NerdKits - electronics education for a digital generation

You are not logged in. [log in]

NEW: Learning electronics? Ask your questions on the new Electronics Questions & Answers site hosted by CircuitLab.

Basic Electronics » Thermocouple wiring??

September 04, 2013
by Ralphxyz
Ralphxyz's Avatar

My project has a thermocouple with a standard two prong male end.

I have a wired female connector.

Now what type of wire would I use from the female connector to the pcb?

Does it have to be thermocouple wire?

If I had a pcb female module the pcb traces would not be thermocouple traces!

Thanks,

Ralph

September 04, 2013
by sask55
sask55's Avatar

Ralph

You should just use regular hook up wire from the connector to the pcb. Technically it is necessary that both of the thermocouple wires connect to the hook-up wires at the same temperature. In your case that is most likely going to be the case since the connections are both in the same connector body.

September 04, 2013
by Ralphxyz
Ralphxyz's Avatar

Thanks Darryl that is what I thought, the package mentions using thermocouple wire in it's installation instructions so that confused me. I couldn't find female PCB connectors that would have been preferable.

Ralph

September 11, 2013
by mongo
mongo's Avatar

I use thermocouples in my field. When running a connection from a sensor on a machine, I use the wire made for the type of thermocouple I am using. Usually type J or type K but I have used others as well. The biggest reason is that whenever there is a junction or connection between dissimilar metals, such as thermocouple wire to standard wire, it too becomes a thermocouple. Using the type J wire on a type J thermocouple eliminates these errors.

Most of the instruments also use connectors that are typed for the same. So the pins and receptacles of the plugs and connectors are also the same.

However, for hobbyist purposes, since there is really no need for the kind of accuracy I generally seek, regular wire is OK. Temperature readings will be a little bit off and non-linear but not by enough to make much of a difference.

Also, if you use good thermocouple wire from the start, you can make thermocouples by simply stripping a little bit on the end and twisting the wires together tightly. It works in a pinch when you can't find a manufactured unit.

September 11, 2013
by scootergarrett
scootergarrett's Avatar

While you are discussing Thermocouples, I have one from this DMM which comes with a K type thermocouple. Sometimes in the morning when I’m board I like to monitor the temperature of my warm beverage. Is putting a thermocouple in a liquid affect the measurement (Other than the liquids temperature obviously) due to current flow through the fluid? from experience it doesn’t seem to mess it up but I have no data to back that up. Also is it bad to drink?

September 12, 2013
by sask55
sask55's Avatar

Eventually there will always have to be two connections between the two dissimilar types of metal in the thermocouple the connection metal within the reading interment. Since these two connections are in opposite directions the thermocouple effect will cancel each other out as long as they are at the same temperature. I have used thermal blocks and have seen examples of ice water baths to insure that the two connections are maintained at the same temperature. My point is there is always going to be at least three thermocouple connections involved. To be very accurate it is important to cancel out the connections that are not of interest by keeping them at the same temperature, how far from the reading insterment this occurs is irrelevant, they are always there somewhere.

September 14, 2013
by mongo
mongo's Avatar

Even in the instrumentation I deal with, there are dissimilar metals at some connection point. The trick used with these are that the metals used are not as reactive and some counter each other to nullify the effect. Gold contacts are one way to do it. Some are gold alloys.

The type K thermocouples are generally used in the food industry. The metals used are basically stainless. (Nickel, chromium, etc.) Wire colors are red and yellow. Type J are Iron and copper wires but there are new alloys that do not corrode so easily. Mainly replacing iron with constantan. Wire colors are red and white.

So, no, the type K t/c won't affect your beverage.

I have several temperature measuring meters, which use a variety of probes. The little plugs and connectors all use metals of similar properties to the wires to minimize the effects.

September 15, 2013
by Ralphxyz
Ralphxyz's Avatar

But what happens when one goes to a PCB? All of the traces are copper, so what the difference if I use copper wire to connect to a breadboard or pcb?

Ralph

September 15, 2013
by mongo
mongo's Avatar

The interface and built-in error correction deal with that. The non-reactive metals also reduce the error. From there, it is more or less just a signal level.

For these reasons, I have pretty much gotten away from T/C junctions, except for high temperature applications. I much prefer the platinum RTD elements. Three wire sensors that are much more stable. The third wire ensures that all you get is the signal over long distances. RTD sensors do cost more but they are much easier to use in the circuitry too. T/C junctions need to have some sort of "cold junction" compensation to be reasonably accurate, and that is a tricky thing to design in a device.

There are a lot of instruments that use a variety of T/C sensors and RTD sensors. They are programmable and all have the built-in error correction. All you do is select the type of sensor you are using and the internal software and algorithms do the rest.

Now, if you connect a T/C directly to the PC board, the copper/copper junction won't make any difference. But the constantan lead and copper trace just might. The metal of the solder can also affect it. That's why they generally use some sort of connector or plug, made for the specific type of junction in use. Even the fingers (edge contacts) of a commercial PC board are coated differently than the rest in the guts of many instruments, just to further reduce the error. I'll get some pictures with examples tomorrow when I go back to the shop. I might have some things in my garage but it's still raining here in Denver. I don't want to be tracking water all over the place.

September 15, 2013
by Ralphxyz
Ralphxyz's Avatar

I have a PID and I have some ThermoCouple female connectors (wired type not PCB) my thermocouple has a two pin male end.

Some people are just bending the male pins to fit the thermocouple terminals on the PID.

Since I have the female thermocouple connectors I thought that would be better and just asked how to wire the female to the PID terminals.

If I had some K wire I could use that but since I do not have any K wire it seems like copper will have to work.

Well actually I might have some K wire I'll have to check all of the thermocouples I have, I am sure some of them are not terminated and I could snip a couple of inches off the end. But then I do not know what the thermocouple wire might be.

Is it worth the effort?

Ralph

September 19, 2013
by mongo
mongo's Avatar

If you have a K sensor, you can generally use copper leads. This is because the junctions where the copper wire and sensor are connected kind of cancel each other out. The two metals in a K probe are so similar that the copper junctions see both side pretty much the same. So you have a positive junction at one side and a negative junction at the other, effectively nullifying each other.

If you had a J probe, the copper to copper junction would be null. But the copper to constantan junction becomes another J thermocouple externally. Varying temperatures at that point will affect the accuracy of the active junction. In that case, it would be best to run the entire length in J wire.

The facility I am at uses Athena. Because of the temperatures we run, they are J probes. The terminals and edge connectors use non-reactive alloys for the connections but I do run J wire.

When I was at the R&D pilot plant, everything was in K type. Even there, we used K wire but that was largely due to the distances that were in use. Some sensors were up to 2000 feet away.

September 20, 2013
by Ralphxyz
Ralphxyz's Avatar

Thanks Mongo, here are the OMEGA Thermocouple Connectors I will be using.

I was asking about wiring the female. There would not be any direct connection between the thermocouple and my connection wires I would connect to the terminals inside the female connector. I am sure copper will be fine the leads will only be three or for inches long connecting to the PID my PID is the smallest one.

Thanks everyone for the help.

It sure will be sad if this forum disappears along with the store, to bad.

Ralph

September 20, 2013
by sask55
sask55's Avatar

Ralph

If you are interested you could take a look at this thermocouplers it explains in some detail the theory of thermocouples, including regarding the reference connections temperatures.

Mongo has much more practical experience using a thermocouples then I do. My experience is limited to replacing existing thermocouples as well as a few physics lectures and labs studying in some considerable detail the Seebeck effect. Mind you that was 35 years ago but the principals remain.

Thermocouple temperature readings are always the result of the difference in the temperatures between the dissimilar metal junctions you cannot avoid that fact. By moving the junctions that are not of interest back to your interment it will by possible to maintain a more consistent temperature and if a great deal of accuracy is very important you could measure that reference temperature using some other means. Then your instrument can compensate for any possible fluctuations in the referance temperature connections. I believe that this is the technique used in most thermocouple reading instrument to insure accuracy, and explains why it would be important to move your dissimilar metal junctions back to the instrument.

I also have concerns about the longevity of this forum I certainly hope this forum continues. I learn so much just following the discussions. The situation with the library and the store cannot be a encouraging sign.

Darryl

September 20, 2013
by mongo
mongo's Avatar

Ralph,

Those are the same type I use with my K's. They come in different colors for the different types. I think the T type is blue, J type is black... (I could be off a little) K is yellow as you already see.

They use metals for the contact surfaces that match the type to avoid the errors. Most controllers have built-in cold junction compensation and it's actually programmed algorithms instead of the old fashioned physical junction.

September 21, 2013
by sask55
sask55's Avatar

Ralph

It has occurred to me how difficult it may be to accually get a temperature reading using your thermocouple if you intend to use the Atmega 168 ADC to read the signal. First you will definitely need to know what type of thermocouple you are reading. As Mongo has said since the temperature /voltage relationship is not linear you would require a algorithm or a very extensive table to make the convertion. Most importantly you would require a very good analog amplification system to bring the very small voltage changes into the realm where you could possibly detect reasonable temperature changes using the ADC.

Take a look at this table K Type for a type K thermocouple. A temperature change from freezing point to boiling point of ether the reference connection or the point of interest would result in a change of 4.5 mA to the ADC. That is less then you would be able to detect (5V/1024steps =4.8mA). therefore your temperature resolution is in the order of 100 degrees C or 212 F per step of the ADC.

I think you will want to put some kind of very good op-amp or transistor circuit on the input to the ADC. There is no point in concerning yourself with details like how close to the board the reference temperature point is if you cannot resolve a change of 200 degrees F at the referance point with the ADC.

Darryl

September 21, 2013
by Ralphxyz
Ralphxyz's Avatar

Hi Darryl, for this project I am using a PID (see reference above) not going to a MCU.

If I was going to use a mcu I have a IC that would go in between it is made to read thermocouples and output a appropriate ADC signal.

Ralph

September 21, 2013
by mongo
mongo's Avatar

An alternative is the 2 wire t/a (Temperature to current) transducer. Analog Devices AD590 is a very accurate temp sensor and takes very little to interface it.

It is a current source, so wire distance really isn't a problem. One side to 5V+ and the other side to a 250 Ohm resistor and then ground. The voltage read across the resistor is 1mV/°C (I think, it's been a while since I have done much with them). The reading is absolute, so at 0°C, you would be reading 273mV. 100°C would then be 373mV. The rest is in the instrumentation amp you choose to use to scale to degrees F, zero and span.

September 21, 2013
by mongo
mongo's Avatar

Here is the data sheet: AD590

September 22, 2013
by sask55
sask55's Avatar

I somehow missed the PID reference you made. Did the IC you are referring to read thermocouple output to an ADC level come with the thermocouple? What IC is it? The thermocouple temperature range is very wide. most other systems have a much lower max temp level.

I am still intending to investigate the one wire system used on the bin temp cables I have. There is a lot of very critical, timing related, coding to that system, lots of information available on line, it is not simple to read. The power and data from up to 100 sensors all on one wire and the ground. I looked at the voltage level with my scope, there is a lot going on. It will require some considerable code to sort out what signal is coming from what sensor and read the outputs.

Post a Reply

Please log in to post a reply.

Did you know that negative numbers are represented in two's complement notation in binary? Learn more...