I built a small motion detection device (with a tri axial
accelerometer) and I want to test it, I need to put it through a fixed
set of known motions (1 to 6 g's) so I can verify it is working
correctly. I suppose I could glue it to a hot wheel and let it roll
down a track with a loop but I don't think that will be too repeatable.
Any ideas for a simple way to test it?
If you rotate it slowly, gravity will give you a 1G at any angle you
hold it. Then if you put a variable length arm on it, and rotate it at
constant speeds in the vertical plane, you will be putting a constant
component on it. The faster you rotate, or the longer the arm, the
greater the acceleration. Don't forget your output will contain a +/-
1G sin wave, because gravity will add or subtract to the rotation. So
let the sin wave be your standard to measure by ratio the constant part
you add with the rotation.
If you want to eliminate the sin wave, turn the apparatus over, and
rotate in the horizontal plane.
Randy M. Dumse
Caution: Objects in mirror are more confused than they appear.
If you need to attach wires and the rotation creates a problem, maybe you
could hang it from a spring (along with weights of different sizes) and let
it bob up and down. You should be able to calculate G force based on the
distance the spring moves and the time it takes. You can control the
distance easily by just pulling the spring down to a known starting point.
And your hardware I would assume could give you accurate time measurements.
Though getting it up to 6 Gs might require a fairly strong and maybe long
You could also use a fixed length pendulum but creating 6Gs I assume would
be even harder in that case.
You can position your device towards the gravitation vector. It will
give you the 1g reference in any direction. If you insist on more or
less gravitation, you can go to the Moon, Jupiter and so on.
DSP and Mixed Signal Design Consultant
I easily managed a 5-G test by swinging the sensor (and logger and
battery pack) on a 15-foot nylon cord out in the side yard.
A better test would be to swing in a vertical circle off the edge
of a building. In that case you get +1g on the bottom and
-1G on the top of the circle. From this you can, with a bit
of analysis, measure the period. With this you can calculate the
rotation rate and the centripetal acceleration. From that
you can also get the real gyro rate for calibration.
Most accelerometers are probably pretty linear between 1 and 5Gs, so
just inverting the sensor gives you the -1 and +1 points so that you
can scale the output.
Thermal and time-related drift in the gyros is the major problem that
I have encountered with the less expensive sensors.
I fully agree with you on that, Mark. The MEMS devices are linear in
their operating range to the accuracy of about 1e-3. Zero drift is the
major source of error, and it can be as high as several per cent. In the
addition to the thermal drifts and aging, zero may be permanently
shifted by the accidental shock.
DSP and Mixed Signal Design Consultant
But given that you probably don't want to wait until you can actually
afford bringing your device under test to one of them, I guess you'll
have to make do with what's available: a centrifuge or something like it.
I don't think the turntables mentioned before will yield those 6 gees
you wanted. The device-under-test would have to be mounted about 1
meter away from the axis at 0.5 revs/sec --- the motor will most likely
not be able to move the weight and air drag of a disk or lever arm that
big. A bicycle's rear wheel (gear shift included) coupled to a suitable
motor would be a better bet.
Given the level of accuracy in the low-cost MEMS units, it seems to me
that the average kids playground will have exactly what you need. And
it's free. You just have to chase some children away.
You know the diameter of the merrygoround, and using a fixed position
get an approximate RPM. Ther mass of the thing acts as a flywheel, so
any slowdown in speed is gradual. Sit on the edge of the thing with your
accellerometer in your lap, and take readings.
Also don't forget you can measure g forces quite well with sudden stops,
which are reproducible given a simple setup rig. This is, after all,
what many of these MEMS units are made for: airbag deployment systems.
Keep in mind that that piezo accellerometers will not register dc
gravity, so what folks have been saying about the constant gravity
component doesn't apply to piezo units. These only register change in
Well I'm trying for 6-7 bit accuracy, which the cheap MEMS units are
capable, about 1 out of 100, that means 1 degree off and the gravity
vector will probably screw my readings up, so laying it on my lap will
I see now what I need is a "rate table" as descibed on that web site.
Maybe something will turn up used on ebay.... I see one just sold for
You can actually build your own rate table. All you need is a piece of
(plastic plywood, etc) cut into the shape of a circle. Then you spin it up
at a known number of rpm's. The centripetal acceleration is proportional to
both the angular speed and the radius (radius being the distance of the
accelerometer from the center).
I am new to this forum, and, as I said last week when I posted 'where do I
start' and got lots of help, I have finals the next few weeks, but the
equations are fairly simple for me. I have a BA in Physics and I am now
studying for my MSc in Applied Physics.
So if you can build a circular platform, and run it with a motor at varying
speeds, and know to within a reasonable degree how many rpms it is going, I
can calculate the number of rpm you will need to run it in order to get the
number of g's you need. I also worked with guidance and control systems for
satellites and other vehicles after I received my undergrad degree, so I
know a little something about rate tables. We used to monitor the signals
coming from the unit under test (usually 3 gyros and 3 accelerometers) using
a plug that connected to the unit thru the slip rings on the table. That is,
when the table rotated, the signals would be routed thru the table, down
into the assembly which consisted of conducting 'brushes' and we picked up
the signals from the guidance package with a computer. The computer would
then analyze the data and we could tell if the unit met the customer's
requirements. I don't' know if you need to get that sophisticated, but maybe
you could supply me with a link to that particular accelerometer. You said
it was triaxial, so you can measure all three axes? X,Y,Z, or as we used to
call them, roll, pitch, and yaw. I didn't even know they made them for the
general public. I have been out of that business for awhile.
But that aside. I know the basic physics of what you are trying to do, and
if you can do the mechanical part, let me know. I am kind of busy right now
studying for finals, getting homeworks done, etc but I would probably have
some time over the weekend to work it out for you if you would like.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.