One thermocouple feeding two devices 2

[analogkid2digitalman]

I'll be doing additional research on my own, but thought I'd post here in case anyone has had similar needs/solutions.

We have an extruder that we'd like to monitor and log data on using data acquisition hardware. The majority of the signals are J and K type thermocouples that are fed into PID temperature controls. Unfortunately the PID controllers do not have retransmitter analog ouputs to scale the readings into V or mA. That would have been very convenient to interface to the appropriate DAQ hardware.

Thus, is it possible to connect a thermocouple input based DAQ to the same thermocouple inputs to on the PID controllers without affecting the signal? One thermocuple feeding two inputs- one to the PID, the other to the DAQ. I do not have the input specs yet for the PID controllers, presume the are >1Mohm. I would be using the appropriate thermocuple wire type as needed. Wiring runs would be short and DAQ hardware in same cabinet to minimize thermal gradients. Thermocouple inputs on the DAQ would also have CJC.

Thanks

-AK2DM


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[GTstartup]

Since the output of the T/C is mV, I believe the answer is yes. I haven't quite done what you have described exactly but I have read a T/C going to a control system input card with a recorder and a multimeter in parallel. I would think it would be quite like this

 

[JLSeagull]

Damian may need to weigh the need to control cost against the requirements for accuracy and reliability. The output of a thermocouple is a low-level milli-volt signal. Each junction of dissimilar metals such as the thermocouple extension wire forms another thermocouple. Low-level signals are subject to noise.

Consider duplex thermocouple elements among the potential solutions to the problems. One thermowell accommodates multiple thermocouple elements within the insulation inside a common stainless steel sheath.

Another solution is to use a transmitter. The 4-20 mA output is a high-level signal as compared to a thermocouple. Splitters and isolators are readily available in the 4-20 mA world.

 

[GTstartup]

Thermocouple extension wire is not dissimilar to the actual T/C. Also since the additional measuring device is in the same cabinet as the original, it's not an issue anyway (no Delta T).

 

[itsmoked]

This is a very poor idea and will likely NOT work. Do yourself/company a huge favor and just run a second T/C as this will also give you, an always useful, backup resource and comparison. T/Cs can drift badly as they are about to fail. Having a second one with which to note this is a very GOOD thing. You can often get dual T/Cs in one probe so there are no real hindrances to just swapping out the exiting one with a dual.

You are also not realizing that the PID controller will have a T/C break detect system that can totally screw up your 'two units on one T/C' scheme.

Public service announcements:
"Friends don't let friends hook T/Cs to more than one device."

"Just say no."

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[danw2]

The heat treat biz has paralleled thermocouples to two devices for longer than I've been alive. One connection to a controller, a parallel connection to a recorder.

There is always an issue of thermocouple break circuits introducing noise into the other device's circuit, but that has diminished considerably over the years. I don't know why.

There are a number of heat treaters in the area that interpret AMS2750 rev D, the pyrometric standard for the NADCAP aviation components producer crowd, to actually require paralleling T/C's, or using a digital protocol, but not analog retransmit for type A instruments.

My advice is to try it and see. If the you don't get spikes from the controller's T/C break circuit and each device floats enough to avoid a ground loop, go for it. The folks heat treating bolts, landing gears and who-knows-what-components for Boeing do.

Besides a transmitter option, dual element T/C's are quite inexpensive. Two elements in one sheath, close to each other.

Dan

 

[dcasto]

I'd put a TC to mA "puck" in. I'm not a EE, so forgive my lack of EE terminology. Then you can go through an optical isolator and split the 4 to 20 mA, one to a redcorder the other to the PID loop controller.

 

[JimCasey]

DanW2 sounds like he has done it and gotten away with it but I still think you'd have to be shrewd AND lucky. The T/C signal energy is so incredibly tiny. The devices you would have to use to read it have to have almost infinite impedance and if you're splitting the signal you just can't claim to have enough energy to go around. OR if there is any stray emf, your T/C signal could easily be dwarfed by the noise.

Suggestions of dual T/Cs in a single sheath (readily available) are prudent, or a single TC outputting to a 4-20 ma hockey puck is totally pragmatic. A 4-20 signal can usually drive through 600 ohms, so chances are you can drive a controller AND a recorder in series without resorting to an isolator/repeater.

And yes the connections of a t/c lead to each terminal of the hockey puck transmiter may also be dissimilar metallic junctions, but the guys who make hockeypucks understand this challenge and compensate for it. The compensation amy either be electronic, or the terminals may be the same alloy as the T/C wire.

I'm on Dcasto's and JLSeagull's team on this one.

 

[analogkid2digitalman]

Thank You all for your input and advise.

I may well stay clear of parallel inputs, every solution is going to incur some substantial $$$. DAQ hardware I am proposing is National Instruments compact FieldPoint to minimize induced noise and wiring runs. It is basically an ethernet controller that I can gang up T/C modules and voltage or current input modules. Both the T/C modules and the PID controls have open T/C detection, I can forsee some issues with that.

The approaches I will propose will be either install dual T/C's, which may be difficult mechanically due to current extruder head setup, or replace the PID controls with units that have retransmitters with voltage or current outputs that can provide signals to the DAQ.

I have also may propose T/C to 4-20mA converters, this would require reprogramming and rescaling of the PID's in use. I will have to assess the impact of that scheme.

Again, thanks to all.

AnalogKid2DigitalMan

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[danw2]

Any commercial temperature transmitter should be able to be scaled (whether in the field or at the factory) to mimic the T/C direct input range of your controller, so that no PID reprogramming or rescaling should be necessary.

Dan

 

[DaveScott]

AK2DM, you didn't mention what kind of accuracy you wanted in your measurement. Each device in your cabinet will produce a counter emf depending on its dissimilar metals and local temperature. I suppose that these will be averaged and each device will have more error (one will read higher than normal, and the other will read lower than normal).

 

[analogkid2digitalman]

Update:

Thanks to all of you for your experiences and suggestions. Based on them, and presenting the potential issues to coworkers, I chose the route of replacing all the PID temperature controllers with units that have 4-20mA retransmitter outputs. Controls and Compact Fieldpoint hardware is on order, looks like I'll have some new tows to play with!

Thanks Again

-AnalogKid2DigitalMan

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[roydm]

I can't believe what I just read. Thermocouples are not low power devices, they produce lots of current. A case in point. The pilot light in your gas appliance uses several in series (thermopile) to energise the solenoid. it's just that you loose it in the extension wire. It's quite acceptable to drive two or more high impedance devices. The only thing you have to watch is that the inputs are isolated from ground.

Roy

 

[itsmoked]

Yes but any current at all messes with the itty-bitty voltages they put out rendering them nearly useless as accurate temperature devices.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[analogkid2digitalman]

I recently purchased a datalogger to record process temperatures (not related to my original post). Facilities doubled up the thermocouple that went to the PID control and wired it to the datalogger also, even though I stated not to do it that way. Not good, we ruined a batch of product since the PV was skewed too low and the datalogger data did not match true temperature conditions.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

[itsmoked]

Ouch!

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[roydm]

Keith,
I don't agree, that the additional load messes with the "Itty bitty" signal, a typical instrument might have several MOhm input impedance. The thermocouple impedance (<1 Ohm) is very low in comparison. The leadwires however do have quite a lot of resistance.
For example if the T/C were putting out 10 mV and the leads were 100 Ohms, and the measuring device 1 MOhm the loss in signal at the terminals is only 0.01%. Besides the datalogger will only be sampling periodically. To double up on all the wiring is an un-necessary expense.
Don't take it from me however, try adding an equivalent resistor across an input, see what effect it has
Regards
Roy

 

[glowing]

Installing a E/I transducer "puck" is a not too bad of idea...or install a new TC.
While your extensions may be of equal material, they are unlikely to be the same length. this will only cause a small error. The bigger concern is the splice points. If you insist on doing this ensure the welds are atmosphere free welds (metal cup, mercury, oil and a welder does the trick).

 

[Standingback]

Dear All,

Just to add to Keith's view and perhaps to put some figures on "itty bitty", A "type K" T/C operating at 360 Deg F would produce 6.094 mV. A 4 degree change in temperature (just over 1% increase in temp) produces 6.183mV. So 0.089mV (1.46% change in mV).

So even if the imput impedance characteristics of the second device was close to infinity, the increased risk of inducing microvolt changes purely because there is another set of terminals to pick up background.

So it is a matter of risk, return and requirement. If you are happy that there could be a discrepency added by the second set of terminals, but were ok with the size of the possible errors, then why not. But is you are producing quality products of high value, why would you want to risk incorrect readings for the sake of replaceing the controller with one which had and auxillary output.

I believe AnalogKid2DigitalMan did the right thing. But thanks to all for an interesting thread.

I trust this helps,
Mlv

http://www.coulton.com

 

[analogkid2digitalman]

I strongly believe I did the right thing, thanks to many of you who provided me with input, personal experiences, and the potentential benefits and tradeoffs.

Eng-Tips is a valuable resource to have- the experiences and expertise of many at one's fingertips.

-AK2DM

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"It's the questions that drive us"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~