MEMS accelerometer: noise reduction and improve resolution

[elberto]

Sorry, 2,8mg of noise is wrong value, real noise is pk-pk noise from formula.

No, I take care only of noise generated inside sensor.

But I'm thinking to leave digital output accelerometer and use analog output ones, so It would be much easier use oversampling.

[johnsold]

With the analog sensor you have complete control over your sensing and signal processing. What you lose is the optimizations which Freescale or other manufacturers have put into thier digital devices. However, if those optimizations do not match your specific situation, then doing it yourself is what you need to do.

If you choose to go to an analog accelerometer, look closely at noise, sensitivity, hysteresis, and related paramters to see if the device is good enough for your needs.  If those parameters are specified at 0.1% of full scale or larger, it may be very difficult to get to 16 or 18 bits of effective resolution.

Lynn

[Henrik_Volkers]

---

@elberto wrote:

Target accelerometers has got about 300ng PSD value, at 24 bit resolution, I would likr to know what can I do with commercial accelerometers.

---

Good analog accelrometers have a linear useable dynamic range over 6 decades, but you don't use them for seismic measurements.

It's hard to beat noise 🙂

In electronics the resistive thermal noise is your enemy,  with a membrane type microphone the thermal noise of the air,

the very low seismic mass to gain more sensivity...

More data or more sensors just give you 1/sqrt(n) less noise and building an analog frontend to resolve that 24 bit is an art!

 No matter how many folding rulers you use and how many times you read them, you will not be able to resolve µm 😄

Switch to cheap commercial geophones! 

Seismometer have typical resolutions of 10e-11g, I would use my time in building one of those instead of trying to beat physics.

http://alabamaquake.com/equipment.html

and here a nice german side:

http://forum.untertage.com/viewtopic.php?f=2&t=6016&start=30