How well do cheap thermocouples work?

[TheBorg]

I've been thinking about getting a cheap thermocouple to measure my hot air rework station output, and probably use it for measuring PCB heat while reworking. Are cheap thermocouples even semi-accurate? I don't need really precise measurements, only within 10*C.

Cheap thermocouples are everywhere, I mean, this one is $6 on eBay.
http://www.ebay.com/itm/Digital-LCD-Thermometer-Thermodetector-Meter-TM-902C-K-Type-Thermocouple-Probe-/251531789729?pt=LH_DefaultDomain_0&hash=item3a907685a1

[Richard Crowley]

Well, if you need ~2% accuracy and the gadget claims 1% that would appear to be suitable.
Perhaps in your application, REPEATABILITY is more important than absolute ACCURACY?

Cheap thermocouples appear to be made from "extension grade" wire vs. actual "thermocouple grade" wire.
http://en.wikipedia.org/wiki/Thermocouple#Grades

It doesn't seem like it would be a big issue to me for hobby-grade experimentation.

I would like to see a tear-down of one of those things.  I wonder where they took short-cuts to make the cost so low.
Or maybe the technology is sufficiently mature that it really isn't that expensive.

[mij59]

Hi,

Electronupdate has done a tear down.


The cold junction compensation, is on the pcb, it should be mounted on the thermocouple connector.

[Richard Crowley]

Electronupdate has done a tear down.


The cold junction compensation, is on the pcb, it should be mounted on the thermocouple connector.

Ha!  mij59 posted a teardown video even as I was adding that comment to my post!      Thanks!

I didn't really see anything that looked like cold junction compensation, or anything that looked like a sensor for it.  Unless maybe that Q1 SMD transistor is being used for cold-junction measurement?  But you're right, it seems pretty far away from the actual junction, where the copper wires connect to the (presumably) proper-alloy connector.  And I wonder what is the Seebeck constant of those blobs of solder? 

[mij59]

Electronupdate has done a tear down.


The cold junction compensation, is on the pcb, it should be mounted on the thermocouple connector.

Ha!  mij59 posted a teardown video even as I was adding that comment to my post!      Thanks!

I didn't really see anything that looked like cold junction compensation, or anything that looked like a sensor for it.  Unless maybe that Q1 SMD transistor is being used for cold-junction measurement?  But you're right, it seems pretty far away from the actual junction, where the copper wires connect to the (presumably) proper-alloy connector.  And I wonder what is the Seebeck constant of those blobs of solder?

I have attached a photo of  my 25 years or so meter, the resemblance is striking.
The cold junction temperature sensing transistor is in good thermal contact with the thermocouple connector.





 

[Richard Crowley]

I have attached a photo of  my 30 years or so meter, the resemblance is striking.
The cold junction temperature sensing transistor is in good thermal contact with the thermocouple connector.

Wow. It seems pretty clear that is the unit they are imitating.

Here is a photo (from the Wikipedia article on Thermocouple) showing how Fluke did it.
Large thermal mass connections, large thermal mass PCB pads, and a sensor thermally glued right between the terminals...

[mij59]

I have attached a photo of  my 30 years or so meter, the resemblance is striking.
The cold junction temperature sensing transistor is in good thermal contact with the thermocouple connector.

Wow. It seems pretty clear that is the unit they are imitating.

Here is a photo (from the Wikipedia article on Thermocouple) showing how Fluke did it.
Large thermal mass connections, large thermal mass PCB pads, and a sensor thermally glued right between the terminals...

Yes that's how it's done.

Checked the day code on the  ICL7106 it's only 25 years old.

[TheBorg]

I'm mainly looking for reliability and repeatability. Looking at the reviews of these it seems that they might be off a few degrees but that is fine as long as it is within a few degrees. My main concern is that the thermocouple wire might not stand up to 250C. If it can then I'll just go for the won hung lo version.

[Richard Crowley]

250C seems pretty modest for K-type thermocouple, which is long-term rated up to 1100C.

[mariush]

The type of insulation on the thermocouple and how the tip is protected restricts the maximum temperature.  Cheapo thermocouples have insulation that starts melting above 250c and those probes rated for 800c+ usually have the tip in a steel sheath for protection.

[retrolefty]

I'm mainly looking for reliability and repeatability. Looking at the reviews of these it seems that they might be off a few degrees but that is fine as long as it is within a few degrees. My main concern is that the thermocouple wire might not stand up to 250C. If it can then I'll just go for the won hung lo version.

 Thermocouples are probably the most reliable temp sensor there is, being just two dissimilar insulated wires welded together at the sensing end. RTDs are the more accurate sensor but less rugged. Thermocouple wire maximum temperature is determined by the type of insulation used. I've used TC wire that used special mineral insulation that could handle a couple of thousand degrees F.

[charlespax]

I've been working with K-type thermocouples quite a bit lately. For your needs even the cheap ones should be fine. I would stay away from the cheap thermocouples with the stainless steel probes. I've had a few internal solder connections break in these.

If you're looking for a temperature datalogger, check out my project at paxinstruments.com. In September I'm kickstarting an open source four channel thermocouples temperature datalogger. Please sign up if you'd like to know when the kickstarter begins.