Making A Thermocouple

A thermocouple is a device used to measure temperature by utilizing the Seebeck effect. This instructable will show a very simple method of making a thermocouple so you can precisely measure the internal temperature of your thanksgiving turkey.

I'm sorry that the one picture I have came out terrible. The picture with the flash off was even worse (low battery syndrome). I was debating weather or not to post - but considering I couldn't find instructions anywhere on the web on how to do this, I decided to go forth and conquer. At least the video in step 2 is somewhat amusing.

If I can - I'll build my own bridge and make my own meat thermometer probe. Why? you ask. Why not? What's cooler than measuring the temperature of your turkey with your trusty multimeter. And in my case, my trusty cheapo multimeter and calculator :P

Step 1 Materials

1. Thermocouple wire : we got ours from http://www.omega.com/prodinfo/ThermocoupleWire.html

You can probably do it with your own wires - but the must be dissimilar materials and your results may vary. If the materials are not suitable, you will know as soon as you try to calibrate.

2. A capacitor bank or other method to spot weld a tiny wire.

Comments

[Oracle1729]

You're putting 2 dis-similar metals into the meat. The meat will act as an electrolyte and you'll have a simple battery. The effect will be metal ions flowing from your thermocouple into the bird. Most metal ions are toxic except in tiny amounts.

[trebuchet03]

It can be put in an inert carrier without changing its ability to measure temp (although it may increase response time).

[Oracle1729]

Of course it can, and that's what a commercial unit does. However, this instructable shows using it directly without an inert carrier, so using it as shown here will posion the meat.

[trebuchet03]

Fair Enough :) I also found out today that the wire insulation is not food service safe. Not sure why though -- maybe lead or something? So, for said time/temp plot - I got a cheap meat probe ($15) and I'll keep record :) I'll even calibrate it :P

[Technophile]

The insulation might or might not be harmful. To rate it as food service safe the manufacturer would have to spend time and $ to design it to be food-safe, then spend more time and $ to perform testing, plus monitoring during production. Not something they'd want to do without a significant reason.

[android_18]

is it possible to construct a thermocouple which could produce 3 volts or higher?

[Technophile]

Not easily. You'd have to apply many junction pairs in series. Hot junction, cold junction and repeat.
From http://www.omega.com/temperature/z/pdf/z218-220.pdf, using Type K thermocouple wire and (pick a number) a 500F temperature difference you get about 10.6mV (0.0106V) per junction pair, so that's 300 junction pairs for 3.18V.

[dagenius]

thermocouples do not produce electricity... they change their resistance to electricity.

[trebuchet03]

No, the Seebeck effect (which is the principle in which a T-couple is driven) is fairly well understood. As defined:
The Seebeck effect is the conversion of temperature differences directly into electricity.

[dagenius]

*embarrased*

[trebuchet03]

No worries, it happens to the best of anyone ;)

[dagenius]

Oh, I think I am being mixed up with thermistors.

[dagenius]

Don't Peltier units operate on the Seebeck effect as well?

[kegancrispin]

Put a couple of twists on the wires, cut off all but the last twist, and use a oxy-acetylene torch to heat the twist until it fuses into a small bead. Don't solder 'cause you're just introducing more metals into the junction, and it's very unprofessional. We used to snicker at guys using twisted or soldered junctions (sorry godot01). You want only the two dissimilar metals ( for example, "J" thermocouples are Iron vs. a Copper-Nickel alloy called Constantan) and you want as small a junction volume as possible. If you want to measure the voltage directly, put one of the wires (stripped) in an ice water bath (for a reference junction zero point) before you plug into a voltmeter. Otherwise the measurement won't be correct or stable. That's basically what the circuitry in commercial thermometers (that use thermocouples) is doing--providing an artificial zero point and then converting that raw voltage to the corresponding temperature for that pair of metals. There are lots of handbooks, the mother of them all is N.I.S.T. Monograph 175 (I think the CRC Chemistry and Physics has them? or omega.com), with the raw voltages vs. temperature for all the common thermocouples. And yes, calibrate at a couple of knowns. Test labs use a very expensive zero point reference cell in their calibrations. For a turkey...they don't.

[dagenius]

So... how would you make the leads longer?

[royalestel]

Cool! Favorited!

[NachoMahma]

Great job! A few comments.
1) TCs are not linear, especially at the extremes of their ranges. That said, for most DIY purposes (0-100 degC) "common" (J/T/K) TCs are close enough.
2) I like to do a third calibration check at a temperature close to what I expect to measure (usually ~50% full-scale), but I'm a bit compulsive (13 yrs as an Elect/Instr Tech).
3) Make sure you use actual TC wire and not extension wire if you are concerned about absolute accuracy. Extension wire is OK if you're only worried about repeatibility.
4) I haven't looked at one in a while, but the Omega Product Handbook for Temperature (Omega) was my TC Bible back in the 80s-90s.

[godot01]

I guess it depends on the application but welding a thermocouple seems like overkill. We always silver solder them. We read them out thru a fluke t/c reader or a micromite. Just twisting T/C wire is generally good enough for many applications although not very durable. If this is not accurate enough for your application then a T/C is probably a bad choice anyway. We use RTDs in some applications.

[0.775volts]

OK, you did a good job making your thermocouple (I commented on making thermocouples in that instructable that was either about spot-welding, or playing with a big-ass capicitor bank, i can't remember which). ok, the best way to explain this is to highlight the process by which a friend and I have made these before. sorry if i'm a bit stylistic grammatically, i'm not on my first beer of the evening. ok, so take your length of thermocouple wire (I prefer type t) cut the jacket off the conductors, and strip both of them approx 1/2" on one end of the thermocouple wire, twist the conductors together. connect this end to the + side of your capacitor bank/welder/bigass current source on the other end of the thermocouplewire: take the conductors and cross them right where the insulation ends. twist them together as tightly as possible and cut them after one good, tight twist. then, take something conductive and flat and connect it with a big-ass wire to the - side of your current source. get a pair of needle nose pliers, clamp them on to the thermocouple wire about an inch away from the free end. wrap a rubber band/string/hairtie/whatever around the handles of the pliers so they stay clamped on the wire. put on welding guantlets and sunglasses and charge your current source once charged, pick up the needle nose pliers and get ready to spot weld the thermocouple to spot weld, just touch the thermocouple wires to the metal plate, creating short circuit and welding the metals together. pull it away after about 1/10 of a second, ie: get a good pop and move it away and you're done. if you stick it togeter for longer, you're likely to melt something and start a fire. we've done this many times, and it produces a good thermocouple. of course, you'll need to calibrate it when you use it, but you always did that with new thermocouples anyway, right?

[trebuchet03]

We're required to calibrate everything every time we use it ;) Even if Tcouple, transducer, scale etc. etc. came with a calibration chart - we are required to re-do it. These thermocouples were used for an exercise in calculating time constants and comparing to soldered connections. When a thermocouple has more than one contact point (such as wrapping onto itself) your measurement will change - instead of measuring at a point - you'll be measuring the average of the connection points. For the purpose of our lab, wrapping would give inaccurate results (higher time constants). Don't get me wrong, your method works too (just not so well for our application :P).

[0.775volts]

I'll have to send you pictures of our set up, we wrap and cut after the first wrap to have enough material to spot weld the wire, it makes a little blob that is solid, so there is only one (rather large) contact point. we had that issue with some of the first thermocouples we built, they were all over the place in terms of response. that's when we realized we needed to change our method (it took us 50' of type T wire to figure out where we were going wrong). when I'm back in my home town for the holidays, I'll duplicate the setup and take pictures so you can see what I'm talking about.

[ewilhelm]

You are now required to post a time/temperature plot of your turkey!

[trebuchet03]

"Just for fun" - I started writing an instructable titled: "A Statistical Analysis of Turkey Cooking." I've already come across some interesting little tidbits about turkey geometry and how they play a role in cooking time :P I think I've brought myself to a new dork low (or high?) :P

[vatosupreme]

Yeah, but we need the data in real time on instructables. So how about hooking it to a webserver :P

[zachninme]

That would be awesome! Stick in about 10 probes, and make a small AJAX site to display info, and allow us to yell at you when you should do something :-)