Sudden spike in strain measurement

[johnsold]

Henrik,

Good points. We will just have to wait to see what Shweta can contribute.

Lynn

[Shweta_JP]

Hi guys!

Sorry for the delay .

The impact test procedure involves a manual drop of the load from a particular height. There are no solenoid switches involved. The user simply starts recording, then drops the load manually.

Could it be that some variation in the excitation voltage is causing it?

Regards

Shweta

[Henrik_Volkers]

Quick solution: SW median filter

But to find the cause a good starting point is correlation: Can you reproduce the spike without falling mass?

Another source of spikes is static discharge.

And yes, fast changes in the exitation can result in a spike in the signal.... (not in an ideal world of ratiometric measurements, but what is ideal?? )

[Shweta_JP]

Hey Henrik!

Using a filter is not an option as it will change the charatceristicts of the impact signal. But could you please elaborate more on the "static discharge" which could cause this?

I also tried recording the excitation voltage of the 4331 - it's almost constant at 2.5 V even when the spike is observed. Also, the spikes are observed even when no load is applied to the plate under test.

Regards

Shweta

[Henrik_Volkers]

http://en.wikipedia.org/wiki/Static_discharge

So keep on capturing data (try only one channel and the test panel in loop... )  so you can monitor when a spike occure and open your senses

is there a flicker in the room light ? Or other electric stuff .... turn stuff off and have a look

There are a lot of possible causes....

Since all channels are affected.. If you connect one DMS/sensor directly (well very short cable) to the Input. Is that channel spiked also?

Some more Very detailed information about your setup would be helpfull, since my cristall ball is in recalibration (no solenoid 😞  😄 )

About the filter: Take a pulse , apply the filter , calc the difference and have a look how much certain filter and samplerate settings will affect the pulse.

[johnsold]

Shweta,

As Henrik said, we need more information to really help you.  The best way is to identify the source of the spikes and prevent them from getting into your data.

Other options: Since you are concerned about the effects of a filter on your data, try a noise blanking approach which was developed for radio receivers decades ago.  You detect the noise by looking somewhere that has no signal.  When the noise is detected, then it is suppressed on the signal.  Use an empty channel to record the noise spike.  If you do not have an available channel, look for the spike on all channels simultaneously (maybe +/- 1 sample if the sampling is not completely synchronous). Since the spikes only last two samples, replace the data with the previous value or the mean of several previous values on each channel.  Replacing with zero or NaN may also be options, depending on the effects on your later processing.

Lynn

[Shweta_JP]

Hey Henrik & Lynn,

I'm sorry for not being clear earlier. The test involves dropping a weight on a steel plate. The steel plate has a fixed number of strain gauges bonded on to the plate at specific locations. Before the weight is dropped manually, they start recording the test. Once they drop the load, the recording is stopped. The strain gauge wires are connected to lugs from which it is wired to the respective 4331 TBs.

I guess the test scenario is now clear...

Like I said earlier, I would be more interested in eliminating the occurence of spike rather than filtering it out.

Regards

Shweta JP

[johnsold]

Shweta,

Thank you for the more detailed description of your setup.

Do you ever see spikes when monitoring the strain gauges but not dropping the weight?  Looking back , you said that you do see them in an earlier post. 

You said you are dropping the weight manually.  This seems to imply that there are people moving around the test rig. How about static electricity discharges? They can be very fast and may contain a lot of energy (lightning!). Does walking around and just touching the metal of the support frame trigger a spike?  Are all of the pieces of the apparatus electrically bonded together.  I have seen situations where minute static discharges across paint have caused spikes.

Lynn