Accelerometer signal (+/- 10,000 g) help

[StathPol]

Hi all, 

I am using an accelerometer for vibration measurements. 

The accelerometer is connected with an external excitation device to provide excitation of the accelerometer (Coupler-Kistler). The coupler is then connected with the NI PXIe-6361 module which is connected on the NI PXI3-1073 chassis. Based on gain of 1 and excitation voltate of +/- 10V, the accelerometer can provide +/- 10,000 g. I am using DAQ assistant at the moment to see the signal and then moved to DAQmx. I have selected the voltage option instead of voltage with excitation and I have set the parameters +/- 10V for max and min signal values, RSE for terminal configuration. I have also used the scaling table option for -10,000 g for -10 V and 10,000 g for 10 V. The accelerometer does not seem to be working properly as the signal does not change at all. Is there any possibility to advice me what is going wrong with the acceleration setup  please?

Thanks.

[Henrik_Volkers]

Please provide a setup schematic and the type of accelerometer (IEPE? Charge? Piezoresistive?)

And what device are you using to provide 10k g ?  (I use Hopkinson bars)

[StathPol]

Hi Henrik_Volkers, 

Thanks for your reply. Please find attached a schematic diagram of the sensor with the hardware and the relevant information for your convenience to visualise the instrumentation setup. The accelerometer is connected with the Coupler to provide excitation voltage to the accelerometer. The Coupler is then connected with the NI 6361 card thought a BNC to 2 - wire which represents a voltage output and ground. The sensor sensitivity is 0.516 mV/g.

The gain should be 1 in order to provide 10,000 g. I have tried several things to make the sensor work properly but unfortunately there is no signal at all from the sensor (i.e. voltage, voltage with excitation etc). I was also hoping to see a voltage output from the DAQ assistant before I move to DAQmx but the signal is not properly measured. 

Thanks a lot. 

[Henrik_Volkers]

First some test to check the sensor and coupler:

Needed equipment: a Multimeter (handheld) or a scope, 20V range , a BNC  T-connector, cable BNC to DMM or scope.

Maybe a 2.5 kOhm resistor, (well something in the range of 2k to 2.7k is fine, I build mine out of 4 10k 😉 )

check the sensor coupler setup:

Insert the T-connector at the sensor input at the coupler, connect the sensor, apply power to the coupler and measure the bias voltage.

You should read ~11V ( something in the range 10V - 12V) bias voltage.

if not, the sensor or the cable or the coupler(setting?) is not OK

check the supply current:

replace the sensor with the 2.5k resistor, you should read 10V with a 4mA supply current (U=R*I)

If you have scope and you can see the 11V in DC mode, switch to AC mode 5mV/div and knock on the sensor, you should be able to see it in the signal.

Next check your DAQ input: apply a known signal and try to read it: Your fingertips and a long unshielded open cable should give line hum+ noise,  or use a line out or phone out of a PC ( or smartphone, or ...) as a signal generator,

(if you are afraid, couple it with 10k and 100nF (or 1µF) in series in each line)

audacity is a nice free tool to create signals at the sound output, however you can write a LabVIEW tool  or use the provided 'Generate sound.vi' example 😉

[Henrik_Volkers · Accepted solution]

How to setup the coupler/measurement chain:

Your sensor has a sensitivty (say it's constant for a given frequency range) S_ug [mV/g] and you add the coupler with a gain of G_c [V/V=1] . so you get S_total [mV/g] = S_ug*G_c

Your DAQ can read +-5V (read the spec in detail, usually they can read up to 5.5V)  and the max. expected acceleration is? 

You can always decrase or increase the G_c  to measure your signals in a optimal range.

You read a voltage u(t) and want g(t) = u(t) / S_total.    You can do that conversion in your program, or tell MAX / DAQmx to do that already in the driver.

You have to read the manual of the coupler on how to setup the gain and/or sensitivity.

If you measure pulses, do a FFT on the signal and lockout for signals higher than 10 kHz ...  you can still use the sensor ( up to maybe 40-60 kHz) but some more signal processing is involved in that case. (or a higher uncertaincy)

If you have the transfer functions of your transducer and the coupler, you can convert your signal into the frequency domain, compensate the transfer functions and convert bach into time domain, to get a better estinate of your acceleration.

[StathPol]

I have just followed your suggestions. The coupler is the problem. 🙂 

Many thanks, 

[StathPol]

Hi Hernrik_Volkers, 

I have written a small DAQmx code to check the voltage readings from the accelerometers. I am wondering whether there is any test to perform in order to ensure that the accelerometer reading is correct. 

Thank you, 

Pepis21

[Henrik_Volkers]

Do you have a shaker?

[StathPol]

Hi Henrik, 

Yes I do have a shaker 🙂 ..What for?

Thanks. 

[StathPol]

Well, we had one but it broke. Is there not an alternative simple way to possibly verify the readings?