Tilt Switch

All physical systems bounce. You have to debounce it with electronics, or microcontrollers.

Damp the bounce (capacitor), or use it to fire a trigger (transistor) that does not 'bounce' like the hard contacts do. That would 'fire' upon the first 'make', and ignore any subsequent intermittent 'make' 'break' sequences.

Ummm, Modelling Clay fired from a slingshot doesn't bounce.

It splats. :-)

Silly Putty bounces.

That "splat" is actually the sound of billions and billions of tiny, clay-particle-sized bounces. ;-)

Cheers! Rich

Sounds like you've been watching Feynmans youtube videos!

There are electrolytic inclination sensors that don't bounce; some are very nicely damped. Of course, their output is analog, not a contact closure.

I should think that there would be mercury tilt sensors that have hysteresis and don't bounce. Mercury relays can generate nanosecond edges without bounce.

But yes, the simple fix is a little electronics, like an RC lowpass filter and a Schmitt gate.

John

Nope!

More like millions of little clay freight cars getting smashed the f*ck up by the string of freight cars behind them.

Less than 0.000001% bounce. i.e. do a 180.

ISTR a video on Tru TV or the like where a whole string of passenger cars all telescoped into each other at a bridge abutment.

I don't think a whole lotta bounce was goin on there either. Robert Plant had already left the building...

There are gyro-on-a-chip deals around that would allow for a far bigger gamut of control over any tilt angle sensor, and likely not cost much more.

Hell, a strain gauge and the rolling ball would work without bounce. Put the ball in a cardboard tube,and put the strain gauge at the sense end. Set the tube angle so that a 30° tilt causes the ball to roll to the gauge end, setting off the sense electronics. Far better than a hard switch.

Thanks a lot for the ideas, John.

Use a pair of accelerometers, or a dual axis accelerometer. As Einstein pointed out, gravity looks just like acceleration, and if you orient your accelerometers correctly, both will register zero acceleration/gravity when your object is in its regular orieintation, and one - usually both - will register some kind of "acceleration" when the object is tilted.

Analog Devices offer a range of such parts

They do contain moving parts, but the part involved doesn't move much, and doesn't hit anything, so contact bounce isn't a problem.

-- Bill Sloman, Nijmegen

Faced with calibrating the oft-erratic run-back timers on several *noisy* shakers, I had the idea of connecting a vibration switch to a lithium-cell powered timer module: Shake the switch and the clock counts minutes & seconds...

After some thought, I mounted two sensors back-to-back.

To get them to work reliably, I just had to put a small capacitor across the contacts. I can't remember the exact value, possibly 4u7 or 10u, because I grabbed a range from 0.1 u upwards, and tried them until it worked...

FWIW, I built the timer module into a water-resistant lunch-box and used

The rig was sufficiently sensitive that I was later able to use it for placid 'orbital' shakers, too.

;-)

--------------------------------------- Posted through

Since John Fields is both dishonest and stupid, he chose an expensive SPI output accelerometer and then tacked on all the cost involved in decoding the SPI output in order to be able to claim that my solution was expensive.

The analog output two and three-axis accelerometers on the same page of the Analog Devices web-site are appreciably cheaper, and lend themselves to rather cheaper signal processing. What this would look like rather depends on what the OP actually needed. For all we know a digital or even an SPI output might actually be more suitable and worth paying extra for.

-- Bill Sloman, Nijmegen

-- Bill Sloman, Nijmegen

oth

is

ne

one

...

But rather more reliable - fewer moving parts,

That's what he says he wants. Customers tend to become moe demanding if they are exposed to more capable components, not that you would be likely to have run into this.

-- Bill Sloman, Nijmegen

These people

are the classic liquid inclination sensor folks.

Somebody, Hineywell maybe?, made a cute sensor, two patterened printed-circuit boards in a can, half filled with silicone oil. The capacitance between the boards changed as the thing was tilted.

Fun, but probably you could use a cheap mechanical tilt switch and some simple debounce electronics.

John

te:

, both

uit is

d one

of one

you

on,

htm...

uch,

The sensitivity wouldn't be a problem. If the OP's system was digital enough to look at a digital output, the "pot" would just be a memory location. Everything is subject to temperature and supply drift - the interesting question is (as always) how much drift the OP can live with.

And the "more expensive" does rather depend on what it is being compared with, and by whom. Your last "costing" threw in the cost of the development system for a micro to handle the SPI output from one of the more expensive accelerometers on the web-page. It was nice of you to concede that - after a second look - the web-page did include cheaper accelerometers with outputs that might be cheaper to process, but one has to wonder how you managed to miss them the first time around.

-- Bill Sloman, Nijmegen