Sudden spike in strain measurement

[Shweta_JP]

Hi!

I have a PXI 1062Q Chassis, 8108 controller (Windows OS), one PXI 4472 module(Slot 2) & two PXI 4331 modules(Slot 3 & Slot 5). The test scenario involves measurement from strain gauges mounted on a plate which is subject to some impact. But when I try to acquire strain data in the quarter bridge configuration from all the 16 channels at 100 K sampling rate, I find that there is a sudden spike in the data apart from the one observed due to the impact (Please refer attached pics). This spike is seen at exactly the same instant across all the channels. The following are the settings used for strain channel acquisition.

Acquisition Mode :Continuous Samples
Sampling Rate :100K
Samples to Read :10 K
Maximum Limit :1.5m
Minimum Limit : -1.5m
Strain Configuration :Quarter Bridge I
Voltage Excitation Source :Internal
Excitation Voltage :2.5V
Gage Resistance :120
Gage Factor :2

Note: Offset nulling & Shunt calibration are enabled (100K shunt resistor in R3 location)

The application was built using LabVIEW 2011 and DAQmx driver version 9.3.5

Can somebody please tell me what could be the possible reason behind this phenomenon?

Thanks in advance!
Shweta JP

[johnsold]

Shewta_JP,

Since it occurs across all channels and is much faster than the strain signals, I would suspect some external electrical transient.  It appears to go both positive and negative.  How many samples are in the spikes?

Lynn

[Shweta_JP]

Hey Lynn!

It's just a 1 sample spike in the positive & negative direction. Could an electrical transient result in such a small duration spike??

Shweta JP

[johnsold]

Shweta JP,

I think an electrical transient is more likely than something mechanical.  A transient which lasts a few microseconds would show up that way.  If you were sampling at >=10E6 samples per second, then you might see more details.

If it is almost always a 1 sample spike in each direction, you could just replace those samples with mean baseline values (or zero).   Your real mechanical (strain) data always last longer than 2 samples.

Lynn

[Henrik_Volkers]

---

@Shweta_JP wrote:

Hey Lynn!

It's just a 1 sample spike in the positive & negative direction. Could an electrical transient result in such a small duration spike??

Shweta JP

---

EMC spikes could be of any length....

A good software spike filter is the median filter, here a filter rank of +-3 should do the job...

Looking at your picktures it seems that they appear in a close relation to your trigger ...  

Crystal_ball_mode: You wrote impact test.... is there some sort of solenoid/relay/inductive release involved ?   Without snubber they can be a source for such spikes.... 

Spoiler

[johnsold]

Henrik,

Nice portrait.

One of the spikes was before impact and one after.  That makes me think it is not something a simple as being related to the impact mechanism.

Lynn

[Henrik_Volkers]

Lynn

I didn't mean the impact pulse, my cristall ball told me that the impact energy is controlled by the drop higth and the trigger is coupled to the release ... the solenoid of the release is powered and switched off and than we have that current in the inductivity that just don't really want to stop flowing......

and here comes the spark ... spike 😄

Imagine a simple pushbutton driving the release solenoid.. and the triggers at rising edge the DAQ . rigth near the DAQ at the desk... no snubber... *EVIL_GRIN*   (Lets see how long the protection diodes at the triggerinput will last...)

[johnsold]

Henrik,

Inductive spikes make life interesting. And sometimes make smoke, too!

The reason I discounted spikes from the solenoid switch was that one of the spikes occurred after the impact. See Data with spike_1.png above.  I do not think the release mechanism is using a crystal ball to predict release times prior to the switch opening.

Lynn

[Henrik_Volkers]

Lynn,

I'm not talking about the impact. Usually you have some sort of mass that has a potential energy and grace to gravity it will magically get kinetic energy 🙂  This transformation can take some time ( fall time) under (nearly) constant acceleration.

So even if the impact is on different  time positions (as later as higher (the inital mass and the impact signal  😉 ), the spike is at about the same time, indicating a manual release under the asumtion that the beginning of the release pulse is used as the trigger.

How long would you press a mechanical RELEASE IMPACT MASS button ? 

Sweta migth tell us.

[Henrik_Volkers]

Maybe the trigger is not even hardware, just layer 8 😉

Spoiler

The operator hits the release button and the mouse button to start acqusition at the same time