How to calibrate a thermocouple?

[Michael Fowler]

OK, technically, since its non-adjustable, all we can do is verify.
How can I be sure that the EGT gauge is reading properly? We discuss where is the best location...( closest to the pistons, IIRC). Do we all just assume that 1200 degrees is REALLY 1200 degrees?
Dunking it in boiling water will show 212 degrees at sea level--or close to it, but that is really low on the scale that we need. Is the temperature of just melting solder, or lead consistent enough to use that as a reference point?

Backstory---I am tasked with getting our soils and concrete lab "certified", meaning inspected to show that we comply with what the Standards require. Part of that process requires me to measure EVERYTHING we use and verify that I can trace the accuracy back to NIST. So I am sensitive to calibrating/ verifying EVERYTHING. What good would it do if I watch the EGT gauge religiously, and keep the readings under 1200 degrees if the real temperature is 1600 when my gauge says 1200?

My first impression is that thermocouples are pretty standard, and reliable. The voltage they generate is a repeatable phenomenon, and probably always right. I just want to be sure.

With snow-megeddon now occurring this question is more hypothetical than immediate---I am not going out in this weather, or the single digit temperatures that follow it to install the gauges.

 

[laserjock]

As I am sitting here watching it snow along side you all I can say is this on thermocouples. Typically I believe pyros are type K thermocouples. They generally have a rated linearity. Type K goes way up there (we use them in process furnaces over 800 C). They are pretty linear probably +/- a percent or so. at 1200 degrees that's 12 degrees. At 1200 degrees, does 12 degrees matter? I would argue probably not a lot because you have several other factors that figure into things like response times and placement etc. I would do the easy 2 point calibration. I would go with the ice bath and then boiling water. If you read good at those two point, you should be within the stated accuracy of the thermocouple at temperature. Now, that being said, a lot of times when people have errors in their thermocouple readings it's not the thermocouples fault. The wire running to the thermocouple is critical achieving an accurate reading. Just using regular copper wire introduces an error in the reading that is proportional to the lenght of the copper wire used. You have to use the appropriate wire based on the type of thermocouple because its the juntion potential that creates the voltage read off. Also have to have a reader that is precise enough and really the pair should be calibrated together if you really want to get down to brass tacks.

Probably more information than you wanted.

 

[franklin2]

Thermocouples are not calibrated, the gauge or instrument they are hooked to are calibrated.

 

[laserjock]

Thermocouples are not calibrated, the gauge or instrument they are hooked to are calibrated.

Strictly speaking you are correct. That is why I usually check the assembly if I need REALLY accurate readings. Majority of the time, just having the reader in calibration is good enough.


Sent from my iPhone using Tapatalk

 

[Michael Fowler]

Agreed that 1% is not critical.
Also --not watching the snow---I'll have to shovel it soon enough.

 

[OLDBULL8]

Here is the different types of thermocouples.
http://en.wikipedia.org/wiki/Thermocouple#Type_K

As far as verifying the temp read on the EGT gauge as to the actual temp at the thermocouple, with a noncontact Laser gun, read that temp and compare it with what the EGT gauge reads. A type K thermocouple is most commonly used on our engine exhaust, range is from -330*F to +2460*F. or measure the micro volts and convert to temperature. When you buy a thermocouple (Pyro) it comes with a certain length of wire, and it's cautioned not to adjust the length.

 

[laserjock]

Here is the different types of thermocouples.
http://en.wikipedia.org/wiki/Thermocouple#Type_K

As far as verifying the temp read on the EGT gauge as to the actual temp at the thermocouple, with a noncontact Laser gun, read that temp and compare it with what the EGT gauge reads. A type K thermocouple is most commonly used on our engine exhaust, range is from -330*F to +2460*F. or measure the micro volts and convert to temperature. When you buy a thermocouple (Pyro) it comes with a certain length of wire, and it's cautioned not to adjust the length.

I took a snow day today. I think it's the first time since the massive blizzard a few years ago.

Oldbull, you can lengthen it so long as you use the correct thermocouple wire.

If you want to know anything about temperature measurement and control, Omega Engineering publishes a nice set of hard back books that they will generally send you free. Covers process control, pressure, strain, temperature, and level.

www.omega.com

Yeah they are tilted toward their products but it really is a good general reference set.




Sent from my iPhone using Tapatalk

 

[mblaney]

I am having a slow work day so I put this together. I took too long and there were a bunch of replies before I could post so some of this has already been said.


This is what I do! I am the manager of the largest commercial calibration provider in Canada - we calibrate a huge variety of equipment from torque tools, pressure, dimensional, multi-meters, scopes, spectrum analyzers... even aircraft weight scales and altimeters. (Pylon Electronics)

Your post asks a lot of questions - I could talk for hours about calibration, traceability, etc but I will try to only offer brief comments. I am not sure if you are really talking about your truck EGT sensor or a work project but I will assume this is for your work and that your ‘standard’ is probably ISO 9001, ASTM or other more demanding quality system (ISO 17025).

How to calibrate a thermocouple?

There is a lot more to calibrating a T/C; as laserjock said, the extension wires and cold junctions are critical to make your system read correctly. The key word here is ‘system’. You can’t check one and assume the rest is ok. Most T/C system failures could be grouped into:

- display meter / temperature logger is out of specification
- improper installation of T/C (cold junctions)
- T/C failure (hopefully obvious because the temperature will be way off what you expect)

Here are a few suggestions you might consider depending on your installation and risk. These suggestions would require you to have a decent handheld temperature calibrator – they run around $1000. Google Martel Electronics PTC-8010 for example. Omega Engineering, Fluke Electronics and many others make equivalent products.

1. If the T/C is in a furnace then you can simply verify it using a separate meter and sensor. This is really easy but you would have to purchase a separate tool and temperature sensor that can be sent out to a calibration provider for traceability. Your verification meter and sensor need to be more accurate than your process requirements. Four times better (4:1) is a commonly used.
- This is very good to do because you are using a completely standalone measurement system to verify your furnace.
- The problem with this is that if you need to adjust anything it takes a long time going back and forth making adjustments while waiting for your furnace to stabilize and it will not help you troubleshoot problems. This is not a good method for making corrections.

2. If you do not have direct access to be able to measure your temperature in parallel then you can simulate the T/C using an external calibrator. Typically you would remove the thermocouple and apply a simulated input to test the full range of your instrumentation; if possible include and extension wires (and junctions) in this test. Then reconnect your T/C and if it reads ambient acceptably then it is probably OK. The complications with this are obviously knowing what the ambient temperature is (if installed in something) and this does nothing to check your T/C at operating temperature. Usually if the T/C is reading OK with this test then it will pass tests at other temperatures.
- This is very fast for adjusting your meter/display and can be used at different test point to troubleshoot installation problems (improper junctions).
- Not doing a direct temperature measurement at your operating temperature(s) has its risks.

3. Do #1, then – if required - #2 and then #1 again. This will give you the most confidence in your furnace (or whatever) and therefore reduce the risk that your product, etc will fail.

What you decide should be based on an assessment of risk vs cost. Your ‘standard’ (ASTM, ISO, etc) should give you a good idea of what your needs are for accuracy, frequency of verification, etc but you will also have to consider the business side; what happens if you discover your furnace was off 100 degrees? 200 degrees? If you are heat treating aircraft parts the chances are that an out of tolerance will have a significant

Now, about getting your lab certified.

This is not as easy as checking your thermocouples Assuming you are working towards ISO there is a lot more than just getting traceability. Procedures, training, and documentation are requirements for an effective and reliable process. I only mention this because the procedures, training and documentation that are required for ISO should pretty much dictate what your measurement needs are (accuracy and frequency of verification for example). This is especially sticky for companies that want to do calibration.

This is why people like me have a job calibrating ‘everything’ for other companies. Usually it is far more cost effective to outsource your calibration needs rather than invest in assets, training, and a management system to cope with something that is not your core business. I am not suggesting that you can’t do it – but when you say ‘everything’ for a soils and concrete lab I envisioning load cells, dimensional equipment, ph meters, pressure transducers, maybe tensile testers.

So much for brief...

 

[mblaney]

laserjock - I second the Omega books. They are an awesome resource!

 

[laserjock]

Now I know who to go to with my cal questions. ;sweet

Great discussion.


Sent from my iPhone using Tapatalk

 

[Michael Fowler]

mblaney,

Thanks for the detailed info. However, the EGT is for my IDI truck--not worth the investment in test gear that rivals the value of the truck.
No ISO for us. We are doing AASHTO on only a few tests. The Load cell on our concrete compression tester is calibrated yearly by a firm that specializes in that. Most of the linear measurements in the Standards I must comply with are only to 0.01 inch. I will be using digital calipers 0-6 and 0-12 inches. Most of my items are either 2 ", 4", 6", 8" or 12", so I am going to invest in three caliper calibration standards, 2", 4" and 8". ( The 12" measurement is only +/- 1/16", and steel rule will do for that) The calibration standards are to 0.0001 inch, or 10X what I can measure, or 100X more than the precision I need.
I was tempted, really tempted by the ebay vendor who sells Joh block sets and claims NIST traceability, but that seems to be a false claim.

 

[mblaney]

I wasn't sure if you were using your EGT as an example of what you were doing at work. I agree with others here, your EGT should either work or it doesn't. What you are looking for is temperature changes relative to what you expect - not necessarily accuracy. If you daily driving shows 900 and towing pushes 1200 then that should be good enough to tell you (approximately) what is going on.

Note: about non-contact (infrared) thermometers - This may be a little overkill for EGT purposes but for accurate measurements you need to know what the surface emissivity is. The IR thermometer can be set to compensate for this to give you corrected measurements. IR thermometers will either have a fixed emissivity (0.95 is common) or are are adjustable but either way you have to know how to determine the emissivity of your target. This won't matter for measuring temperature differences between similar objects (like checking manifold temperature for each cylinder to find a bad injector) but if you want to use the IR thermometer to verify your EGT sensor then you need to know the emissivity. Rusty steel can be as low as 0.7. Assuming your thermometer is fixed at 0.95 your correction factor would be 0.95/0.7 = 1.36. This will cause a major discrepancy with any T/C.

Determining surface emissivity usually involves a sample of the product you are measuring and a surface contact thermocouple. Alternatively, you can greatly reduce the emissivity difference by painting your manifold surface flat black.