Spikes in readings from cFP-2000

[AndrewFMills]

Hello,

I'm hoping some one will be able to help with a problem we are experiencing with our temperature and voltage measurements. We are noticing large spikes in temperature being logged that are not physically possible, and hence must be some form of interference effect. The spike seem to be entirely random, and I can't seem to be able to replicate the problem. They may happen once a week on average, and its not at the same time every week so its not as if another event is causing it.

Our setup is:

- Type-R thermocouples connected to TC-125 modules via compensating cable (some via long runs)

- 3 cFP-2000 units in an enclosure, with power supplied by individual Quint 24VDC power supplies. These are connected to the mains via a UPS.

- Also there are AI-110 modules connected to the same field point banks. These also seem to be affected too. These monitor -/+10V voltages from LVDT signal conditioning, and the cabling come in through the same bunch of cables as the thermocouple compensating cable.

- Each thermocouple is connected in parallel to a Eurotherm process controller (2416 and 3216 models) and the cFP.

- The LabView software is set to take a reading every 10 seconds, so I presume the spikes we are seeing are when we've been unlucky enough to catch when the interference is happening.

I'm hoping someone will be able to help! I haven't a clue where to start making changes to cure the problem, especially when I can't replicate the problem. Any suggestions would be approriate.

Andy.

[Andy.G]

Hey Andy,

How long are the thermocouple cables exactly? Also would it be possible to get a screen shot or attachment of the code you are using? I suspect interference in the wires, but this is just speculation.

[AndrewFMills]

Thanks for the reply Andrew,

The longest run of compensation cable is probably close to 100m, the shortest is a few metres. We are seeing just as much spiking in the ones with the longest runs as the shortest runs.

I don't have a copy of the code unfortunately. The program was written for us by a 3rd party. Instead I've attached an example of some of the data collected.

[Andy.G]

Hey Andrew,

Was the program working when you got it, has the spiking always been there? As if this is the case it could be down to the company who wrote the program initially. Since the issue isn't dependent on the length of the cables, the next step would be to check the break out boards. Is there any signal conditioning being used (as in cold junction compensation) also ensure these are boards correctly wired to the instrument. If this does not solve the issue then the issue maybe with the programming, in which case I would need a screen shot or sample of the code being used.

[AndrewFMills]

Thanks for your quick reply,

Yes the spiking has always been there. I hadn't even considered it could be a software problem, I didn't think this was very likely. all the program does is checks readings every 10 seconds - if the reading has changed from the previous value by an amount greater than a set deadband it logs a new point on the graph. Theres no control or processing element to it. My initial thought was it was EMI in the cabling.

Just for further information the cable lengths vary because the field points are centrally located in one room. The cabling runs to another 4 rooms with the test rigs in.

Signal conditioning is being used for the LVDT measurements to produce a -/+10V output. The TC-125 module's own cold junction compensation is being used for temperature measurements. I've checked the field points are correctly wired and everything appears to be correct.

As an experiment we've fitted out three of the test rigs with ferrite blocks on several of the cables in an attempt to suppress any EMI.

[Andy.G]

Hey Andrew,

I can't imagine it would be EMI, as this would cause a constant interference in the signal. What are the test rigs? If they are consuming large amounts of power a transformer could produce a magnetic field powerful enough to disturb the thermocouple significantly. Also what I could suggest is having a control, such as another thermocouple held at a constant temperature clear of any EMI source and see if that produces spikes. Also are you using the same Signal conditioning block for all the wires, could this be changed?

[AndrewFMills]

Hi Andrew,

My reason for suspecting it to be EMI is the test rigs contain a large transformer which is on constantly, and a motor which kicks in periodically. My suspicion was when the motor kicked in it was causing the EMI, but I've never been able to reproduce it.

The rigs are for testing creep in materials. This is one of the rigs we use http://www.denisonmayesgroup.com/Images/Tc50.

We are using the NI cFP-CB-3 connecting block for all the thermocouples, and the NI cFP-CB-1 connecting block for the LVDTs.

[Wayne.C]

I would focus on the motor because its operation is intermitent.  Its very possible that starting the motor causes a disturbance that is showing up on the power supply to the cFP. 

Is power to the motor being switched using a contactor or motor starter?   If its a contactor, make sure you have a snubber installed across the coil of the contactor.

[AndrewFMills]

The motor is activated by a mercury level tilt switch. I don't think there is anything else in the circuit but I may be wrong. I will follow it through.

[Wayne.C]

If a spark or arc occurs in the mercury switch as it makes or breaks contact, that will generate electrical noise as well.