Accelerometer data with high skewness for USB cDAQ-9178 chassis with multiple NI-9234 modules

[EvenLund]

I appologize. Here is a subset from what I consider bad data. I have worse data than this, but the overall question is what can cause this skewness of the accelerometer data?

[Henrik_Volkers]

There is no info with that data..

SR?

scaling? Raw voltage?

[EvenLund]

SR=5688.89 samples/s

Units are m/s^2

I understand that devil is in the details, but in a general: what can cause spikes lasting for 1-2 samples?

[Henrik_Volkers]

3 harmonic of the line freq is a little to much off for my taste .... but maybe a slipping engine?

What's that 4.7Hz? Data is to short to do integration with a high pass filter lower than ~8 Hz ...

The 9234 will adapt the aliasing filter to the samplerate. So if you have short pulses (oh maybe a crack? or other short shocks, I don't know your setup, but there is some frequency content in the 1.5 to 2 kHz , some pressure spikes from the pump cause the spikes?) in the acceleration data, these will be low pass filtered (integrated) ... Get some data with max SR to validate..

Since there is always some offset error, integration should be done with a high pass filter matching or little lower than the HP of the signal chain (0.5 Hz here) so 10s of data would be fine 🙂

I used the sound and vibration vi's ...  so only a picture

[Henrik_Volkers]

and don't assume the 'movement' unrealistic... you don't have an inertial system, small changes in the angles together with gravity can cause such signals 😄

How long are the cables you use?  Since IEPE sensors running with only 2mA drive current have a slewrate limitation to positive signals, but can better follow the negative slope, this also migth be a cause of more negative spikes.

This is inherent to IEPE sensors. Can you measure or make an educated guess of the cable capacity?

See

https://www.ni.com/de-de/support/documentation/supplemental/08/using-long-cables-with-iepe-sensors.h...

[Henrik_Volkers]

If you tell us (or me)  more about what you want to analyse (and your setup) , we can give better hints.

If you don't want to make all stuff puplic, you can send me a private message.

[BradW]

You say you are applying the triax to a pipeline.  I'm assuming the signal is basically flow noise transmitted into the pipe. Could there be triboelectric charging of the pipe wall?  We've seen issues in the past w/ the 356A triax having some ground potential or noise issue when the structure wasn't well grounded.  If we are gluing them on, we've sometimes used a trick of applying a single layer of Kapton tape (very thin, stiff, high dielectric strength) to the surface and then gluing the accel to it, and presto, noise or offset went away.  Might not be your case, but just wanted to put it out there.

[EvenLund]

Very interesting that you mention this as this was actually our theory too. The stretch of pipe was suspended on rubber pads that were sliding on top of teflon discs. I think this combo is not good regarding the triboelectric effect. We found when using duct-tape between the accelerometer and the pipe the spikes were gone, but this signal occured when one of the connectors got in contact with ground. These are natural gas pipelines and there is an ongoing discussion whether the electric discharges can be dangerous.

[BradW]

Well, looks like you might be getting somewhere.  Applying a scope probe to the pipe wall might show voltage relative to ground.  And, it could have both a DC and an AC component.  If you don't have differential scope inputs, but have a battery powered scope, you can put one probe to the pipe and one to an earth ground and see if the traces are showing a deltaV.

 The duct tape will be much more compliant compared to kapton, due to the thickness and the rubbery nature.  It could be acting as a mechanical low pass filter and if you are looking for flow noise, it might lose some content.  The kapton reduces that effect, or at least pushes it to higher frequency. As for the connectors, if you see issues when the barrel of a connector touches the pipe, then yeah, you have ground potential.  If you can't ground it out, then wrap the connectors w/ electrical tape.  All AC coupled inputs that I've used have a max offset voltage they can tolerate before the signal starts moving.