Build or buy? Multi-level vessel temperature logger

[paulca]

"A stainless steel stick with some form of temp sensors distributed along it's length."

* Must be food grade
* Must be tolerant to T>=0<=105*C
* Must be tolerant to high concentrations of:  Chlorine, mild acids, mild bases, alcohol.
* Must be cleanable/sanitizable - ideally detachable and completely immersible.
* Accurate to 0.1C +/- a bit.

The circuitry for the logger I can cover myself I think.

What I'm curious about is....   has anyone seen a commercial product like this anywhere?

Also, if I am to make it what is the best sensor type to go for?  Using "foil" or wire tip thermocouples isn't going to be very durable.  (Thinking of cleaning and storing, think kitchen, not lab).

If I was to make a crude as hell prototype, just to give you the idea, I could take 6 standard curve "stainless steal kitchen thermocouple probes" and lash them to a stainless steal rod at various heights.  Then connect all 6 to the logger board.

While this would work.  It would be difficult to clean with the lashings, handles and cables for bacteria to live.

[BennoG]

what about a stainless steel hollow pipe and put the sensors inside, and the pot the whole thing.
1 side you can weld close and the other you put some pt1000 sensors in or even better some DS18B20 or so (then you have only 3 wires coming out of the tube)
then fill the whole thing up with some food-grade potting and you are done.

Benno

[mianos]

The term for the tube with a closed end is call a "thermowell". Normally they are stainless. The ones for brewing can be very long, 30cm. You can pick up all sorts of them from aliexpress.
Then, as BennoG says, I'd stuff a bunch of probes down them. Personally I'd much prefer proper 3 wire pt100 probes, you can also get them from "funny bunny"'s store.
He has the ones without covers, they are pretty easy to work with mechanically as the junction is only a mm or so. That would be very easy to wire and stuff.

This said, the normal MAX31865 is an spi device and you only get one probe per chip.  (I have good multiple board esp-idf drivers if you go this way and want them).

Your other option, DS18B20 would be much easier, they can all be run straight to your MCU with no driver board, they are quite precise. With .5 degrees of a PT100 based on my experience using them quite a bit. (They need pullups btw). The biggest down side would be getting a lot of them in a thermo well. 

ps, 0.1 is quite easy for the pt100. That's why I aim for.

I also have a pt100 on a bonded teflon tube. You could just fix these together and pull them apart for cleaning. Food safe, acid safe to the max. Not safe amongst forever chemical fear-mongers, unless you want top go past 550.

[voltsandjolts]

If you're looking for a project, using optical fibre to get a distributed temperature measurement might be more interesting than multiple rtds.

[max_torque]

Aside: This ^^ has been done many times!  I use an optical fibre temperature sensing array to measure various temperatures inside a running high voltage permanent magnet machine. Works by firing a lazer down the fibre and measuring the phase shift that is driven by a temperature dependant scintilating end "reflector"



Back with the OP, whats best probably depends on how dynamic the system is going to need to be, ie how quickly temperature changes and how fast it must repond.  For slow systems then a thin pcb with NTC resistors (or serial output temp sensor ICs etc) along it's length could be easily sealed into a length of food grade silicon tube pretty easily.  That silicon covering is going to limit thedegree of isotropic conformity to it's surrounding as compared to a more thermally conductive protection medium (ie stainless steel etc)

Again, you need to consider just how precise, accurate, responsive and complex the measurement device actually needs to be!

[mag_therm]

https://www.watlow.com/products/Sensors/Thermocouples/Multipoint-Thermocouples

[paulca]

Yea, then it only has a "Request a quote" button, it's too expensive.

Interesting.  So for the DIY approach, a silicon or stainless tube. 

8mm ID (which I have) will just about take a TO-92, so it will take DB8120b 'one wire' senors, requiring only a pair of wires for the whole "bus" of sensors.

I can probably very easily layout the apparatus without any PCB needed.  "Pull it through" the pipe/tube and then, gravity fill the tube with epoxy.

The only bit that sounds like it will go wrong is the last part.  I don't have any thing resembling a vacuum to pull the expoxy (or silicon sealer) through.  Not unless I get really runny, designed for the purpose epoxy.

EDIT:  I also need to make sure the epoxy (and everything) is okay at 105C.

What if I filled it instead with something liquid, non-conductive and then expoy shut the two ends?  (Can you solder shut stainless with hot air?)  How runny is "potting" compound?

[mag_therm]

Food grade, you might have standards for this, for example if the thermowell breaks, leaks  etc. I was thinking of suggesting silicon oils but that may be toxic.
On the other hand I was reading recentlty that FDA in USA  does not actually prohibit machinery bits and pieces rats mice cockroaches etc.

They only prohibit that stuff from exceeding a certain ppm. !

[voltsandjolts]

Aside: This ^^ has been done many times!

Yes, it's been done many times.
We use it several km underground and it works well.
But I haven't seen any low budget options, which might be an interesting project for a multi-level vessel temperature logger (which is the thread title btw).

[jbb]

Please note there are a lot of knock off DS18B20s out there. They might appear to work fine at room temp and then be wildly inaccurate at lower or higher temps.

[mianos]

If you are just stuffing DS sensors down a thermowell, I'd cut silicone sheet of some sort to keep them in place.
I use thermal compound to ensure good heat transfer to the walls. It makes a huge difference to the speed with my pt100s.

I second jbb's note on fake DS18B20s, the clones can be insanity inducingly flakey. Specially if they only start becoming unreliable when they start to heat up, as he says, when they are mounted in use and you can't debug them. I'd pay the digikey shipping for proper ones next time I use them.

[CatalinaWOW]

Food grade, you might have standards for this, for example if the thermowell breaks, leaks  etc. I was thinking of suggesting silicon oils but that may be toxic.
On the other hand I was reading recentlty that FDA in USA  does not actually prohibit machinery bits and pieces rats mice cockroaches etc.

They only prohibit that stuff from exceeding a certain ppm. !

Do you know of anyone who prevents such things?  How would you know if there were parts per million or parts per billion or even smaller quantities?   

Also think about the mechanics of this.  Lets say you examine a bin containing 1,000 liters of grain, and after carefully combing through 100 liters of grain you find an insect wing.  (Tedious job, but maybe you are volunteering).  Statistically you would assume that there were about 10 insect wings in the entire container.  You would have to do this to several 1,000 liter containers to have any kind of error bars on that wing count.  And make some assumptions that all of those containers were equally contaminated.  And there are other pitfalls.  Did you take your 100 liters off the top or find some way to randomly select through the volume?  Do you believe that the contamination is equally distributed through the container, or maybe concentrated on the top surface where insects and other critters have easier access.

Finally, other than ick factor, what is the harm from low levels of such contamination.

I think you will find that the US, and probably most other countries which have standards for these things have done a pretty good job of balancing real benefit and practical measurement.  Where there are differences I think you will find the justifications fuzzy and subjective.

[Mr.B]

...8mm ID (which I have) will just about take a TO-92, so it will take DB8120b 'one wire' senors, requiring only a pair of wires for the whole "bus" of sensors...

...EDIT:  I also need to make sure the epoxy (and everything) is okay at 105C...

If memory serves me, without googling the datasheet, the DS18B20 tops out at 85C.
(edit: my bad... Accuracy tops out at 85C, still reasonably good to 125C.)

I also second the statements above about knock off DS18B20s.
I only purchase mine from Mouser.
I have only had one bad one over the last 10 years and many, many hundreds purchased.
They are a very reliable, accurate, repeatable sensor.