Noise in Analog Voltages from tandem DAQ Modules

[KyleL12]

I have a cDAQ-9178 chassis with NI-9205 for analog voltage and NI-9214 for thermocouples.

In our experiment, we have several pairs of spot welded thermocouples along a metal strip.  We measure the voltage drop between two adjacent thermoelements to determine the electrical power into the strip.  The TC extension wires are shielded twisted pairs.  The shields are connected to a terminal block which is then connected back to the common of the NI-9214.  The thermoelements are tapped at the NI-9214 to the NI-9205 analog card.  We measured the voltage drop between three different pairs, we also have a hall effect sensor and fluxgate magnetometer connected the NI-9205. 

I found out that the voltage signals get noisy when the hall effect sensor is disconnected, so it is being used to eliminate this noise.  It's low/ground is connected to the common of the NI-9205.  It acts like a pseudo-ground.

However, there is noise in the in the voltage signals now.  I have tried several different things.  I have found 1 microF capacitors eliminate the noise, but over time causes a sinusoidal fluctuation in the temperature measurements. 

Are they any suggestions to eliminate this noise without the capacitors, so that I do not have to try to eliminate this effect?  I am not sure if it is grounding issue or DAQ cross-talk.  I would expect minimal cross-talk between the NI-9214 and NI-9205 since they are part of the same DAQ chassis (I could be wrong) and I would thing the COMs would be shared between the modules due this chassis.  

Is there something else I am missing or has anyone come across a similar issue?

[MichaelDB]

The source of noise in an analog signal can be difficult to track down, unfortunately there's no silver bullet.  There are a couple things I can suggest. 

One is that you mentioned connecting your cable shielding to the common pin of a module.  Typically we suggest connecting cable shielding to the chassis ground, which may or may not be connected to a particular module common.  It depends on the module.  The module datasheet will contain a wiring diagram that shows the connections.  Usually we distinguish a common pin from a ground pin.  A common pin provides a common reference for measurement which may or may not be grounded, while ground is always connected to chassis ground. 

Regardless of the module, connecting shielding directly to the chassis ground is best practice.

Another suggestion is to make sure you haven't inadvertently introduced ground loops through cable shielding or other wiring.  A ground loop where both the sensor and DAQ module connect to ground through different paths can introduce noise similar to what you describe.

We publish a field wiring guide to address some common sources of noise in analog measurements.  I'd recommend giving it a once over, just to serve as a list of common issues to check.

https://www.ni.com/en/shop/data-acquisition/measurement-fundamentals/field-wiring-and-noise-consider...

If you're still having noise that's interfering with your measurements, we'd be happy to discuss it with you in more detail.  Please feel free to call or email us to open a service request.

[SteveD123]

Are the wires of each thermocouple welded separately to the metal strip? Ar you using resistance or induction heating, either of which will produce currents in the strip? If so, it could be potential difference between the thermocouple points caused by the voltage drop between them.

Are you sure that your measured noise is picked up in the wiring, and not simply a true representation of the instantaneous voltage at that point on the strip?

[KyleL12]

SteveD123,

The thermoelements are welded separately and there will be a potential difference across them, this will affect the measured temperature.  It is recognized in the testing standard we use.  This is not an issue for AC currents, but our measurement procedure can eliminate this bias when using DC currents.  The testing standard also requires the voltage drop to be measured across the strip and requires this voltage to be me measured between to thermocouples on the strip.  We choose to measure between the same thermoelement of the desired thermocouples and make the connections from the analog voltage DAQ to the desired thermoelements at the thermcouple DAQ. 

We have an older system where we use two different DAQs for measurements.  There is additional noise by piggy backing the voltage DAQ from the temperature DAQ.  I have verified this myself.  This type of cross-talk should be to a minimum for this DAQ Chassis and modules that we are using.  This noise i also occurring in from the voltage measured from a fluxgate magnetometer that we use for measuring the applied current to the strip.