Tilt sensor

Have you considered a pot with a weight hanging from the spindle?

How about a joystick with potmeters? Remove the springs. Mount it with the stick facing the earth and attach a small wheight to the stick. Now the stick will always point to the earth.

Use an incremental pulse potentiometer,with a small weight hung onto the shaft. (example:SHARP GP-1R52C 360 pulses/rotation)

As mentioned, a pot with a weight would be cheap and easy to signal condition.

These are electrolytic:

Or I suppose you could use mems accelerometers, Analog Devices maybe.

John

John

You want a tilt sensor for student projects. Is sub-$15 okay through distribution? Both digital and analog output formats are available for this one series of devices from STMicroelectronics:

On Mon, 10 Apr 2006 20:38:47 +0100, "Slurp" Gave us:

You used to cook your soup with a plumbed pot? Hehehhe... just kidding.

Most pots have a fair amount of resistance to turning.

A round slit wheel, and optical sensor array would work. Those found in old mice might work, but probably don't resolve well enough. The slit wheel from the back of an old industrial motor (surplus) will resolve to single or half degrees usually. Some have a dual, staggered sensor so the wheel slit resolution gets doubled.

An analog version might be the tried and true pinball machine tilt mechanism with a capacitive sensor in the ring section. That will sense any tilt in any direction, but without sectoring it, you would only know the amount of tilt, not the direction. With a sectored ring, and quite a bit more circuitry, you could detect tilt amount and direction.

On Mon, 10 Apr 2006 19:38:55 GMT, snipped-for-privacy@puntnl.niks (Nico Coesel) Gave us:

Great transducer as long as all resistance (damping) to movement and any return to center springs are removed. Then, the problem of the read circuitry comes into it. Probably only a slight variant of the calibration applets old joysticke used to come with. Most of those gave coordinate readout. One would have to read all four pots at all times, and extrapolate tilt value and direction from the four readings. Also, it would likely only work with a fairly big (heavy) bob on a fairly long rod. His package size requisite wasn't mentioned.

On Mon, 10 Apr 2006 21:45:30 +0200, Sjouke Burry Gave us:

If the weight could turn the damped shaft, the sensor would only give answers in one plane. Do pulse pots have sub two degree signaling?

No ,just use two perpendicular pots,if you want, however I dont see a request for a 2 directional tilt measurement.

Hello Joseph,

I could never match the talent of all these engineers but as a garage hack and homebred tinkerer your singing my song. (:

Take a 2.5" (6.25 cm) clear plastic bottle, fill halfway with water. Make yourself a calibrated strip to stick on the side, zero would be at the exact point of level water. Tilt it to 60 degrees either way and mark it. I would imagine the scale and the rest of the markings would be linear. You could also use a 1" or larger pvc pipe glued in the center vented for water flow to force the water to stabilize quicker. Electronically you could do fancy things to monitor the water's angle relative to the walls. A larger diameter bottle will give you a greater sweep distance per tilt angle.

It's 3am, I will wake up tomorrow and go "what was I thinking"

Just a thought,

• * * Christopher

Temecula CA.USA

You can use a dual-axis accelerometer, e.g. ADXL311 from Analog Devices: accuracy to 0.1 degree of inclination, available from Digikey for $8.5 in ones.

greetings, Tom

I'd go for pins from the top and a strip at teh bottom. The strip at the bottom is ground; measure teh conductifity from the two other pins that are more-or-less inserted in water (or other liquid).

Thomas

Calibration is easy. Hold the entire contraption in the 'zero' position and hit 'zero' on the read-out device/application. Besides, PC style joysticks have 2 potmeters which are directly to the PC.

I think you will run into huge errors with this method when the angles are small due to noise and inaccuracies in the accelerometers.

Which is why appropriate noise filtering is required in an accelerometer-based system.

The Analog Devices accelerometers have noise specified 320 ?g/rtHz or less. I can't find noise data on the ST parts but a 3-axis accelerometer with digital output has settable decimation values for bandwidths down to 10 Hz.

sorry... 300 ug/rtHz (micro-g)

On Tue, 11 Apr 2006 19:56:38 GMT, snipped-for-privacy@puntnl.niks (Nico Coesel) Gave us:

What makes you think a PC is being used? I was talking about using the calibration software as a primer for the software he would need to write to do the job.

On Tue, 11 Apr 2006 08:05:45 +0200, Sjouke Burry Gave us:

He merely said "tilt". On a pinball machine, that is any direction. In industry, it likely refers to any direction as well.

Two pots would require two swing bobs, and would have to be separated, and both would have to be read to extrapolate the answer.

This suggestion is very similar to the joystick suggestion, but much more obtrusive.

On 10 Apr 2006 16:07:46 -0700, "John_H" Gave us:

Nice item, but it also requires a lot of support circuitry, as well as software. Hell a G force resolver would work, if you want to get into advanced technical solutions.