DPDT RELAY'S SWITCHING THERMOCOUPLES

[fifty60]

I have come across a job that has DPDT relays switching between two different thermocouples. 1 thermocouple is for a local reading and the other thermocouple is the remote reading. I checked the Data sheet and the minimum switching current for the relays is 1mA at 1VDC. I don't think there would be 1mA through the thermocouples, and surely we are only dealing with tens of millivolts. Any thoughts on using DPDT relays to switch between thermocouples? Should I be concerned about this minimum current voltage rating?

 

[kwired]

I have come across a job that has DPDT relays switching between two different thermocouples. 1 thermocouple is for a local reading and the other thermocouple is the remote reading. I checked the Data sheet and the minimum switching current for the relays is 1mA at 1VDC. I don't think there would be 1mA through the thermocouples, and surely we are only dealing with tens of millivolts. Any thoughts on using DPDT relays to switch between thermocouples? Should I be concerned about this minimum current voltage rating?

I would say yes, with such low current to start with even a very small contact resistance will impact the load. You need switching device designed for such low currents, and not general purpose switching device.

 

[fifty60]

I can find some gold plated bifurcated relays that have a minimum current rating of 100uA, but they still have a minimum voltage rating of 1VDC. I am not sure I will be able to find lower than that.

 

[Jraef]

I would say yes, with such low current to start with even a very small contact resistance will impact the load. You need switching device designed for such low currents, and not general purpose switching device.

I completely agree. I also don't know of a relay on the market that would NOT impact a TC circuit. Bad design.

I would use a transducer to convert the TC signal to something like 0-10VDC, one that provides offset and gain adjustment to compensate for resistance losses, then switch that signal.

Or better yet, a transducer that takes in two TC signals and gives one output that you can select via an input. I'm pretty sure Red Lion makes them, probably others too.

 

[Jraef]

Forget that last statement about a switchable input, I checked Red Lion and I was wrong. You could do it with a process controller, but it would be very expensive.

I did find however several dual channel transducers where you can take in two TCs and get two analog outputs, that would save money. Just nothing that allows switching inputs to one output. With that though you could take those two analog signals into a "smart relay" that can accept two analog inputs and in the relay, you can retransmit one analog output based on an input selection. All told it would likely cost around $350 though. Unfortunately I could not find a smart relay that has direct TC input modules available. You have to get into a micro PLC before you get that.

 

[kwired]

Forget that last statement about a switchable input, I checked Red Lion and I was wrong. You could do it with a process controller, but it would be very expensive.

I did find however several dual channel transducers where you can take in two TCs and get two analog outputs, that would save money. Just nothing that allows switching inputs to one output.

Would something using mercury switches work or is there enough resistance even in those to cause problems?

 

[gar]

131009-0919 EDT

You should be very much concerned. The issue is not that it is a DPDT relay, but the nature of the contacts. Also you need to be sure that there is not a shunt (parallel) path from the thermocouple probe back to the measuring circuit.

The contact material and any thermal gradients in the switching system are of importance.

Also do you really need a DPDT vs a SPDT? See the first reference.

Some solid-state switches might work. Gold contacts with sufficient thickness, palladium, mercury wetted contacts, and gold cross bar are possible materials. Sealed reed switch construction is an important method.

http://www.digikey.com/Web%20Export/Supplier%20Content/Meder_374/PDF/MEDER_Reed%20RelaysSwitchingLowMicrovoltsSignals.pdf?
redirected=1

http://www.leachintl2.com/english/english2/vol6/properties/00050.html

http://www.ni.com/white-paper/7200/en/

.

 

[gar]

131009-1007 EDT

What temperature range, accuracy, resolution, and stability are required for your application?

.

 

[Jraef]

Would something using mercury switches work or is there enough resistance even in those to cause problems?

Hmmm... Interesting thought, they were know to be low resistance, although mercury relays are pretty much taboo now. But low is still not zero, and ANY resistance in the circuit that is not able to be compensated for is going to negatively affect a TC circuit. That's why TCs come with a fixed lead length and you must not cut it, the wire resistance itself is already factored into the overall calibration. That's what makes RTDs a better choice for temperature monitoring now, the signals can be calibrated in the field.

 

[petersonra]

there is also the problem with T/C of having additional junctions inline that can affect the reading.

 

[GoldDigger]

there is also the problem with T/C of having additional junctions inline that can affect the reading.

You would have to be careful that all of the wires coming into and out of the relays as well as the relay contacts themselves were at the a uniform temperature unless you have all of the relay internal wiring and contact surfaces made of the same material as the corresponding wire in that leg of the thermocouple.

BTW, the resistance dependence of some thermocouple circuits stems from the fact that the sensor, for economy and simplicity, is really a current sensor, while the TC junction is actually a voltage source. If you have a sensor which is truly a voltage sensor, the loop resistance will not be important.

 

[fifty60]

Thanks for all of the responses. Bringing the T/C's into an input module and then working with seperate 0-10V outputs would definitely be the best option, however the pricetag on that puts it out of range. I have seen some controllers that will select the sensor input based upon the status of a digital input, but unfortunately that is not a capability of this controller.

I believe my best option would be to switch sensors to 100 Ohm Platinum RTD's. I have 2 seperate wire wound, 3 wire, RTD's. I am trying to understand why the RTD's would be a better option. The resistance of the sensor plus 2 of the leads is subtracted from the resistance of 2 leads. I do not understand how the circuit accomplishes this. These calculations and adjustments must be done by the controller. If anyone can offer some insight into this it would be appreciated.

I believe the current involved with RTD would be much more than an T/C but it is probably still less than a MA with voltages less than 1V. This is still well below the ratings of most gold plated contacts. How would I find out the exact voltages across the RTD and mA's through the RTD? Would this be a characteristic of the controller or is it a characteristic of the RTD?

 

[Jraef]

With a TC, the sensor itself is generating a mV signal because of heating the dissimilar metals. The device at the other end looks at that mV signal to determine the temperature.

With an RTD, the field device is just the sensor, an element with a fixed specific resistance. The excitation current sent to it comes from the transducer or input module it connects to and you calibrate it at that end to match the conditions after they are installed. So without knowing what you are going to connect it to, you won't know the voltage they use. But it's always low to avoid heating up the RTD sensor, usually between 5 and 8VDC.

 

[fifty60]

The resistance added by the contacts should be accounted for by calibration using RTD's?

 

[fifty60]

Thanks for the links, they are very helpful. We use gold plated contacts that are rated at 10A. It seems like the optimal gold plated contacts would be rated for less than 3 Amps. When I add the contact resistance to one leg of the RTD, the entire resistance goes up by the value of the contact resistance. Is this resistance change eliminated by the 3 wire configuration or does the added resistance have to be included on 2 of the 3 leads to be ignored in my measurements?

131009-0919 EDT

You should be very much concerned. The issue is not that it is a DPDT relay, but the nature of the contacts. Also you need to be sure that there is not a shunt (parallel) path from the thermocouple probe back to the measuring circuit.

The contact material and any thermal gradients in the switching system are of importance.

Also do you really need a DPDT vs a SPDT? See the first reference.

Some solid-state switches might work. Gold contacts with sufficient thickness, palladium, mercury wetted contacts, and gold cross bar are possible materials. Sealed reed switch construction is an important method.

http://www.digikey.com/Web%20Export/Supplier%20Content/Meder_374/PDF/MEDER_Reed%20RelaysSwitchingLowMicrovoltsSignals.pdf?
redirected=1

http://www.leachintl2.com/english/english2/vol6/properties/00050.html

http://www.ni.com/white-paper/7200/en/

.

 

[GoldDigger]

The resistance of the sensor plus 2 of the leads is subtracted from the resistance of 2 leads. I do not understand how the circuit accomplishes this. These calculations and adjustments must be done by the controller. If anyone can offer some insight into this it would be appreciated.
View attachment 9198

You have to look at the fact that the controller cannot sense the resistance directly to understand how it does the subtraction. (A digital controller could actually calculate both resistances and then subtract, but an analog circuit can also get the same result.)

The following may not be the way a particular controller does it, but it is one way it could be done:

1. Start by driving a known current, I, in the same direction through both R2 and R3, with the sum of the two currents coming over the return path via R1.
Now measure the voltages at terminals 1, 2, and 3.
2. Now note that V₁₂ is equal to 2I(R1 + Rb) + IR2 and V₂₃ is equal to I(R2 + R3).
For this three terminal network to deliver the result stated, you have to be able to assume that R1 is equal (or very close to) R2 and R3, call it R. This will be the case if they are the same diameter conductor run in a single 3 wire cable with any intervening connectors also having matched resistances on all contacts.
3. The equations in 2 then become V₁₂ = 3IR + 2IRb and V₂₃ = 2IR.
If we run these two voltages into an op amp with the input and feedback resistors set up so that the output voltage is (1/2)V₁₂ - (3/4)V₂₃/I, then the output voltage will be directly proportional to IRb. At this point either take advantage of the fact that I is known and regulated or just use another op amp-based divider circuit to get Rb. From there, do the calculation (or analog manipulation) to find the corresponding value of T.

To use one controller with two RTD sensors that have different length leads, you will want to use a three pole relay unless you are willing to parallel the connection of one of the leads.

 

[ptonsparky]

Interesting. We are rebuilding a Flaker for a feedlot. They use a series of three switches to feed three thermocouples to one readout that has a single input. Switch one is lower sensor, two is middle and three is top. What would be the difference of using a switch vs a relay?

Granted we are looking at degrees +- a few, and not .01.

 

[GoldDigger]

Interesting. We are rebuilding a Flaker for a feedlot. They use a series of three switches to feed three thermocouples to one readout that has a single input. Switch one is lower sensor, two is middle and three is top. What would be the difference of using a switch vs a relay?

Granted we are looking at degrees +- a few, and not .01.

The same contact resistance and temperature differential considerations would apply equally to both switches and relays. The switches just make it a manual operation.
Now there are available switches designed specifically for use with thermocouples whose terminals and internal leads are made of the same material as the corresponding thermocouple wire(s). It is less likely that you would find a relay built the same way.

 

[gar]

131009-2107 EDT

fifty60:

First, you need to define what thermocouple type you are using, what its output voltage is at the minimum and maximum temperatures of operation, what is the required accuracy of temperature measurement, and what stability of measurement is required.

Second, what is the environment that the relay contacts will be subjected to.

Third, what is the input impedance of the measuring instrument to which the thermocouple is connected.

Provide this information.

A 50 year old mercury wetted contact relay, P&B JM1 112 11, has less than 10 microvolts drop anywhere from 0 to 30 microamps. This relay has not been used in 50 years. We should assume that your measurement load is a high impedance. Thus, it is important how the contacts function at low current levels. This relay plugged into an ordinary octal socket, and then some wires to binding posts. Measurements across binding posts.

A more recent small reed switch relay same results.

The 10 microvolts is the smallest increment on the meter.

This does not address thermal effects, but you have to provide more information.

.

 

[fifty60]

"To use one controller with two RTD sensors that have different length leads, you will want to use a three pole relay unless you are willing to parallel the connection of one of the leads."

I have decided to use a 100ohm platinum RTD. I will most likely have to revisit the T/C switching at a later date, but from the information gathered it seems like switching T/C's with a relay is far from the ideal situation.

By paralleling one of the connections do you mean taking the third leg from both RTD's to the same terminal, and never switching them, while having the DPDT relay switching between the 1st legs and 2nd legs of the 2 RTD's? So really the third legs of both RTD's will enter the same terminal of the controller and never be switched.

If this is correct, would I then have more flexibility with my relay selection? I would not have to use bifurcated relays, but could use standard gold plated relays?