Thermocouple Connections
Duane Degn
Posts: 10,588
I'm finally moving a oven controller project from a breadboard to an Propeller Protoboard with an enclosure.
I plan to use two Type-K thermocouples. I plan to use a thermocouple amplifier AD595 with each thermocouple. I'll use either a MCP3002 or MCP3208 to read the voltages from the amplifiers.
I'm connecting the AD595 the same as in the toaster oven controller schematic.
I will usually use the thermocouple reader in one location but I'd like the option of using it in a different location.
I'm wondering about the pros and cons of using some sort of connector on the thermocouple. One the breadboard I used screw terminals that were plugged into the breadboard. The thermocouple wires were connected to the screw terminals. I worried about the different metals interfering with the thermocouple readings but the readings seemed accurate.
Do any of you know of a reason not to solder the thermocouple wires to a connector such as a headphone jack?
Another concern I have is using a small TV screen as a display. The electronics of the TV is likely to be close to the thermocouple amplifiers. Any suggestions on how to minimize noise? I haven't tested the thermocouple reader using the TV display yet. I'll make sure the readings are still accurate before committing to the design.
I'd appreciate any insights, from those who have experience with thermocouples, may offer.
Duane
I plan to use two Type-K thermocouples. I plan to use a thermocouple amplifier AD595 with each thermocouple. I'll use either a MCP3002 or MCP3208 to read the voltages from the amplifiers.
I'm connecting the AD595 the same as in the toaster oven controller schematic.
I will usually use the thermocouple reader in one location but I'd like the option of using it in a different location.
I'm wondering about the pros and cons of using some sort of connector on the thermocouple. One the breadboard I used screw terminals that were plugged into the breadboard. The thermocouple wires were connected to the screw terminals. I worried about the different metals interfering with the thermocouple readings but the readings seemed accurate.
Do any of you know of a reason not to solder the thermocouple wires to a connector such as a headphone jack?
Another concern I have is using a small TV screen as a display. The electronics of the TV is likely to be close to the thermocouple amplifiers. Any suggestions on how to minimize noise? I haven't tested the thermocouple reader using the TV display yet. I'll make sure the readings are still accurate before committing to the design.
I'd appreciate any insights, from those who have experience with thermocouples, may offer.
Duane
Comments
The screw terminals will be accurate so long as both sides are at the same uniform temperature. Position the terminals as close as possible to the AD595, and keep it all at the same temperature, that is, not near some IC or resistor or lamp that leads to a gradient of temperature across the terminals or the AD595 itself. If everything is at a uniform temperature, the voltages of the additional metal junctions all cancel out.
The inputs to the AD595 are differential, and therefore can reject noise picked up from external sources, within limits. The TV screen should not be a problem. There are things you can do if noise turns out to be a problem, evidenced by fluctuations in the readings, for example, by rerouting, or by slipping grounded braided shielding over the wires, or by averaging.
Type K wires (chromel: chromium-nickel and alumel: aluminum-nickel) are very difficult to solder without aggressive fluxes. Screw terminals are a better bet, I think. The thermocouples measuring junctions themselves are better formed by welding.
Thank you for the reply.
After I posted my questions, I was explaining how thermocouples work (as best I could) to my wife. I was explaining that there was the hot end and the cold end (I think the proper terms are hot junction and cold junction). It then occurred to me that as long as both ends of any connectors were the same temperature, there shouldn't be a problem from the metal junctions.
I am glad to have you confirm my suspicions.
Thank you for the suggestions. I haven't tried it with the TV yet. It would be nice if I can use the thermocouples with the TV since I've already mounted the TV in an enclosure I plan to use.
Now you tell me. You're right about the wires being difficult to solder. I think I got them soldered on to an 1/8" jack. I haven't tested them yet. I might need to go back to the screw terminals.
Thank you very much for your assistance.
Duane
The best reason for not soldering the thermocouples to anything is that every junction between different metals creates a junction and can effect the accuracy of the "system". Remember a thermocouple is simply to dissimilar metals in contact with each other.
In theory you're right but in the real world it doesn't ever seem to work like that! I've run into problems when I attached alligator clips to thermocouple wire to ease/speed calibration of controllers and recorders. Soldering thermocouple wires is about impossible with anything but a torch. I've also found it best to use silver solder when soldering the stuff. The best way to connect thermocouples and thermocouple wire is with screw terminals made for connecting thermocouples.
Special screw terminals made of the same metals as the thermocouples ease the application, but usually end up transferring the problem a little further down the line or inside the instrument. It eventually comes down to a uniform temperature wherever the transition to dissimilar materials occurs. When using an AD595 or a high resolution ADC, special attention has to be given to keep the IC itself at a uniform temperature.
Type T couples (copper-constantan) are easy to solder with conventional solders and fluxes. So long as the entire junction subject to the same temperature, the presence of lead or tin or silver solder in the junction will not affect the reading. (Law of intermediate materials). A good reference is the guide to using thermocouples from Omega Engineering. Type T overall is the most accurate and easiest to work with at relatively low temperatures (up to 350 degrees Celsius).
The tiny change in temp caused by my fingers touching the clips took a surprisingly long to stabilize. It didn't take long for me to realize that using them was a mistake and I went back to using thermocouple wires of the type I was measuring! :-) I mostly worked with plastics extruders, injectors, etc and furnaces and my work was mostly limited to type J and type K thermocouples.
While I agree that every connection from the "hot" junction to the "cold" junction is bad the best I've found are soldered with compression fittings made of the same material as the thermocouple. Since the two wires making up the thermocouple and extension wire are different metals you have to be sure to wire therm properly! In the process of calibrating thermocouple based systems I've found that ALL junctions are a source of error and even slight corrosion can cause large errors. The fewer the better and something like a headphone jack is just asking for trouble and error. I know that the theory is that as long as all the junctions are at a uniform temp the errors offest and cancel. In the real world however I've found that the errors multiply. :-)
I've given up on the headphone jack idea. (Though it would have been cool to be able to plug the thermocouples in that way.)
I've been having trouble moving from a breadboard to a Propeller Protoboard.
I initially thought it might be because I switched my ACD from a MCP3002 to a MCP3208 but switching back to the MCP3002 didn't solve my problem.
The output from the ACDs stays at zero volts. There are still things I can check but I've decided to wait a couple days before continuing work on it, in an attempt to preserve some of my remaining sanity. The breadboad version is still working fine and has been for over a year.
I'll report back in a couple of days after I try working on this project again. I'll let you know if I'm still having trouble or if things are working correctly.
I really apreciate both of your suggestions. I was hoping someone here knew more about thermocouples than I did. Thank you.
Duane
The odds are that you're going to need to amplify the signal from the thermocouples with a high accuracy op amp with at least 1% resistors (.1% are better) to give you more voltage to work with when converting it to digital. The voltages you get from thermocouples are in the low mV range.
I thought that was what the AD595 was for?
Duane
Just checked the spec sheet and it looks like the output for the AD595 is 1V at 100 degree C. I could be wrong because I only took a quick look through the data sheet. can your adc deal with those voltages.
I'm drying soil/ore samples are part of a moisture test. I generally set the oven to 104 C. A couple of degrees high or low wont matter much.
As I mentioned above, I've been using an AD595 with a MCP3002 for over a year on a breadboard.
Duane
I actually have a MAX6675 chip.
The breadboarded oven controller was one of my first microcontroller projects. I had previously used an ADC so I felt comfortable with those and the SPI interface of the MAX6675 intimidated me. I'm presently comfortable using SPI so I might dig out the chip and give it a try.
I appreciate the other suggestions. I minored in physics so I've had some theory but most of my practical electronics knowledge comes from posts like yours on the forum.
Thank you again,
Duane
I'm attaching my own cold junction compensation/linearization object for type K thermocouples. I use that in conjunction with a high-resolution ADC that resolves to 7.8125 microvolts and also reads a reference temperature right at the point where the thermocouples attach to the ADC. There are two methods in the object. One takes as input the reference temperature in units of 0.1 °C, and returns the microvolts that a thermocouple would produce at that temperature when referenced to a second thermocouple in an ice bath. The second method takes that reference microvolts and also the differential microvolts produced by a measurement thermocouple, and returns the temperature of the measurement thermocouple in units of 0.1°C. The DAT table entries are in increments of 10 degrees Celsius, for interpolation.