Accelerometer gives wrong signals

[Gregory]

That datasheet makes it look like 0.532mV = 1g, or 1V = 1880g. Above you said 1V = 1000g, am I reading it wrong?

[StathPol]

The supplier said 1 V = 1000 g, yes...but we record g and then scale it to Volts, and then re-scale it back to g again?

[preston_johnson]

can you share the make and model of the accelerometer, we can look up the scaling factor

[altenbach]

---

@StathPol wrote:

...but we record g and then scale it to Volts, and then re-scale it back to g again?

---

That seems circular, starting with g and ending with g after a detour. That makes no sense.

[StathPol]

Thanks to all for your kind replies.

Preston Johnson -- Please find attached the certificate, setup of the accelerometer including the power supply, NI 6361 (voltage measurements) and the PXI-1073.

Regards

Stath

[StathPol]

In total, there are 4 accelerometers. Thats why the sensitivity in this particular certificate is 0.508 mV/g

[preston_johnson]

The data sheets indicate 0.5mv/g so 1V = 2000g.  I see the range of the accelerometer is 12,000g, or +/- 6V or so. 

You are sampling as RSE, why not differential?  I see from the photo it looks like you are sampling 4 channels. 

In the block diagram, you separate the channels into separate arrays, then immediately put them back into a multichannel waveform array again.  If you want just the one channel, why not simply index the array to the channel you are interested in. 

Is it possible to share a screen shot of the captured waveform?

[StathPol]

The range settings (g) on the power supply have or +/- 5V or +/- 10 V. Shall we then select 5Volts and specify in the MAX scale, the reverse setttings that 

- 10000 g= -5 Volts

0 g= 0 Volts

+10000 g = 5 Volts?

[preston_johnson]

I would leave the signal conditioner at +/- 10V to avoid clipping the signal. 

[Henrik_Volkers]

My hints:

-Set the conditioner to gain 1

-Use +-10V input range

-Scale the result in your LabVIEW program.  If you have access to the sound and vibration suite use it for aquisition and scaling.

You don't loose that much resolution ...  your measurement uncertainty is 2.15% for sine calibration accoring to the cal-sheet and since you measure transient small peaks with power in frequencies >10kHz  you migth end with 5% anyway....  If you match gain and range it's just a factor ~4  but you wouldn't capture out of sensor range peaks that easely occure with such tests.

Finally: If you do further calculations you migth end up in using metric units (m/s²) anyway 😉