Thermocouple good at room temp is good always?

[cupoftea]

Hi,
If you are about to do thermal tests with thermocouples...then to see if they are ok...can you say they are ok, if when in the thermocouple reader, with the end junction bits loose on the lab table, then if they read room temp properly , then those thermocouples are OK? (at all other temps too). We only need accuracy to within about 4 to 5 degrees.

 

[betwixt]

There isn't much that can go wrong with a thermocouple! They should be near identical and accurate but check the manufacturers spec for linearity. Also make sure you are measuring them correctly, the connections between board and wires can make far more difference to the reading than the junction. Use cold junction compensation if possible.

Brian.

 

[KlausST]

Hi,

thermocouple measurement is a differential measurement of cold junction temperature and thermocouple temperature.
If the cold junction is at room temperature and the thermocouple is at room temperature .. then it measures ... nothing!
So if you want to measure absolute temperature at the thermocouple you need to measure absolute temperature at cold junction.

Klaus

 

[cupoftea]

Thanks, i guess a good quick check is to boil the kettle, dip the thermocouple wire end in and see if it measures about 100degC.

 

[betwixt]

The boiling water test might work but you would be measuring 100C vs ambient temperature rather than 0C. Obviously, you can mathematically adjust the figures but to measure accurately you have to at least subtract the measured temperature at the 'cold' end of the thermocouple wires. If you are building this measurement system from scratch or incorporating it into a product, I suggest looking at the data sheets for the AD594 / AD595 which detail how it all works.

After reading your first post I demonstrated how thing could go wrong to a student here. I simply wrapped the two (type J) thermocouple wires around the probes of a DVM set to a low voltage range and it showed a 0V measurement. I then used a hot air gun to heat the thermocouple junction to about 120C and the meter reading went up as expected. I then let it cool down and the reading dropped back to zero. So far, so good. But then I heated nothing but one of the DVM probes and again the reading went up, I let it cool and then heated the other probe, the reading went negative. So you can see the voltage is generated whenever the two wires join to some dissimilar metal.

Those AD594/5 ICs have built in thermometers that compensate for ambient temperature, effectively making the connection to the thermocouple wires appear to be at 0C. You still have to be very careful to connect the wires as close to the IC as possible to minimize the temperature difference between solder joints and the IC die and to use the same solder on both wires. Note that not all solder will readily bond to thermocouple wires.

Brian.

 

[cupoftea]

Thanks, we are using a TC-08 (Pico) thermocouple reader into the computer.

TC-08 Thermocouple data logger | Pico Technology

Measure and record temperature with the TC-08 thermocouple data logger. USB connected and powered. Supplied with easy-to-use PicoLog software

www.picotech.com

Its a few hundred quid and we hope it solves a lot of the thermocouple issues for us.

Thanks, your info on thermocouples is really good....i will store this away, and i am looking at these AD59X chips for future stuff.
As you kindly tell, there is a great deal can go wrong in thermocouple readings.

 

[FvM]

Thanks, we are using a TC-08 (Pico) thermocouple reader into the computer.

So you also have sufficient cold junction compensation, presumed you are using correct TC connectors.

 

[D.A.(Tony)Stewart]

There is exist a proper way to calibrate thermocouples. Room temp does not measure gain or offset errors exactly but it gives a warm and fuzzy feeling if it reads 21 'C and warmer fuzzy feeling if it reads 25'C

Example Temperature Ranges for Calibration:​

• Type K Thermocouple: Calibration points might include temperatures ranging from around -50°C to 1000°C.
• Type J Thermocouple: Calibration might cover temperatures from around -40°C to 750°C.

Always consult with manufacturers calibration procedure.

When done, typically, the accuracy of a Type K thermocouple is within ±1.5°C or ±0.25% of the measured temperature, whichever is greater, over the temperature range of -40°C to 1000°C.

 

[oscillationeer]

Hi,
If you are about to do thermal tests with thermocouples...then to see if they are ok...can you say they are ok, if when in the thermocouple reader, with the end junction bits loose on the lab table, then if they read room temp properly , then those thermocouples are OK? (at all other temps too). We only need accuracy to within about 4 to 5 degrees.

Any standards-compliant thermocouple wires will give good enough accuracy for your '4 to 5 degrees' specification,
but your 'thermocouple reader' and extension wires, and thermocouple in combination, can NOT be tested adequately
with only one temperature measured. The best practices for thermometry include regular checks for calibration
of readers, and resistance monitoring to ensure no breakages occur, as well as certification of the
wire and insulation materials.

In particular, a reader at ambient temperature will report ambient temperature with a short across
the thermocouple terminals, which short could be ANY wire whatever, or with polarity of the
thermocouple backward, or with a good thermocouple. That test is too easy to pass.

 

[cupoftea]

Thanks, with the TC-08 and out proper thermocouple connectors, that go into the TC-08, it seems that if it reads room temp correctly, and then the temp starts increasing when the fet heats up, then it is highly likely that the reading will be correct to within 4 to 5 degrees C, you would agree?
Taking it that the TC-08 is in good calibration.

 

[cupoftea]

Unfortunately, we can only do one channel with the thermocouple directly glued to the metal tab it seems, from the TC-08 documentation.

 

[FvM]

You need to asure that the TC is electrically isolated from your circuit. When using glue, this can be best achieved by applying a thin glue layer, curing it and then glue the TC. Most easily with UV-curable glue.

 

[D.A.(Tony)Stewart]

In particular, a reader at ambient temperature will report ambient temperature with a short across
the thermocouple terminals, which short could be ANY wire whatever, or with polarity of the
thermocouple backward, or with a good thermocouple. That test is too easy to pass.

i.e. Calibrate it near max temp, not room temp. e.g. 100'C in boiling water might be a good test.
Type K thermocouple is within ±1.5°C or ±0.25% of the measured temperature, whichever is greater.

so it must be within 1.5 °C
Kapton tape is the std. adhesive for tips with a multi-channel recorder.

 

[cupoftea]

We have an SOT223 which has 18V on the tab. We will glue the thermocouple to it as you kindly say.
The TC-08 is good to 30V, so it'll survive
We will just have to turn off the power supply when we take the thermocouple reading , otherwise we will read garbage.
The idea of a thin film of insulating glue first sounds good though...but it seems we'll get better contact if we just glue the thermocouple right to the metal tab.

Also BTW, as a rule, how long a length of thermocouple wires do you strip a type K thermocouple? Eg, 5mm with all that 5mm of wires twisted together is optimal?: Then glue that whole 5mm twisted thermocouple junction to the SOT223 or DPAK tab?

 

[KlausST]

Hi,

The TC-08 is good to 30V, so it'll survive

You should know the difference between
* Absolute maximum ratings -> not necessarily operating properly.. but destructive beyond the limits
* operating conditions --> which are given as +/-7.5V

***
Also nothing says that the channels are isolated against each other.
And nothing says how the "common mode voltage" is driven internally.

--> I would not rely on it to work with any external connection to the thermocouple signals

Klaus

 

[cupoftea]

Thanks, i see your point...our circuit is isolated and its ground is not the same ground as the TC-08 thermocouple reader. As such, with our thermocouple actually touching the transistor's metal tab, we can only get a thermocouple reading every now and again by momentarily turning off the power supply to the equipment......then the metal tab is not "electrified", and is just a "piece of metal"...when we get the reading, we turn back on the equipment so the transistor keeps getting hot, then we turn off again in the near future...and take another reading....etc etc

 

[betwixt]

Be careful here, two points you raised:
1: "how long a length of thermocouple wires do you strip a type K thermocouple?"
If you mean to extend the thermocouple wires, a good electrical connection is all you need but the wires MUST be the same as the ones going to the tip or you produce another source of error voltage where they meet. DO NOT twist two wires together to make the sensor tip, they should be welded (not soldered!) to give consistent results.

2: "we can only get a thermocouple reading every now and again by momentarily turning off the power supply to the equipment..."
Touching the thermocouple to another metal will produce an error. The voltage is produced at the barrier of the two thermocouple wires, adding a third metal contact, even if it is isolated will introduce an error. You should aim for a good thermal contact but keep it electrically isolated. A quick and cheap trick is to place a small blob of heat sink compound on the component and dip the thermocouple in it. The compound will make some change in the heat distribution but with care it can be quite small and you can wipe it clean afterwards.

Brian.

 

[cupoftea]

Thanks, we cant afford a spot welder...we have indeed twisted the thermocouple wires together and have soldered the twisted end.
Without a spot welder, thats all we can do.
The temperature appears to be going up when we turn the power on.
I will put it on a solder iron set for 100 degrees and see if the thermocouple reads it right.

I have seen docs saying soldering was "not the best way", but acceptable.

Sorry to confess, but Most of the places where i have worked solder the thermocouple ends together.....if you asked for a spot welder you would be ridiculed...they would insist on proof that soldering them did not work.

f you mean to extend the thermocouple wires,

No sorry, i mean the stripped bit at the end of the sheathed thermocouple wires

 

[cupoftea]

Put a soldered thermcouple into a drip of solder on a weller iron tip.
..then cooled it to 110degc on solder units temp indication.....thermcouple TC-08 reader said 95degc....when i yank it about i sometimes get it nearer 110degc.
When i warm up to 207degc, the TC-08 says its 160degc.
Then i repeat it with a better soldered thermcouple end...and i get 100 degc when weeler says its 106degc.

Then i get a spot welded thermocouple and press it against the solder drip.......and sometimes its measuring same as weller...sometimes its 30degs less.
I am wondering about thermocouples sometimes.

I was about to throw the spot welded thermocouple away...then i tried it again against the 110degc drip of solder on the iron...and it read ....110 degc.......its difficult to get repeateable readings......theres another metal at play here...but it should be same temperature of both thermcouple metals, so it shouldnt matter.

 

[KlausST]

Hi,

I am wondering about thermocouples sometimes.

Be sure it´s not the problem of the thermocouple.
A thermocouple works rather reliably.

It´s the "outside" that causes errors on the measuremnt.

Klaus