JBC T245/C245 Thermocouple thermal coefficient

[sahko123]

So there seems to be little to no consensus as to the type of thermocouple used in the C245 cartridges for the JBC T245 handles.

I have seen some references to Copper-Nichrome with a seebeck Coefficient of 18.5uV/c and some mention of 40uV/c elsewhere with everyone questioning everyone else's measuring methodology.

I'm designing a base station for the Jbc T245 handle and wanted to have measure temperature with some modicum of accuracy. I finally got it decided to test the temp coefficient at work after hours using a

Keithley 2110 5&1/2 digit multimeter with Thermocouple input,
Keithley 2000 6&1/2 digit multimeter with 100nV resolution  ,
K type probe,
Metcal MX5000 soldering station,
Kapton tape,
and little confidence in anything I do ever.

The Keithley 2110 was used to measure the K-type probe which was strapped as close to the tip of the C245 cartridge directly (T245 handle not included). The voltage across the unknown thermocouple was measured simultaneously with the Keithley 2000. The system was initially set-up and then left to equalise over some arbitrary amount of time until it reached a stable state. I used this as the cold junction measurement which ended up being 24.14c with 30.3uV.

Then comes the hot part. I used the Metcal soldering iron to heat up the C245 tip and let the system equalise. The Metcal station does not display temperature only input power into the attached iron. After the temperature stopped changing at around 274c the voltage was 7510uV. The soldering iron and tip were connected with molten solder blobs to ensure optimum transfer.

Now it becomes interesting. I couldn't be arsed writing a script to log results and setting all that up in the soldering section of the lab with a loaner work laptop since it would have been too much of a pain especially since I haven't done much logging with meters. I decided instead to save time simply record what happened next with my phone pointed at the two screens of the meters. Then I simply turned the metcal iron off and moved it away and let the C245 cartridge cool passively in the lab while recording the readings on my phone using video. The cartridge cooled back down to room temp after about 11 or 12 minutes or so. After I got home I scrubbed through the video and added the data to an excel sheet and graphed it. here's the graphed data up to about 275c and then with a dotted linear trend line just to demonstrate linearity.



Afterwards I averaged the uV/c rating of each reading which came out to be roughly 40uV/c as someone stated previously in another post. the trend seems to be quite close to 40uV/c with roughly 7uV/c over the range measured 23c-275c. I believe this corresponds to a Nichrome-Nickel thermocouple which has a seebeck coefficient of a combined 40uV/c (25uV/c-(-15uV)).

I'm quite happy with this result and believe its good enough to use for fairly accurate measurements. Anyone have any objections?

[thm_w]

Nice work.

This would make sense to me as they sell a Type K thermocouple (41uV/C): https://www.jbctools.com/ph218-thermocouple-type-k-product-550.html
So only one lookup table in software is needed.

[SiliconWizard]

I've made a T210 controller so far, and for sure the coefficient is about 10µV/°K.
Not tried T245 yet, but from what I've gathered, it's almost certain it's about 40µV/°K.

In both cases, the thermocouple is an integral part of the cartridge design, so don't expect it to match any standard thermocouple exactly. You can find the JBC patent to see how this is constructed.

[sahko123]

I haven't actually thought to check the patent. Might be a bit of a pain to find though if it exists. I'll search through the website to see if there is any trace of one

[floobydust]

From the 341 page Unisolder development thread, I'd saved this graph someone did but can't remember what page. It's about 17.3uV/°C he gets and no idea if done properly.
JBC tip patent EP1086772 but no voltage info.
Hakko is 20-21uV/°C patent US6087631
I'm pretty sure the heaters use Kanthal wire but there are slightly different alloys depending on manufacturer.

[DavidAlfa]

Yes, T12 is around 21uV/°C, C245 41uV, and IIRC, C210 about 10uV

[floobydust]

The T12-clone controller boards supporting JBC option use an op-amp with Av=250 and 3.3V MCU such as STM32F103 etc. That would roughly give 13.2mV full-scale for 660°C/1,220°F for a 20uV/°C thermocouple, or half that for a 40uV/°C thermocouple 330°C/626°F - which I think is too small. Where's the miscalculation here?

[SiliconWizard]

I haven't actually thought to check the patent. Might be a bit of a pain to find though if it exists. I'll search through the website to see if there is any trace of one

The basic one is this: https://patents.google.com/patent/EP1086772A2/en

While it shows that the thermocouple is an integral part of the tip, it won't give you numbers. But at least you get to see how this is formed.

[sahko123]

[UPDATE]

I've done another experiment this time instead of measuring directly on the contacts of the cartridge I instead had the cartridge inside the T245 handle -> handle connected to its mating connector -> soldered wires to said connector -> and wires grabbed onto by alligator clips going into the same Keithley 2000 multimeter. Obviously this is a very un-ideal setup but that's essentially the point. I also had the C245 powered on its own (roughly 10W from a random power supply at work no PWM yet) instead of heating it up externally.

Shown below is the graph I end up with. Logging was done the same way and for the orange graph its only measured between 79.85c-372.62c. Blue=Test1 direct contact. Orange=Test 2 through several connections


This showed an offset between the two voltages. This is caused several hundreds of uV of offset. Also the heating must have travelled down to some other connections (most likely C245 to T245 handle) which caused a different slope so the offset changed over the temp range.

Because the thermocouple was assumed to be a k type from the measurements. For shits and grins I turned on the temperature measuring function of the Keithley 2000 to see if it measures the K type probe correctly. It did not.... I think. The reading was wildly different between the real k type probe (measured with Keithley 2110) and the 'k type' probe. The Keithley 2000 would read 210c when the Keithley 2110 would read 430c or there abouts. I think this is most likely caused by the large amounts of terminations/connections. But it seems like the measurements should be closer together no? Both meters have terminal cold junction compensation but its seems that either the difference was larger than it seems. My maths may also be off but I have no idea at this point. In the final product I will have an ambient temp sense on the terminals of the connector.

Should I trust the initial testing? Tomorrow I will try and probe temperature from the pins of the connector instead of attached wiring.

[thm_w]

The T12-clone controller boards supporting JBC option use an op-amp with Av=250 and 3.3V MCU such as STM32F103 etc. That would roughly give 13.2mV full-scale for 660°C/1,220°F for a 20uV/°C thermocouple, or half that for a 40uV/°C thermocouple 330°C/626°F - which I think is too small. Where's the miscalculation here?

What part is too small?
The gain on the T12 station here is 160-300 right: https://github.com/deividAlfa/stm32_soldering_iron_controller/blob/master/BOARDS/Quicko/Schematic/Quicko%20T12%20STM32.PDF

660°C * 20uV = 13.2mV.  13.2mV * 161 gain = 2.13V
3.3V / 161 / 40uV = 510°C

[floobydust]

If the JBC thermocouple output is greater than about 35uV/°C then all the existing T12-clone controller boards out there won't work with the JBC option selected. You'd think people would have noticed that. Because they all use a TC amp fixed gain of 250 and do calibration in firmware instead of a trimpot. The A/D would clip above 377°C/711°F
This is what makes me skeptical the TC is so high at 40uV/°C
These TC's are not at all off the shelf i.e. type K, they are what you ended up with after constructing the heater/tip . There are many TC junctions actually in the metallurgy, and nichrome/Kanthal wire are two of them.

OP when making TC voltage measurements, any dissimilar metals in the two test leads will make an extra unexpected TC and can corrupt your results. You have to have identical test leads, clips. probes for the EMF's to cancel out. Also the heater current when on will cause a voltage drop in the ground wire and an extra hot spot in the tip, which can give bad results. Ideally you'd measure TC voltage 100's msec after turning off the heater. There is a dynamic tip temperature aspect as well. Or use something else to heat the tip to a known high temp.

[sahko123]

Firstly I di t know why people are talking about the t12clones I don't really care about them and am designing my own controller.

I'm aware of the effect of dissimilar metals on readings of thermocouples. I just expected the effect to be different to how it turned out to be. When measuring the temp no power was applied to the heater element.

[gabiz_ro]

... when making TC voltage measurements, any dissimilar metals in the two test leads will make an extra unexpected TC and can corrupt your results.

But in real usage scenario we have contacts between tip and handle (inside),contacts between metal and wire (inside handle) , contacts between wire and metal inside plug,contacts between plug and socket (few extra if use extenders) , contacts from socket-wires-PCB.
But I think some maybe compensate each other, in that case if identical test leads also compensate each other TC resulted in contacts.
metal_x (test lead) - metal_y (tip) ---TC--- metal_y (tip) - metal_x (test lead)
unexpected TC ---TC--- unexpected TC
ex
-100uV +400uV +100uV (at some temperature)
Is that right or I'm wrong?

[sahko123]

The hope is that these junctions don't change much in temperature so the offset stays constant and can be easily calibrated out. But because inherently the device self heats it becomes more difficult so the actual temp Co response changes and instead of 40uv per degree you might end up with 37uv because contacts heat up from current passing through or from heat travelling from the cartridge

[floobydust]

If no one has the T245 thermocouple characterized, I look to other T245 controllers that people have made, including the T12-clones which claim to support JBC.
Trying to get even a ballpark on the TC voltage. I see two different T245 thermocouple sensitivities people measure, and design in for hardware. They must be making a mistake wiring it or measuring TC voltage with the wrong blanking interval etc.

JBC Soldering Station CD-2BC - Complete Schematic & Analysis uses OP07C/LM358 Av=245 3.3V ADC, as well as cancelling GND wiring drop.
https://github.com/Muny/DIY-JBC-T245-Station uses MAX31856 and linear curve fit.
https://hackaday.com/2017/11/06/roll-your-own-jbc-soldering-station/ uses MAX400 op-amp and Av=401 5V ADC.  files

"No T12 station will work correctly with the C245". OK, got that confirmed.

pic of a tip internals taken from Dangerous Prototypes forum topic: Really universal soldering controller page 210

[analityk]

I have compare temperature reading between UTi720A thermal imager and Siglent SDM3065X and from siglent menu i have pick thermocouple NITS90 - and readings between UNIT and Siglent were this same (+-4K). Any other Siglent thermocouple setup doesn't give any copmarable resutls between thermal imager and siglent. 
Temperatures was this same from -20 to +250 deg. C so i can't be fluke.

Outer/Green: Heater-/TC-
Middle/Blue: Heater +
Inner/Red: TC+

Tip JBC C245-765 (228049) have NITS90 TC.

https://srdata.nist.gov/its90/download/type_n.tab
https://www.wika.com/media/Technical-information/English/ds_in0023_en_co.pdf