Need advice on multi channel temperature logger with disposable sensors

Hello dear all,

I'm planning to build an Arduino based temperature logging system for an academic research project. The goal is to measure temperatures inside a composite material while it is being processed. Due to the nature of the process and the fact that resins are going to be cured and hardened, i'm planning to use cheap thermocouples or thermistors from aliexpress and dispose them with the test specimen after each experiment.

I'll be needing:
-At least 1°C accuracy
-Cheap sensors
-Measurements every 5 seconds or longer
-0°C-400°C operating range(maybe reduced into 250°C if limited by sensors)
-Sensor cables around 50-100cm

My questions are:
-If i buy let's say cheap 100K thermistors and build a look up table for the sensors i bought, would that look up table be valid with enough accuracy for all the sensors of the same batch with only just a simple room temperature calibration at the device start-up?
-The same question above for K or J type cheap thermocouples?
-Do you know any kind of interface/shield for connecting multiple thermistors/thermocouples to an arduino with the required circuitry inside, which is a resistor for thermistors and an op-amp for thermocouples?
-Do you recommend any solution out of my current scope?

Thanks a lot!

Is your material moving, or is it locked in place?

Paul

For me, the DS18B20 is cheap. If you're buying them by the thousand then they are very cheap.

It depends if you are going to exceed the DS18B20's maximum temperature. When it exceeds the max, it simply stops outputting data. It knows how hot it is. (Most other components will work outside their published boundaries. This one doesn't.) It's also a relatively bulky device.

I think there are some manufacturers of PT-100 thermistors who claim that their thermistors are interchangeable. For most, even thermocouples, you do need to 'zero' them to one known temperature such as room temperature. For sensors to be bonded into a layup, you may or may not have an opportunity to gather them all together and plug them in to grab that zero calibration. It would be a significant step of extra work.

MorganS:
For me, the DS18B20 is cheap. If you're buying them by the thousand then they are very cheap.

It depends if you are going to exceed the DS18B20's maximum temperature. When it exceeds the max, it simply stops outputting data. It knows how hot it is. (Most other components will work outside their published boundaries. This one doesn't.) It's also a relatively bulky device.

I think there are some manufacturers of PT-100 thermistors who claim that their thermistors are interchangeable. For most, even thermocouples, you do need to 'zero' them to one known temperature such as room temperature. For sensors to be bonded into a layup, you may or may not have an opportunity to gather them all together and plug them in to grab that zero calibration. It would be a significant step of extra work.

DS18B20 seems useful, but i checked all the temperature sensors of maxim and they dont measure temperatures higher than 150C.

I can first plug the sensors in, do the zero calibration and than place them into the mold, it's okay.

Paul_KD7HB:
Is your material moving, or is it locked in place?

Paul

They will be stationary inside a mold, under a hydraulic hot press.

I think thermocouples are the only sensible way to handle 400C.

You'll need a shield - see eg the MAX6675, and probably an analog mux. Better than relays, which can give a contact EMF.

I've use the MAX6675 shields, and they work fine - but a word of caution. In strong electric fields - ie close to mains - they can give erratic readings. Add a 100uF capacitor between the thermocouple -ve input and ground to cure this.

If you buy 'bare' themocouples - normally type K for this sort of job -they're not too expensive. and have a low thermal mass and hence fast response. If you want to do this on the cheap, buy a length of thermocouple extension lead, cut to length, and melt an end together with oxy-acetylene. I've done this out of curiosity, and it worked well.

And, unlike cheap thermistors, thermocouples are very reproducable from unit to unit.

Allan

+1 for thermocouples.

For 1° accuracy 0-400 look up table is a lot of SRAM.

What controller are you using?

I'd put lookup tables in flash - why waste scarce SRAM?

If you're taking a lot of readings I suggest you stream them as raw data to a PC running a terminal program eg RealTerm using serial.Write().

Save to a file and import into eg Excel for further processing .

Allan

Had a quick look on ebay....

4 quid for 10 metres. Good for 500C

So if you cut it into 10 that's about 40p a pop plus a few seconds of oxy-acetylene.

If you haven't access to the kit, visit a local backstreet car repair garage.

Or you could always cut off the embedded bit after your experiment and re-melt the new end.

Allan

allanhurst:
I think thermocouples are the only sensible way to handle 400C.

You'll need a shield - see eg the MAX6675, and probably an analog mux. Better than relays, which can give a contact EMF.

I've use the MAX6675 shields, and they work fine - but a word of caution. In strong electric fields - ie close to mains - they can give erratic readings. Add a 100uF capacitor between the thermocouple -ve input and ground to cure this.

If you buy 'bare' themocouples - normally type K for this sort of job -they're not too expensive. and have a low thermal mass and hence fast response. If you want to do this on the cheap, buy a length of thermocouple extension lead, cut to length, and melt an end together with oxy-acetylene. I've done this out of curiosity, and it worked well.

And, unlike cheap thermistors, thermocouples are very reproducable from unit to unit.

Allan

Cheap reproducability is a good thing, thanks. Regarding that matter, i've also considered buying 200-300 meters of T type thermocouple wire and weld/crimp them together.

For the analog mux thing, i can buy seperate max6675 shields for each channel so i dont need to rely on an analog mux.

larryd:
+1 for thermocouples.

For 1° accuracy 0-400 look up table is a lot of SRAM.

What controller are you using?

I'll probably use an arduino mega or uno. Yes 0-400 look up table would take lots of SRAM but i can also put the lookup table in the progmem.

allanhurst:
I'd put lookup tables in flash - why waste scarce SRAM?

If you're taking a lot of readings I suggest you stream them as raw data to a PC running a terminal program eg RealTerm using serial.Write().

Save to a file and import into eg Excel for further processing .

Allan

+1. I plan to save the data in .csv files on an SD card and later process with Excel or similiar software, to make the system more portable. The hydraulic press that will be used in the experiment is not something that you would like to keep your computer near.

Arduino programming part is not a big matter in this project as far as i've been able to predict.

allanhurst:
Had a quick look on ebay....

0-500c Temperature Range 3mm Width Thermocouple Extension Wire 33ft for sale online | eBay

4 quid for 10 metres. Good for 500C

So if you cut it into 10 that's about 40p a pop plus a few seconds of oxy-acetylene.

If you haven't access to the kit, visit a local backstreet car repair garage.

Or you could always cut off the embedded bit after your experiment and re-melt the new end.

Allan

Thanks for that but the type of the pair isn't certain on the listing.. If i fail on achieving good welds, could i just solder or use small aluminum clamps/crimps at the ends?

Also on a quick look on aliexpress, i might just switch to a proper multichannel thermocouple temperature logger device, but my uncertainities on thermocouples continue:

  1. Am i right that thermocouples generate a potential difference which is a function of almost only the connector side junction temperature and the sensing side junction temperature? I mean, the parts of the thermocouple cables between junctions will probably be passing from other places inside the specimen with different temperatures, including other measurement points. Would i still get a reading that is only dependent on the temperature of the end joint?

  2. Can i just buy hundreds of meters of wire pairs of certain thermocouple types(for example i have T type in my mind), cut them in desired length, make a junction by crimping/welding/soldering(soldering is a bit controversial but i've read somewhere that as long as the solder bead isnt exposed to a temperature gradient it would be okay.) and expect it to generate voltage differences equal to other ready made thermocouples of the same material pairs?

3)Let's say i've bought 100 thermocouples of extra pure grade materials with no measurable inhomogeneities, and i have a zero noise system that can read the voltage in a thermocouple with 100% accuracy+precision. Would all these thermocouples give exact same temperature readings when given enough time to settle or would the junction geometry and some other factors affect readings?

Years ago we assembled thermocouples as part of a project for a customer. They were spot welded by a Seattle company. About 90% were good!

Paul

I don't think oxy is the correct way to weld the thermocouple tip. You dont want the metals mixing. The Seebeck voltage is generated where two dissimilar metals touch. An impact welder or even a hammer can do it.

I would not buy "extension" wire either. It is made to a different standard to the thermocouple wire. But it probably won't matter for your purposes.

Paul_KD7HB:
Years ago we assembled thermocouples as part of a project for a customer. They were spot welded by a Seattle company. About 90% were good!

Paul

The question is, what was wrong with the remaining 10%?? Were they giving noisy readings because of a bad joint? Did they just simply not work? Or they gave highly inaccurate readings? if welding process does/might introduce inaccuraties that is very bad.

MorganS:
I don't think oxy is the correct way to weld the thermocouple tip. You dont want the metals mixing. The Seebeck voltage is generated where two dissimilar metals touch. An impact welder or even a hammer can do it.

I would not buy "extension" wire either. It is made to a different standard to the thermocouple wire. But it probably won't matter for your purposes.

In what ways they differ from thermocouple grade wires? Do their temperature response change or is it just durability etc. I'll take that advice on not welding but making them touch firmly.

if you plan on cheap, and leave in place or discard, then get the thermo couples and learn how to spot weld your own. very simple. lots of microwave transfrormer spot welder DIY links.

get a 1 meter wire, do your test, cut the wire as close as you can, strip the ends and weld, do it over and over until you used up to much of the wire to make more.

I had a guy near me that used to do it. takes all of 10 seconds.

klazzera:
The question is, what was wrong with the remaining 10%?? Were they giving noisy readings because of a bad joint? Did they just simply not work? Or they gave highly inaccurate readings? if welding process does/might introduce inaccuraties that is very bad.

In what ways they differ from thermocouple grade wires? Do their temperature response change or is it just durability etc. I'll take that advice on not welding but making them touch firmly.

Sorry, the other 10% consisted of fell apart, easily pulled apart, all were because of improper welding. And no inspection. I guess the price reflected the bad quality.

Paul

Hi…

to answer your questions…

1/ Correct.

2/ Yep - get a length , cut it up and join the ends . Soldering is no good - would melt at 400C even if you could tin the wires properly , which I doubt.

Silver soldering or brazing might work. Or just fuse together with a hot flame as suggested. Simpler.

Look at a commercial product - it’s obvious that’s what they do. Done it myself. It works.

Haven’t tried spot welding.

MorganS is wrong. It doesn’t matter that the alloys of the 2 wires mix, or even if there is an intermediate metallic conductive substrate at the join. Even in a spot weld there must be some mixing or it wouldn’t ‘stick’.

3/ Correct. The potential is entirely a property of the two alloys used in the wires. Time constant depends on geometry.

Allan

klazzera:
In what ways they differ from thermocouple grade wires? Do their temperature response change or is it just durability etc. I'll take that advice on not welding but making them touch firmly.

I can't find the original reference where I read this. I think it comes down to the insulation used. An extension is usually outside the furnace so it doesn't need to withstand furnace temperatures. The wire inside is identical for a thermocouple or an extension.

Hi,
Interesting reading;

https://au.rs-online.com/web/generalDisplay.html?id=infozone&file=automation/good-practice-in-extending-thermocouple-cable

Tom... :slight_smile:

Thanks for that but the type of the pair isn't certain on the listing.. If i fail on achieving good welds, could i just solder or use small aluminum clamps/crimps at the ends?

If you're dealing with hydraulic presses I'm sure there'll be loads of technicians about who could do the welding for you.

Show them a commercial item and they'll know what to do.

Skilled technicians are wonderful and can do things you'd never believe - I've known quite a few in my time in industry.

Allan

And there are spot welders made specifically for welding thermocouples.

Paul

Not to my knowledge - and why bother?

Look at a commercial product. The ends are obviously fused in a hot flame. Done it myself - see above.

Allan