adxl103 accelerometer as tilt sensor

[jginn11]

I was wondering if anyone has any experience with the adxl103 analog devices accelerometer. I am trying to obtain a sensor capable of measuring angles in an underwater robotic craft. I need single axis rotation measurement, and understand that this is a reasonable low cost sensor.

My problem is that I will be integrating using a digital system with data points every 100 ms or so, and am worried about obtaining a reasonably clean signal. Do you think that this is possible with this sensor and these conditions?

Also, the adxl103 measures "single axis" acceleration. Does this mean that is only measures acceleration in one LINEAR direction. If so, I could easily calculate angle using this, but will the tilt of the sensor cause errors.

Anyway, if anyone has any information or experience or any input whatsoever, it would be greatly appreciated.

Thanks.

 

[itsmoked]

I have used the ADXL202E with great success, (June 2001 Popular Science). It had digital out though.

What you are proposing should not be a problem with that device. I wouldn't hesitate to give it a go.

What might be a problem is soldering it. This is a small part with 0.025" pads this is non-trivial land!!

You will be laying out a board for this correct?

 

[IRstuff]

How do you get it to sense rotation?

TTFN

 

[itsmoked]

IRstuff;

The gravity vector poynting at China! As the device tilts the force diminishes. With a single axis it is harder! With two axis you can do it easier. With the single, you have to essentially tell it "you are now level" then through your actions, deduce which way you are rotating. Counter motions that you command can confirm measured rotations. I'd prolly use a two axis.

 

[sreid]

I'd suggest two sensors at 45 and 135 degrees from the horizontal. This is so you can distinguish between rotation (tilt) and horizontal or vertical accelerations. A little thought and some math will make this clear.

Do some analog pre-filtering to, say, 2.5 Hz to minimize alising-sampling at 10/sec limits the bandwidth to 5 Hz.

 

[itsmoked]

sreid; There is an additional part that's exactly the same but two crossed at 90 degrees. Then you just mount it as you say. I twould do the same. I would lay out the board so when it was mounted vertically it would be properly oriented.

 

[TurboXS]

There are also 3 axis accelerometers such as;

http://www.st.com/stonline/bin/sftab.exe?db=rosetta&type=query&table=444

I would imagine that an underwater craft would need a 3 axis (axes?) accelerometer to help determine it's orientation and rotation in any plane.

I have one on my desk but haven't had a chance to play with it yet.

regards
Pete
(TurboXS)

 

[sreid]

Also take a look at

http://www.microstrain.com/3dm-gx1.aspx

 

[felixc]

You may also want to look at the ADXRS150 from Analog, which is a silicon gyroscope. It may provide better movement feedback than your accelerometer, or complete its inputs for a total 3D movement tracking. If you use it as per their demo board schematics, the performance is nice.

 

[itsmoked]

Ah that is sweet felixc! But the price sure sucks...
$30 in 1000's {not to mention unavailable until this Friday which really means 3 months from now.}

 

[felixc]

Expensive? It depends what you want to do with it. Its closest competitor is a lot more expensive.

Before MEMS accelerometers could be used for tilt sensors, clinometers were (and still are) in the hundreds of dollars.

This gyro is very popular in products like video cameras, for image stabilization, and in cars. It isn't a brand new product, it is in production, but the demand is high. I designed it into a guidance system and it works fine.

 

[itsmoked]

felixc; Huh.. Why does the Analog site say available on 5/20/2005??

Can I ask what kind of vehicle did you design the guidance system into?

 

[felixc]

It may be because they're out of stock. Either because they're in allocation (nooo...) or they've had problems making it.

It went into a tractor.

 

[itsmoked]

Neat! Would've been the VERY last vehicle I would've guessed.

 

[rjgoebel]

TurboXS,

You could use a 3 axis accelerometer to determine orientation in 3d space but that is all you can do. There are 6 degrees of freedom associated with finding out WHERE the accelerometer is (translation about 3 axes), as well as its orientation (rotation about 3 axes). Now the question is, how many 3 axis accelerometers would you need to determine both?

How many 1 axis accelerometers would you need?

One solution that a company markets is based on an array of 9 one-axis accelerometers, and this device they sell is placed in the cockpit of a plane. It is used for determining when the parachute should open up for a fighter pilot that has bailed out. With the spinning and fast movements that go along with bailing out, accelerometers are very important for making sure the parachute doesnt pop open when a pilot is upside down.



Jim Goebel,
Electrical Engineer
Mid-West Forensics, Inc.

 

[Early2]

What I think jgin11 is trying to ask is
if designing with x-axis only mem then can the
design get accurate tilt response from say the 0 degree
angle without getting a mirrored response from 180 degree angle?

Or is dual axis the only way to go when calculating tilt?

 

[itsmoked]

You may never know on an original post this old...