Interference on T/C inputs to PLC

[LANETR]

I have interference on analog inputs to a PLC via a thermocouple input module. There are 8 inputs, we are using 4 on this particular module. They are foil shielded type K T/C's. When the furnace zone approaches the set point temperature (~900°C) ONLY the 1st channel input starts fluctuating in the PLC analog input registers but not when verifying millivolts with a multimeter. I have tried; swapping modules - the same problem occurs with a new module, ferrite cores placed around the T/C close to the input card, Checking that shielding is only connected at the PLC shield grounding terminal, moving the location of the thermocouples (they are in close proximity to glo-bars). This system has worked fine in the past but intermittently presents this problem which seems to go away on its own. I suspect interference of some type causing problems in the MUX internal to the analog input card but cannot measure it. I have considered running the shielded T/C's in a conduit but this should be unnecessary with heavy foil shielding. Does anybody have an alternate solution to minimize this noise?

 

[xnuke]

Just a couple of ideas:

GE Fanuc suggests grounding the shield of analog input cables in the field, not in the panel. It's in one of their PLC I/O manuals. I've never done it myself, but you might want to lift the ground in the panel and try grounding the device end. If you try it, let me know if it works or not.

Have you checked the integrity of the cable on that particular thermocouple? Sometimes an intermittent loose connection can cause a floating input, which can produce erroneous readings. When your furnace temperature gets hot, you may have a loose connection due to expansion of materials located nearby.

Are you getting noise brought in on a cold junction compensation line?

 

[LANETR]

Thank you for the reply on this. Horner Electric suggested grounding the shield at the PLC input module but neglected to mention grounding all unused inputs to ground which I haven't tried yet. Groundng in the field would be risky. The field location has high energy globars connected to SCR power controllers located in the proximity of the T/C's. I would imagine there is more noise there than at the PLC end (remote). Cable integrity is good. Type K T/C input millivolts check against a redundant temperature and a check of the physical connection was done. All connections were redone with new connectors. No cold junction comp. line used in this system. Ex USMC crypto-tech here ...

 

[newsarang]

Please check the filter frequency of the T/C module in its configuration.It usually have a setting from 10Hz to 100Hz and as you select the lower freq. the noise increases and vice versa.This feature is about AllenBradley module,Just check it with your PLC mfg.
Hope it helps you out.
bye

 

[melone]

You don't want to ground on both ends! Any potential difference in the "grounds" will cause current to flow. A change in current causes noise on your lines.

 

[xnuke]

Nobody said to ground on both ends!

 

[Tom001]

Have you got any inverter drives in close proximity to the PLC/thermocouples, i had a problem once where the interference could not be eliminated from the PLC/thermocouple, i fitted ferrite rings to the motor supply instead, this seemed to cure the problem

 

[cbarn24050]

Hi, make sure 1 of the T/C conections is grounded.

 

[dgauthier]

In the old days, we used to have to attach a very small value electrolytic capacitor across noisy TC inputs. This was a stop-gap measure until the cable or other faulty component was identified and replaced. You will have to experiment with various values to get the right filtering. We've also had cables ohm out fine all around, but still be the problem, so if all else fails, have a new cable installed. Also, is the TC grounded or ungrounded? Check to see if maybe the wrong type might have been installed by mistake.

Regards.

 

[chettyc]

This is a problem common with sampling caps in the A/D. When two successive channels are vaery dissimilar in the I/P, the above problem manefests itself. I would install a second TC on an adjacent channel and read only only the second channel, or set the interchannel sampling rate sliightly longer.
Goodluck