching but it takes some digging to understand how much care is needed and why. The why is that the thermocouple effect produces small voltages (21.493 mv is specified for 540C, 44.439 mv for 1080C - that's 0.044439 volts) so that it is easy to mess up the reading if other voltages are created. Connecting thermocouple wire to copper produces a new junction. How to avoid the problem? Keep all the pairs of new junctions at the same temp. At ultra precise labs, this is done with thermal junction blocks. In more practical situations, it is likely that the left-hand terminals of a switch or relay are not much different in temp than those on the right. So, if thermocouple wire is connected to a switch (copper or alloy) and a junction is created, the reverse junction is created at the output a fraction of an inch away and the effect is canceled. This clearly does not apply if the iron in the thermocouple is connected to copper wire which is then run several feet (or more) and connected to iron wire again. Thus, special connectors and wires are available, but special switches are in short supply.

Now that I have used the adapter, and the least expensive digital voltmeter from Radio Shack, for some time, I have found it stable and reliable. I have compared its readings with cones and other thermometers and checked the reproducibility of the readings.

The error specified for type K from 559F to 2282 F (293C-1250C) is 0.75% (while R is .25% and is more stable.) 0.75% is a small number, but taken times a large number produces a moderate number, in this case 15F degrees of error at 2000F.

"Iron rusts at low temperatures where condensation can form. Type K (nickel-chromium (Chromel) vs nickel- aluminum (Alumel)) can be used up to 2,300F in an oxidizing or inert atmosphere. Type K thermocouples exhibit a number of instabilities and inaccuracies, particularly at higher temperatures, changing their emf/temperature characteristics. In reducing atmospheres (lack of oxygen) at temperatures of 1,500 to 1,750F the positive thermo-element forms a greenish chromic oxide, commonly known as "green rot". This causes a decrease in the electromotive force of the thermocouple. Type K is also subject to aging when exposed to 800 to 1,200F for a few hours. Temperature cycling above 1,400F and then below 700F causes a random error due to changes in the composition (inter-granular structure) of the conductors. "

FLAME DETECTOR - When plain wire is heated at one point, the thermocouple effects in each direction cancel. If one could imagine somehow making a metal free connection at the hot point of the wire, one could get the thermocouple effect of just one kind of wire between the connection and the end of the wire. It turns out exactly that this can and is done in one kind of flame detector.
In a furnace, it is often important to detect if the flame has actually ignited, so the furnace does not fill with an explosive collection of gas and air. One way this is done is to put the end of a thermocouple wire into the place where the flame is supposed to be. A hot flame is ionized, so it will conduct. So a voltage can be created between a metal part of the burner head and the wire. A sensitive circuit can detect the voltage, or failing to find it, shut down the gas supply. 2001-08-27, 2009-05-27

Thermocouple question
Posted By: Brody Date: 2/25/2001 - 1:58 p.m.

Hello one and all! Just a quick question...What's the life expectancy of a K type thermo in a furnace? Now before everyone shoots off at me, hear (or read) me out. First I know it's gonna fry. But in how long? 10 sec, 30 sec, maybe a minute or two? I'm trying to get a temp. reading on my furnace, and looked over at an old kiln with a pyrometer on it. And thought I might be able to use it for a "spot" type reading of the temp. I was thinking of using a glove and holding it just inside of the door, MOMENTARILY. Like maybe 20-30 secs max. Will this work, or am I just kinda foolin' myself? Thanks for the answers (go nice!) -Brody

The problem with K type thermocouples is not how long they will last. I have a couple that I have had in my gloryhole for hours at a time and they survive quite well thank you. They are big wire - 8-11 gauge. Small wire will oxidize rather quickly and be damaged to the breaking point.
The problem everyone talks about is that they drift and become more inaccurate. My bone of contention is "How much do they drift and how inaccurate?" The answer turns out to be about 10-20 degrees. For people who are trying to do precision work, this is too much. For furnace workers, it is less than the variation from putting the thing in various locations.
A real problem is that for big wire thermocouples, the time it takes to make a measurement is looooonnnnng. Your hand (or glove) will fry if you try to hold it. Long is 20-30 seconds before settling. Also, if the thermocouple has porcelain insulators as mine do, you can crack them off by heating too fast. You would need to ease it into place.
Mike Firth

Contact Mike Firth