MEMS-Accelerometer and instrumentational amplifier

[Abraham900]

Accelerometer and instrumentational amplifier

Hello.

I'm trying to build a mems accelerometer that can sense very small vibrationers at the milli g (g=9,82m/s^2) level. (It equals to a really low output voltage fyi)

So I have checked out this device at Analog Devices ADXL203

 http://www.analog.com/static/imported-files/data_sheets/ADXL103_203.pdf

.

Since the vibrations are at the milli g level I need to amplify these in order to work with these signals. I'm not very familiar with the instrumentation amplifiers but I was suggested to use INA826 for that part because of its properties.

So basically:

MEMS-Accelermometer (X,Y) --> Amplify --> DAQ Interface --> Labview.

But my problem here is the amplify part. Not entirely sure how to connect the instrumentation amplifier with the X,Y signals (two different signals of course) to an in-amp because I have never worked with them before.

Another issue that is on my mind is the bias part from the accelerometer. According to the datasheet, the bias is 2,5V if g=0 (that is, you get 2,5V on both x and y line if no acceleration/vibration is applied to the accelerometer). So I need to connect the Y or X to the positive input and 2,5V to the negative input on the in-amp to compensate for this bias so I can get a amplified signal or am I doing something wrong here?

Thanks!!

 

[FvM]

You would want to check in a first step, if the specified noise level of the MEMS sensor possibly dwarts your design idea. With full bandwidth, the noise level is in the order of several mg. Measuring below mg will be only possible with rather small bandwidth. Don't know what's the intended application bandwidth.

To remove the static bias, you'll peferably use a high-pass (or band-pass) fílter. There's no particular purpose of an instrumentation amplifier, an AC amplifier with bandpass charcteristic will do.

 

[speedEC]

Re: Accelerometer and instrumentational amplifier

microcontroller details? I think you don't need any amplifier. Just connect the Xout and Yout to your microcontroller and do ADC and read the ADXL outputs before you try with amplifier.

 

[Abraham900]

You would want to check in a first step, if the specified noise level of the MEMS sensor possibly dwarts your design idea. With full bandwidth, the noise level is in the order of several mg. Measuring below mg will be only possible with rather small bandwidth. Don't know what's the intended application bandwidth.

To remove the static bias, you'll peferably use a high-pass (or band-pass) fílter. There's no particular purpose of an instrumentation amplifier, an AC amplifier with bandpass charcteristic will do.

You are right. I also need to calculate the noise. I only know that the vibration is at the mg level. The exact numbers are not known to me so I intend to evaluate it. The bandwidth will be 2kHz. There's a formula in the data sheet so I will use it to calculate it right now

- - - Updated - - -

I'm not using a microcontroller. I will use a DAQ interface that samples the signals to the computer.

 

[FvM]

mg resolution with 2 kHz BW sounds like "lost in noise" or "no way". You need to refer to a more sensitive acceleration sensor, e.g. classical quartz acceleration sensor. https://en.wikipedia.org/wiki/Piezoelectric_accelerometer

 

[Abraham900]

Thanks.

Do you think it is possible to use two accelerometers for better SNR?

 

[FvM]

Averaging two acelerometers reduces the noise only by 3 dB (factor 0.7), not very promising.

 

[Abraham900]

Hello!

Just to update you on this matter I want to say the circuit did work really well (completed it in january so I am a bit late)... I could measure the heart frequency, both low >4Hz and 20-150Hz along with the breathing charateristics.

I can even measure between 1-150Hz and split up the signal into different channels to aqcuire different biosignals from one measurement.