Thermocouple/Ground Problems

Since the thermocouples are grounded through the machine I believe they are conducting voltages through the machine, not by induced noise which can be fixed with shielding. Both A and B are not shielded and both run in the same tray, running by the DC motor. Thus i dont think shielding is the main issue, I believe it is induced noise through the metal components of the extruder which is picked up through the grounded thermocouples.

Does this make sense? Im not an electrical engineer... but i am trying to learn.

The DAQ is not grounded through the power supply, I believe it is grounded through the computer via usb port. I dont believe there is a way to eliminate this ground? Is there any way to establish a common ground between the DAQ and extruder?

I also realize that ungrounded thermocouples should have been purchased, but I did not purchase them. Also it would be quite expensive to replace all 20 thermocouples. Im hoping to find a way to fix this... especially since the system worked before. All i have done is added a few more thermocouples and installed a new DAQ system (OMB-DAQ-2416 with an expansion modual OMB-AI-EXP32).

Thanks for the info... I'll keep researching

First off all you need to isolate the problem, sorry for the pun. But first short out all the inputs and run the system. They should all read ambient temperature. If any temperatures move by 5 degrees then the problem is with the data acquisition and not the Thermocouples.

If you still have the problem isolate the communication cables.

You can try to use a puck, thermocouple in and 4-20 out, http://www.endress.com/corporate#product/TMT182

Yes, I agree that it is probably not a shielding issue, given your additional information that both TC leads run together.

The first thing I'd try would be USB isolator, something like this:

"This isolation barrier eliminates ground loops which are disruptive in measurement applications."

On second thought, the DAQ isolation in to out is pretty good:
Input Isolation: 500 Vdc min between field wiring and USB interface

There's reasonable isolation:
Common Mode Rejection: Thermocouple mode, 110 dB;

But the expansion module is multiplexed:
"The OMB-AI-EXP32 is a multiplexer-based channel expansion module that adds an additional 16 differential/32 single ended channels." Solid state mux switches can share a common that will see common mode.

Have you tried disconnecting the expansion module to see if there's any difference?

A reliably functioning DAQ with low level signals requires a single ground, in the DAQ only. The floating twisted pair (shielded preferably, shield grounded at DAQ only) is connected to a differential receiver, with its excellent common mode suppression.

Introducing any other ground is asking for trouble. Especially a machine's power ground. Any change in that ground - while still well within safety margins - will mess with your signals. And you keep chasing your tail.

I suggest, you bite the bullet, and isolate them all. If the thermocouples are all connected via twisted pairs, all it takes a drop or two epoxy to do that. But first I would put an isolating tape between the sensor and the machine for a iury rigged trial run, before making it permanent.

Dear Mr,

we work with Thermocouple K and custom data acquisition module, normally the problem that you speak is that the groud wire is not correct sized for this application. To solve you can separete the power ground from the signal ground or change the size of ground.

You should have all your instrumentation and data aquisition going to a separate 'clean earth', not sharing a 'dirty earth' with plant equipment .

I believe this is the root cause of your problem, especially when the t/c's themselves are earthed (I have no clue why that is so).

I couldn't access websites that you referred to and see no sketches. Hence my brief earlier response.

You need to state:

• how exactly the t/c's are earthed
• are they located in thermo/wells, if so are they wetted? (I am not familiar with food extruders).
• Are the t/c's for machine protection, or to monitor processing temperatures ? What exactly are they protecting ? eg bearings)
• distance to the data aquisition system. Are you using compensating cable, end-of-line resistors etc..?
• 20 t/c's installed, but only one is acting up ? Are you sure that the machine has not got a hot spot there? That's what the t/c's are there for, after all.

Once the motor and power are turned off the thermocouple readings return to normal.

• Instantaneously ?
  

Hilton,

FYI

The Tc's are botled into the barrel of the extruder, they measure product processing temperatures, barrel temperatures, and cooling fluid temperatures.

When the TC's should have been reading room temp when the power was turned on they were instantaniously jumping up 30 to 50 degrees. then when the power was switched off they dropped down to the correct value.

I believe i have figured out a way to fix it.

Thanks

I agree with Iris, there is almost surely a feedback from the electrical ground system.

You could try to place an opto isolator between the machine and the ground system.

The problem mainly consisted of several ground loops and the fact that the thermocouples were grounded.

I found some old ungrounded thermocouples and ran new leads over to the DAQ, this worked!

I was able to find some of the grounding issues and fix them.

I went ahead and ordered new ungrounded TC's and ran new leads to the DAQ, also I plan on shielding the leads.

Thank you so much for the advice everyone!!!

Thanks for the update.

Iris, Hilton and I are singing from the same songbook. For starter, the thermocouples produce milliVoltDC signals in an environment, where all the machines are driven by hundreds of VoltsAC and tens of AmpsAC. With the slightest leakage in the machines (and all machines made by man do leak), you have a disturber, that readily overwhelms the TC signals.

Rule #1: isolate, and have only one CLEAN ground at DAQ. If you do not have it, make it.

Rule#2: all other grounds ARE dirty. Machines more, computer less, still all dirty.

Muxing milliVolt signals is a questionable endeavor at best, I do not trust it. In DAQ you ought to be able to connect 2 DAQ via 2 ports, in software.

best regards

It looks, that while I composed the note, the problem got solved along the line, we proposed.

Congratulations!

Thank you!