Accelerometer/INS question

Nanotech accelerometers are showing up which offer relatively cheap,
mass-producible devices (to the point that laptops may be able to safe
the heads before the floor arrives ;-) -- but how do they do on
accuracy? Are the higher grade units up there with laser ring
devices? Or the level required for space applications?

/dps

dave schneider wrote:
Nanotech accelerometers are showing up which offer relatively cheap,
mass-producible devices (to the point that laptops may be able to safe
the heads before the floor arrives ;-) -- but how do they do on
accuracy? Are the higher grade units up there with laser ring
devices? Or the level required for space applications?

Lousy.
I think you mean micromachined, not nanotech.
The cheapest units read around 2G or so, with an error of some .1m/s^2 or
so.
Not very good.

Space applications are many things, for some things this level of
precision is totally fine.
For inertial navigation, for over a few seconds, it tends to be a problem.

Not my area of expertise, but...
To do an INS you need 3 accelerometers and 3 gyroscopes.

There are low cost MEM's sensors for both.
The accelerometers are getting pretty good, but the gyroscopes are not so hot.

1)The MEMS gyros I know of are rate gyros, they give you the first
derivative of what you really want.

2)They have a lot of drift.

Both MEMS devices are good enough for in atmosphere flight where you can use GPS and the gravity vector
to remove the errors.

You can assume that the 1Gee gravity vector is aways there
(or will return shortly after brief periods of non 1 G)


Example of the best MEMS gyro I know of:

http://www.analog.com/UploadedFiles/...ADXRS150_B.pdf
Note the drift and zero specs....


Paul




On 28 Jun 2004 12:19:00 -0700, (dave schneider) wrote:

Nanotech accelerometers are showing up which offer relatively cheap,
mass-producible devices (to the point that laptops may be able to safe
the heads before the floor arrives ;-) -- but how do they do on
accuracy? Are the higher grade units up there with laser ring
devices? Or the level required for space applications?

/dps

dave schneider wrote:
Nanotech accelerometers are showing up which offer relatively cheap,
mass-producible devices (to the point that laptops may be able to safe
the heads before the floor arrives ;-) -- but how do they do on
accuracy? Are the higher grade units up there with laser ring
devices? Or the level required for space applications?

The new chip based gyros and accellerometers aren't nanotech
per se, they're micromechanical electronic systems (MEMS).

Laser Ring Gyros are mostly fading out of use, replaced by
fiber-optic gyros.

A good comparison of performance is found in the Crossbow
IMU units:
http://www.xbow.com/Products/product...ls.aspx?sid=26

The IMU400 is their best MEMS based product, the IMU700 is
their best current FOG based (uses MEMS accellerometers
but fiber gyros).

Random walk for the IMU400 is about 2.25 deg/hr^0.5,
(the IMU-400CC-100, which has a 100 deg/sec max rotation rate
limit.. the IMU-400CC-200 has a 200 deg/sec limit and
almost exactly twice that random walk).

Random walk for the IMU700 is 0.4 deg/hr^0.5.


-george william herbert

In article ,
wrote:
Both MEMS devices are good enough for in atmosphere flight where you
can use GPS and the gravity vector to remove the errors.
You can assume that the 1Gee gravity vector is aways there
(or will return shortly after brief periods of non 1 G)

Unfortunately, it's perfectly possible to have a nice clean 1G while the
aircraft is doing something very unpleasant and imminently fatal. In
flight in the atmosphere, what an accelerometer is measuring is the lift
imparted by the wings, **NOT** the gravity vector. (This is why a good
airliner pilot -- or a good airliner autopilot -- can do a substantial
turn, banked to a fair angle, without spilling your drink. The drink is
responding to lift, not gravity.)
--
"Think outside the box -- the box isn't our friend." | Henry Spencer
-- George Herbert |

Unfortunately, it's perfectly possible to have a nice clean 1G while the
aircraft is doing something very unpleasant and imminently fatal. In
flight in the atmosphere, what an accelerometer is measuring is the lift
imparted by the wings, **NOT** the gravity vector. (This is why a good
airliner pilot -- or a good airliner autopilot -- can do a substantial
turn, banked to a fair angle, without spilling your drink. The drink is
responding to lift, not gravity.)

(I'm the pbreed of an earlier post , now responding from a different
computer)

The mechanical gyro hroizion in most aircraft are driven by
differential air presure between the cabin and a motor driven vacume
pump. These devices have weighted vanes that try to erect the gyro to
the local 1G vector. The erection time period is long, but it is
there.

Make a standard rate turn (3 degrees per second) and hold that for 720
degrees, when you return to level you will notice a pronounced
horizion tilt.

When flying on insturments you adjust to these instrument errors by
crosreferencing multiple instruments.

The Horizon Bank angle is backed up by the rate of turn indicator, the
gyro compass, and the magnetic compass, and finnaly the navigation
radios.


The horizion pitch angle is backed up by the altimeter, airspeed, rate
of climb and knowledge of what the aircraft should be doing at various
power settings.

I just finished my instrument rating in January and you spend some
significant time on identifiying failed systems so you can throw out
eronious indications.


As for my earlier statement about MEMS devices being good enough for
atmospheric flight, It would be very difficult to maintain a steady 1G
acceleration with steady state turn rates less than the MEMS drift
rates of ~1deg/sec. Combine this information with 3D sensing of the
eaths magnetic field, and GPS postion and it is possible to have an
excelent AHRS (attitude, heading refernce system) using MEMS devices.

It is my understanding that the New Garmin G1000 avionics suite uses
MEMS sensors. I can't currently find the referencebut I believe that
they did a lot of testing against a traditional FOG (Fiber optic gyro)
based system.

Take a look at the last paragraph he
http://www.flyingmag.com/article.asp...ection_i d=17

Paul

Ian Stirling wrote:
Lousy.
I think you mean micromachined, not nanotech.
The cheapest units read around 2G or so, with an error of some .1m/s^2 or
so.
Not very good.

Okay, MEMS -- I was indeed thinking of the beams with the resonance
cavity etched out beneath them.

And I did expect the cheapest units to be, well, the cheapest units,
but was wondering what the current and expected upper ends looked
like.

Thanks for the help, Ian and George, and for the links.

/dps

wrote in message . ..
Not my area of expertise, but...
To do an INS you need 3 accelerometers and 3 gyroscopes.

There are low cost MEM's sensors for both.
The accelerometers are getting pretty good, but the gyroscopes are not so hot.

1)The MEMS gyros I know of are rate gyros, they give you the first
derivative of what you really want.

2)They have a lot of drift.


And all of those parameters (some more than others) swing
significantly with temperature.

In article ,
wrote:
The mechanical gyro hroizion...
...These devices have weighted vanes that try to erect the gyro to
the local 1G vector. The erection time period is long, but it is there.
Make a standard rate turn (3 degrees per second) and hold that for 720
degrees, when you return to level you will notice a pronounced
horizion tilt.

Which means that it's *not* erecting to the local gravity vector, it's
erecting to the local lift vector. The assumption is that over a
sufficiently long period of time, the aircraft will on average be flying
straight and level, i.e. with those two vectors aligned, and so a heavily
filtered version of the lift vector can be used as a proxy for the gravity
vector.

Remember, gravity pulls on all parts of the aircraft equally. It doesn't
pull on the gyro-horizon vanes any harder (per kilogram) than it pulls on
the wing spar. Such an instrument *can't* sense gravity directly. It
gets erected by torque on those vanes, and a uniform gravity field can't
produce a torque -- it accelerates everything equally. (Very small
torques *are* produced by real gravity fields because they aren't quite
uniform, but those are too small to be an issue for this.)

Consider the aircraft sitting on the ground. Gravity is pulling on it;
why doesn't it fall downward? Because the ground is pushing up on its
tires. *That* is the force such an instrument senses when the aircraft is
on the ground -- not the force of gravity -- because *that* force is
applied locally, only to the tires, which then transmit it to other parts.
It's the transmission of a locally-applied force that can create torques,
and those torques are what erect the gyro horizon.

In flight, lift takes over the same role... but there's nothing that
inherently constrains lift to be (on average) vertical, except the pilot's
desire to survive. :-)

As for my earlier statement about MEMS devices being good enough for
atmospheric flight, It would be very difficult to maintain a steady 1G
acceleration with steady state turn rates less than the MEMS drift
rates of ~1deg/sec. Combine this information with 3D sensing of the
eaths magnetic field, and GPS postion and it is possible to have an
excelent AHRS (attitude, heading refernce system) using MEMS devices.

Yes, I quite agree -- what I was disputing was the statement that the
gravity vector itself is a useful reference, which it's not. It's the
MEMS gyros and the other added sensors, plus the constraints imposed by
the aircraft aerodynamics, which make such combinations workable. Alas,
those don't always read over to other applications.

(Consider putting the aircraft into a very slight bank, while adjusting
pitch etc. to maintain 1G of lift. Because of the bank, the aircraft will
be turning slowly; because it's very slow, the MEMS gyros can't reliably
pick it up. Because of the bank, that 1G of lift is not pointed exactly
upward, so its vertical component is less than 1G... and so the aircraft
is accelerating downward. GPS position change or magnetic field may
reveal the turn, and aerodynamics will try to point the aircraft into the
wind and thus reveal the growing sink rate, but the MEMS gear itself won't
notice unless cued by one of those things.)
--
"Think outside the box -- the box isn't our friend." | Henry Spencer
-- George Herbert |