Calibration of Accelerometers

[Rama2141]

I have a few uniaxial and triaxial accelerometers (voltage-based). How do I calibrate them?

[JÞB]

Generally.   Rotate them for a maximum output and a minimum output on each axis .

If you are somewhere near the planet Earth that should give you points representing +1g and-1g. The scale will be linear.  

Since the Earth doesn't significantly change its mass often its fairly repeatable. 

[Bob_Schor]

I agree 100% with Jay.  We had some students working with little triaxial accelerometers (they were Biomedical Engineering seniors, so maybe we shouldn't have expected too much) and they also couldn't figure out this problem.

One "feature" of many Accelerometer chips is that, internally, they are usually pretty close to having orthogonal axes, and are manufactured with parallel opposite faces, so if you flip it over, the axis giving you the largest (in absolute value) signal should simply reverse its sign.

Something you might consider is to test your accelerometer in a known orientation, namely one that you know is aligned with the direction of gravity (let's call this +Z, or -Z, depending on your convention).  To do this requires you establish an accurate XY plane to place the accelerometer so its face points along the Z axis.  How to do this?  Ever heard of a Bubble Level?

For Extra Credit, there's a way to calibrate (most) tri-axial (or uni-axial) accelerometers using simply a planar surface, without knowing its orientation with respect to gravity.  You only need to know that the package containing the accelerometer is a rectangular solid (opposite faces parallel, adjacent faces perpendicular).  Think about it.

Bob Schor

[JÞB]

The bubble level works well for uniaxial applications.   

However, if the Z axis is normal to a horizontal plane the X and Y outputs will be 0!  Since the Sine of the angle is proportional to the outputs, finding the orientation that zeros X and Y is a lot more sensitive than maximizing the Z output. 

Bob, we won't depend on assuming a V/g scale and using a single point and some basic trigonometry... we are calibrating V/g according to the OP.  Luckily there is a good 1g reference field handy.

[Bob_Schor]

Jay,

     You missed the point where I "assumed" the Accelerometer was embedded in a rectangular solid of some sort.  The one I was using allowed you to hold it "flat" against a surface, "on the long edge" against the surface (a 90° rotation), or "on the short edge" (90° about an axis at right angles to "up" and the previous "long" axis).  Doing this gives you all three axes.

     The accelerometer chip the students were using scaled it in terms of volts/g.  If memory serves, there was a DC Bias of 1.5v, and the gain of the chip was 0.3 v/g.  The trick was that the tolerance reported was ±10%, so if you were attempting to get any useful quantitative measurements from such a chip, calibration was a necessity.  Having two unknowns about three axes requires (excuse me which I break out the Calculator program ...) 6 measurements under 6 different orientations, which I called "upright, upside-down, right-side-down, left-side-down, nose-down, and tail-down".  Given a stable surface (such as a table-top), 5 of these were trivial (hold the chip in the given position without moving it while taking 1" worth of readings), and the 6th just required a steady hand and accurate eye (because the wires came out of one end of the chip and sort-of "got in the way").

BS

[mcduff]

Devices can be bought that can be used to calibrate accelerometers. They usually vibrate at 159Hz with 1 g of acceleration. See https://www.pcb.com/sensors-for-test-measurement/calibration/portable-vibration

Excuse my ignorance and stupidity here, but the OP never said what type of accelerometer they had; only it was triaxial. (Note sure what voltage based means here.)

Most of the work I have done, and the equipment from NI, use IEPE accelerometers. NI makes lots of digitizers that have IEPE excitation. Here a 2 mA or so bias current is supplied to the accelerometer to power it. When this excitation current is supplied the digitizer needs to be AC Coupled. As Jay said earlier, the earth's mass is not periodically changing thus a static acceleration would not be visible with an AC coupled accelerometer. Thus you may need some sort of calibrated shaker. If you have DC coupled accelerometer, like ones in your phone or a myRIO(I think), then you should be able to measure g from the earth.  

mcduff

[Bob_Schor]

Good point, McDuff!  My experience is with the DC variety (like phones, myRIO, hobbyist packages), where having a rather large mass underfoot gives the (static) stimulus usable for calibration.

BS