$4 Thermocouple TM902C quality and accuracy?

[kalel]

Likely the cheapest temperature meter is the TM90C with K type Thermocouple.

Price is around $4 (slightly less) with free shipping and probe.

Some mentioned points:
Specifications:
Measurement range: -50°C to 1300°C
Resolution: 1°C
Accuracy:
0°C to 500°C: +(0.75%+1°C)
500°C to 750°C: +(1%+1°C)
0°C to 20°C: +2°C

I found some images of the inside by searching it on this forum, but not much of a discussion or review (by searching the name). Since now it's an old thing already, that model at least, I guess more people have had it and used it, and perhaps even tested the accuracy/calibration (I guess calibration can be vary from unit to unit however).

If you have used it, I'd like your opinion about the quality and accuracy. It is the cheapest unit I could find, but I'm just wondering what can be expected (is it one worth purchasing for general use).

[MosherIV]

If you have used it, I'd like your opinion about the quality and accuracy.

Quality - poor. The on/off switch actually does not disconnect the battery. Search the web and you will find someone has posted instruction and video on how to do a modification.

I have no complaints, it matches all my other thermacouples (mostly DMM).

[kalel]

If you have used it, I'd like your opinion about the quality and accuracy.

Quality - poor. The on/off switch actually does not disconnect the battery. Search the web and you will find someone has posted instruction and video on how to do a modification.

I have no complaints, it matches all my other thermacouples (mostly DMM).

A little bit ouch about the battery. Although it depends on the current drawn, perhaps it's something that makes it last forever. Anyhow, if the switch is the biggest issue, I guess it might be well worth the money.

Is this the right video?

- It seems a bit complicated, but as some have proposed in comments there might be a simpler solution

Here's one comment from YouTube:

Based on your video I immediately tested it with my Fluke 87 IV and measured the standby current of the TM-902C in the off position.   It was in the micro amps draw range.    I have left the battery in and there seems to be no issue of excessive battery drainage

I guess this might be different for different units, but... hopefully it isn't.

[MosherIV]

Yes, that is the video.

I think the mod is tricky because the switch is left being held in place by just 1 pair of solder lugs.
When I did it, I was scared I knacked my one because the readings were all over the place just after I did the mod. I realised that the circuit was still hot from the soldering.
I had to wait a couple of hours before the unit went back to normal and in cal.

The mod changes the switch from just un-biasing the transistor to actually isolating the battery - it is worth doing.
Or simply remove the battery when you are not going to use it for a while.

[Gyro]

Haha, the first time I've seen that video. Unfortunately it misses a fairly fundamental finding - the IC in the thermometer is actually designed for a 3V battery supply! 

I found this out by accident when the first one I bought smoked when I put the battery in (dead fake regulator). After a bit of investigation I found that the main IC needs 3V. It even has a low battery indicator on the display which cuts in as a pair of AAA alkaline batteries go flat - and before accuracy fails. The battery compartment is also perfectly sized to accept a 2xAAA battery holder.

One other point - there are two versions around, one with 1'C resolution and one with 0.1'C resolution. You want the latter of course. I did a writeup on here somewhere, I'll take a look later.

EDITS:

- Ah, ok. It wasn't much of a writeup, it's the last post in this thread, which includes a rather useful review and comparison test of multiple units:
https://www.eevblog.com/forum/testgear/tm-902c-cheapest-dedicated-k-type-thermocouple-thermometer-review/

- I have a pair of these units which I have calibrated to consistently match within 0.1'C (after decent settling time). The PCB includes a footprint for an adjustment pot as you can see in the PCB photos in this thread:
https://www.eevblog.com/forum/reviews/tm902c-digital-thermometer-info/msg184850/#msg184850

You can also see the SOT-223 regulator that I removed and bypassed on my units for 2xAAA battery operation (note the 'idiot diode' across the battery leads too). The SOT-23 package transistor, Q1 is used as the temperature sensor for the cold junction compensation. On mine I have also added adhesive copper foil around the thermocouple connector and back of the PCB to try to keep temperature differential between the connector and Q1 as low as possible.

I can't remember exactly, but I think I calculated that, by changing from a 9V battery to 2xAAA the battery life extended to something like 500 hours.

You can buy thermometers with dual thermocouple inputs fairly cheaply too but I find that having two of these is more useful... You don't have to worry about ground loops or how far apart the measurement points are.

[kalel]

I read your posts there, and wondering, what kind of battery holder for the AAA that fits inside did you use?

I have to contemplate the possibility of having to change it as well if I get this unit - because as you mentioned, not only is it less efficient (might be something I can live with to spend less time on modding), but the 9v battery can even burn it out, which is not so great.

[Gyro]

Hi kalel,

Just a basic ebay wire-ended one like this (many sellers):

http://www.ebay.co.uk/itm/1X-Battery-Storage-Case-Box-Holder-Leads-Fr-2xAAA-3V-Battery-with-Wire-EA79-/332221431192?hash=item4d59f0bd98:g:s8sAAOSw~FJZGWQK

With a little bit of foam behind it to stop it rattling.

The logic of whoever designed these things really eludes me! Why put an expensive 9V battery and a regulator in when the thermometer chip is actually designed for a 3V battery supply. I couldn't believe my eyes the first time I wound the voltage down and the Low Battery symbol popped up on the display (Edit: at the correct threshold)! 

Measuring against a lab grade total imersion mercury thermometer they're actually pretty linear, even at 0.1'C resolution. Most of the time I use mine for monitoring temperature rise rather than absolute temperature, so having two that track accurately is nice. You don't have to bother of course. Bypassing the regulator is easy, just snip its legs and bridge the appropriate pads. There's no need to mess around with the switch at all as the leakage current disappears along with the regulator.

[kalel]

Hi kalel,

Just a basic ebay wire-ended one like this (many sellers):

http://www.ebay.co.uk/itm/1X-Battery-Storage-Case-Box-Holder-Leads-Fr-2xAAA-3V-Battery-with-Wire-EA79-/332221431192?hash=item4d59f0bd98:g:s8sAAOSw~FJZGWQK

With a little bit of foam behind it to stop it rattling.

The logic of whoever designed these things really eludes me! Why put an expensive 9V battery and a regulator in when the thermometer chip is actually designed for a 3V battery supply. I couldn't believe my eyes the first time I wound the voltage down and the Low Battery symbol popped up on the display (Edit: at the correct threshold)! 

Measuring against a lab grade total imersion mercury thermometer they're actually pretty linear, even at 0.1'C resolution. Most of the time I use mine for monitoring temperature rise rather than absolute temperature, so having two that track accurately is nice. You don't have to bother of course. Bypassing the regulator is easy, just snip its legs and bridge the appropriate pads. There's no need to mess around with the switch at all as the leakage current disappears along with the regulator.

Thank you for that information, it's seems worth doing that mod.

[Nusa]

I had two of those tucked away, so I dug em out and sure enough the battery was dead in the one that had a battery in it.

I put one on the power supply and it was happy down to 2.3V input to the regulator. I will be modding them to the AAA configuration.

[Gyro]

While you've got it apart to solder in the new battery leads, you may as well chop the regulator too - then you'll get full Alkaline battery life (down to 0.8V/cell) and lose the off state leakage current too.

That's unless you plan to really bodge it by just cutting off the 9V battery clip and soldering to the wire ends that is! 

[Nusa]

Never fear, I planned to do it properly. I just didn't have a clue it was needed until I saw this thread.

[MosherIV]

there are two versions around, one with 1'C resolution and one with 0.1'C resolution.
Really

There is little point in 0.1'C resolution when the absolute accuracy of K-Type thermocouple is 1'C
https://en.wikipedia.org/wiki/Thermocouple
Look for the table near the bottom of page.
That means that any temp reading is + or - 1'C from the actual ! That is worst case. Unless you have determined the actual deviation, you will never be sure if it is actual +0.8'C or some other number.
The best resolution thermocouples are T-Types according to the table with 0.4'C.

The only use the fraction of 'C is - is if you want to know relative temp change.

FYI these cheap chinese copies are of an old 1980s model, check out this old thread :
https://www.eevblog.com/forum/reviews/how-well-do-cheap-thermocouples-work/
Look for this picture :


This original contains a lot more circuitry because it has the proper 'cold junction conpensation' circuit. There are modern Cold Junction conpensation chips around but they cost more than the whole cheap circuit!
I suspect they are using a transitor to monitor ambient temperature, hence when I did the mod the temperature read out was screwy for an hour until the SOT23 transistor returned to ambient temp.
Again read all about it on the wikipeia page or search for how thermocouples work and what the proper circuit should be.

(Having cold junction compensation still does not fix the absolute temperature accuracy of the thermocouple itself)

Saying that, I am amazed how well the bodged circuit works! 

[Gyro]

Wow, they have, err, cost reduced it a bit in the past few decades!  (I thought that case looked reminiscent of something I saw in Practical Electronics!). It's a shame you can't get hold of the originals any more, but as you say the bodged ones do work very well, considering. Actually It would be possible to move Q1 off the board and mount it on thin flying leads against the connector pins (with a bit of Kapton for insulation), I might try that.

True enough, you're not going to get 0.1'C accuracy but with two matched thermometers you can certainly resolve fractions of a degree... and get a reasonable calibration of an individual thermocouple over your range of interest, as long as it isn't too wide (assuming that you calibrate it). The original looks to have had adjustments for spot and slope though.

The one you show could well be the origin of the 9V battery supply (I don't see a regulator), but you'd think they would know the supply characteristics of the chip they are putting on the board now - it's a COB after all! The battery holder is an easy fit in the existing battery compartment, maybe they just never tried. 

It's impressive how many variants you can fit into one model number isn't it!

[Nusa]

Modification complete on my two units. All I did was:
1) removed the 9V connector
2) removed the 3V regulator and cleaned up pads
3) solder-bridged the regulator input & output pads
4) added wires from the 2xAAA battery holder
5) soldered the switch mounting tabs for better mechanical support

And that's all there is to it. I did the power supply test again. The battery indicator comes on at 2.3V. The display changes to "Lo" at 2.2V. New AAA batterys provide 3.3V, and that seems to be fine too. Obviously, much more than that would probably kill it with no regulator to protect it.

The regulator that was in it was rated up to 24V input...so the original configuration would've been great to hardwire into a 12-15V vehicle system, or similar scenarios.

[Gyro]

Glad you did it properly, not that I ever doubted you. 

The regulator that was in it was rated up to 24V input...so the original configuration would've been great to hardwire into a 12-15V vehicle system, or similar scenarios.

Assuming it was a genuine regulator! I suspect that the thermocouple is elevated from the battery negative terminal too. Your battery life should be so long now that it won't matter if you forget to turn it off for the odd week or so.


P.S. I'm pretty sure I tested mine up to about 3.6V but I didn't want to push it too far.

[LifeRider]

Hi Chris and all,

please don't hate me for digging out an old thread, but this is very specific.

You wrote here about putting a potentiometer to be able to calibrate it. I don't need 0.1 accuracy, but I'd like to move it when it's close to 1 degree away.

Please see the image of the board included in my copy. Would this version be suitable to put in a calibration potentiometer?

Thank you for your help - I have to admit that I did not pay attention to electrical circuits at school, and now I'm paying the price. The good news is that I'm enjoying the free courses on YouTube, and having fun at almost 40yo

Thank you, Daniel

[ocw]

I use either a $5.78 TE NB-PTCO-153 0.1 accuracy RTD or a $14.95 Adafruit 3984 0.15 accuracy RTD if I need a sealed sensor.

Admittedly, the Analog Devices CN0359 measurement/display electronics adds $131.25 to the price.  The attached picture shows how I use both to measure underground soil temperature and moisture.  The moisture is determined by measuring the AC conductivity/resistivity of Irrometer/Watermark 200SS sensors.  Both are four point measurements.

[Gyro]

Hi Chris and all,

please don't hate me for digging out an old thread, but this is very specific.

You wrote here about putting a potentiometer to be able to calibrate it. I don't need 0.1 accuracy, but I'd like to move it when it's close to 1 degree away.

Please see the image of the board included in my copy. Would this version be suitable to put in a calibration potentiometer?

Thank you for your help - I have to admit that I did not pay attention to electrical circuits at school, and now I'm paying the price. The good news is that I'm enjoying the free courses on YouTube, and having fun at almost 40yo

Thank you, Daniel

Welcome to the forum Daniel,

I see that evolution (and cost reduction) has once again been in action!

The PCB has indeed changed a lot - I see that they have now moved on to push button operation (presumably with timeout). They still seem to have left the option for a slide switch though (P20-22), presumably with a solder blob added to the pads marked SW too.

There is no longer a separate voltage regulator - I can't see what battery is on the end of the Red wires, whether it is 9V or they finally switched to the more sensible 2xAAA.

Unfortunately the ambient (unit) temperature sensing transistor is no longer there, so this is probably integrated into the IC under the blob, further from the thermocouple connector.

There also isn't any footprint for an SMD trimming potentiometer any more and so, no indication of which resistors were related to it for the non-potentiometer factory adjustment. It's possible that one or more of the resistors might affect the calibration but it would be pure guesswork as to which - some of their markings don't appear to correspond with the silkscreen markings.

I'm curious about the other three through-hole pads (P24-P25-GND). it's vaguely possible that they might be intended for a pot, but they could equally be some sort of serial port for on-chip factory calibration, in which case, the accuracy might be quite reasonable). Unfortunately there is really no way to know what's going on under the blob.

You could maybe try measuring the voltages on P24 and P25 relative to GND.

Sorry I can't be more encouraging.

[Nusa]

The C/F jumper looks interesting, at least for North American users. If you don't mind experimenting, it would be nice if tested to see if F mode actually works with the provided firmware.

[LifeRider]

Nusa, it is possible to convert this little unit to Fahrenheit. After I put a blob of solder across the pads marked CF, it switches to F. See the image attached (sorry I haven't got an image of the inside now).

Chris, when I searched online for GND P2.4 P2.5 all I kept getting was pinouts for little microchips, and so it seems that these are some sort of access points, I assume for talking to the chip. The voltage from these to ground is 3V. The SW pins when soldered together switch off the auto switch off function of the thermometer. I like it having that time out, so I keep them separate.

By the way, the units are powered by two AAA batteries, but only the ones with the red button. There still is a TON of stock out there with the black switch and a 9V battery.

[Nusa]

Thanks. Nice 3-d printed case.

The labels on the through-holes are probably port and pin number for the microcontroller under the blob. Port 2, pins 4 and 5 for the mystery pins. The 3V you measured means either the internal pull-ups are enabled or they are configured as output pins. And Port 2, pins 0-2 up by the switch/button area (the SW pads likely change the functionality from a button to a SPDT switch, but that might not be right). Experimentation is the name of the game, unless you find someone with the answers already.

[LifeRider]

Thanks about the case. I hope people like it. I'm crowdfunding this as a product (a long stem composting thermometer based around this unit)

The SW pads swith off the timed off function. By default the unit switches off after a few minutes. Soldering these two pads seems to disable that.

How do I go about hooking up these "P" pins? What tool should I learn to use?

Thank you

ps. Here's the campaign if you're curious (I don't expect you to be interested in the product): https://www.indiegogo.com/projects/compost-thermometer

[Gyro]

Chris, when I searched online for GND P2.4 P2.5 all I kept getting was pinouts for little microchips, and so it seems that these are some sort of access points, I assume for talking to the chip. The voltage from these to ground is 3V. The SW pins when soldered together switch off the auto switch off function of the thermometer. I like it having that time out, so I keep them separate.

By the way, the units are powered by two AAA batteries, but only the ones with the red button. There still is a TON of stock out there with the black switch and a 9V battery.

Well at least it should have decent battery life then - they've learned the lesson. It is a useful finding that you can disable the timeout, especially if you want to monitor something on an extended run. They can take a couple of minutes to stabilise down to the last digit (from comparison of my two matched 'slide switch' units).

If both signals are reading 3V, then yes,  my best guess would be that they are a factory comms port too.

Unfortunately, if you want a 'tweakable' one then you may have to go with one of the older style slide switch ones and hope that it doesn't share the same PCB as the push button version (entirely possible). The only differentiator is now whether it has a 9V battery, assuming that the ebay listings are up to date.

[LifeRider]

OK, so with this board the idea of the potentiometer is a dead end. No worries. I wonder if I can find something else here either before the current enters the IC, or with the pins that come out of it. Regardless, it's not too big of a deal. It's a nice to have. I advertise the readout error as stated by the thermometer manufacturer, and in my tests it usually is close enough.

Thank you for your help

[Gyro]

Thanks about the case. I hope people like it. I'm crowdfunding this as a product (a long stem composting thermometer based around this unit)

The SW pads swith off the timed off function. By default the unit switches off after a few minutes. Soldering these two pads seems to disable that.

How do I go about hooking up these "P" pins? What tool should I learn to use?

Thank you

ps. Here's the campaign if you're curious (I don't expect you to be interested in the product): https://www.indiegogo.com/projects/compost-thermometer

Sorry, I've no idea how to hook up the P pins, one may have a weak pullup resistor (Rx) and the other a stronger drive (Tx) but that's a pure guess. No idea about format.

Hmmm. I've just looked at your campaign page! My first reaction was How much?    but then I read about the long probes, so maybe those are expensive enough to justify the cost of £85.

It looks as you're just buying TM-902Cs (~£3 from China one-off including thermocouple and case), re-packaging the PCB in an Orange 3D printed case and adding the long probe.

I assume that you will  be dealing directly with the manufacturer to get best volume pricing on the PCB alone, so you are probably in a position to ask them the details of the port.

3D printing also isn't the way to go if you're planning to sell any volume - you'd be far better getting the case moulded in China. Edit: From the video, I understand your rationale.