Making a thermistor food probe

[znorris]

I need a thermistor food probe for my first arduino project. I haven't been able to find anything pre-built that is within my price range and still meet the specs for the project. All the pre-built temp sensors for food are cheap but they do not include a resistance to temperature chart! I'm really not trying to do that all myself... I wish I could find a empty metal food probe and I could insert my own thermistor.

It needs a high operating temp (300F) as it will be in a grill
It needs to be in a pointy stainless steel probe to prevent corrosion, facilitate cleaning and not leave a gaping wound in the chicken
Negative temperature coefficient preferred

If anyone could help me it would be much appreciated.

[easternstargeek]

I have a cheapo wireless barbecue thermometer that I bought at one of the Big-Box home center stores. I don't remember what I paid for it, but it wasn't much- probably around US$30.00 or so. It has a detachable probe of exactly the kind you are looking to make, and the cable is heat-resistant and covered with a nice stainless steel braid. The other end of the cable has a standard sub-miniature (2.5mm) phone plug.

One day, I decided to see what was in the probe, and since I suspected it was a thermistor, I hooked up my ohmeter. Sure enough, it was an NTC thermistor, with quite a large coefficient of temperature. Thermistors are cheap and effective, but highly non-linear, so if you are gonna roll your own Arduino project to read it, you will have to craft some kind of linearization algorithm to correlate temperature with resistance. The easiest way to do that would be to use a lookup table to define points on a piece-wise linearization curve. The downside to that, unfortunately, is that you'll have to make a calibration table for your thermistor, which requires an accurate reference thermometer, and a lot of patience.

You can use an ice-water bath for zero deg. C, and boiling water for the 100 deg C point (unless you live in Denver). The in-betweens are going to be more difficult. Years ago, I worked in a calibration lab, and we used hot oil baths for temperatures up to about 450 deg Fahrenheit or so. We used ordinary vegetable oil of the type you cook with. The hardest part of calibrating the probe is keeping the bath temperature steady during the "soaking" time required by both your probe and the reference thermometer at each new temperature point. It is likely that the response times of the two devices will not be the same.

Achieving any kind of high accuracy will be difficult, but it will really give you an appreciation for what it takes to design electronic instrumentation. The learning experience will be well worth the effort.

PS: If you get a thermometer like mine, you can get a 2.5mm plug and jack, plus a switch that will allow you to rapidly switch the stock probe between the stock transmitter and your ohmeter. That will allow you to log resistance versus temperature readings "on-the-fly," using the stock receiver-indicator as your reference.

[sensor geek]

Perhaps you mean thermocouple and not thermistor. A thermistor probably wouldn't survive grilling temps.

[easternstargeek]

Sensor geek: The unit I have is rated to 450 degrees, fahrenheit, and it is definitely a thermistor and not a thermocouple.
If you are trying to measure the temperature of a piece of meat, you'll never need more than 200 degrees, anyway, since the probe will be in something moist. From a hardware standpoint, a thermistor is the cheapest and easiest sensor to read with an Arduino, but you do have some software work to do with the linearization.

@znorris:
By the way, you can get a replacement probe assembly here for less than $US10.00. I can't guarantee that it is a thermistor, but it is exactly identical in appearance to mine, and since it is a cheap piece of consumer gear, it is probably the same one I have.
http://www.smarthome.com/3218CR/Maveric ... ter/p.aspx

[sensor geek]

If you are trying to measure the temperature of a piece of meat, you'll never need more than 200 degrees, anyway, since the probe will be in something moist.

Very true. Seems obvious when you think about it, which apparently I didn't. 450F on a thermistor? Things have improved - I don't recall ever seeing one good for more than 250F.

[znorris]

@sensor geek You're right most thermistors I've seen peak out around 250F but they do have some higher temp ones which are still cheap compared to other methods. I may end up needing a thermocouple if we want to take the temp of the firebox itself.

@EasternStarGeek Those are the exact probes I am currently looking to buy. If I fail at creating a good resistance to temperature chart for calculations I will most likely strip the thermistor out of the probe and insert my own.

Thank you both

[meseta]

Try PT100 probes from eBay: http://www.ebay.com/sch/i.html?_from=R4 ... Categories
excellent temperature range (up to 850F), excellent stability and accuracy, and housed in a nice stainless steel probe.

[easternstargeek]

A Pt100 probe, otherwise known as a Platinum Resistance Thermometer, is a temperature-dependant resistor, standardized to have a resistance of 100 ohms at 0 deg. C. At 350 deg F, the element will have a resistance of 166.98 ohms. Over that range, your linearity should still be better than +/- 2 degrees F- making them much more linear than a thermistor. Now for the bad news:

The tricky thing about using Pt100 probes with an Arduino (or anything else for that matter) is that you have to be very careful how you excite the Pt100 element. This is important for two reasons- most inexpensive PT100 sensors are made using a thin-film process, and cannot pass much current before they burn out. Also, any current will self-heat the element, throwing off your reading. Most applications I have seen use an excitation current of around 100uA (0.1 mA). The bottom line is that you don't have much signal to work with, and so an instrument amplifier and precision current reference is necessary, ahead of the Arduino analog input. In addition, improper amplifier design and even wiring techniques can cause very large measurement errors. To read more about this, follow this link:
http://www.picotech.com/applications/pt100.html

These sensors have been around forever, and much has been written about their use. I found an interesting thread over at AVR freaks, not all of it accurate, but an interesting discussion, nonetheless:
http://www.avrfreaks.net/index.php?name ... 31&start=0

Personally, for what you're trying to do, I'd stick with the thermistor probe, and if you need more accuracy, I think the best solution would be Ladyada's Thermocouple and interface set:
http://www.adafruit.com/products/269
http://www.adafruit.com/products/270
It's quite a bit more money, but it will work right out of the chute. I have a set myself that I use for keeping tabs on my wood-burning stove, and I have been very satified so far. If you choose to use the interface with someone else's T/C probe, be sure you get a Type "K" probe. That is very important! (Type K defines the specific metals that are used for the T/C junction, and different types have very different output voltage responses)

Good luck, and please keep us posted on your progress!

PS: For anyone who is interested in learning how to measure almost anything, I recommend a thorough perusal of the Omega Engineering site:
http://omega.com/
This is the best-known industrial measurement, process control, and heating technology site there is. Most of it is geared towards high-reliability black-box solutions, so you won't learn too much about how to design circuits down to a component level, but as far as excellent white papers and tutorials about sensor technology, it can't be beat.