Inclinometer/tilt sensor help requested

[MacGyverS2000]

Anyone have some info on low-cost inclinometer/tilt sensors, be it driving/signal conditioning schematics, favorite vendors, etc.? I like the simplicity of the solid-state MEMS style units, but the cost can be a bit much for this project ($10+)... I would prefer to keep the cost under $10, preferrably in the $5 range (if at all possible). Electrolytic sensors are far less expensive, but require AC-drive (and generally more signal conditioning) circuitry, which means more time spent on design and debugging.

A tilt range of +/- 45 is probably the minimum, with +/- 90 being ideal. The required accuracy and drift is very low (a degree or two), so long as drift isn't cumulative.

So far, I have the following MEMS units:
* SignalQuest
* VTI

and the following electrolytic units:
* Fredericks
* Spectrum Sensors
* AOSI

 

[VEBill]

Assuming that you have gravity...

Analog solution - A linear potentiometer with a weighted arm attached to the shaft. A typical pot will have a 270 degree turning range. The circuit becomes trivial - just use opamps (if req'd) for offset and gain.

Digital solution - A similar arrangement (the weighted arm) with an shaft encoder, preferably an absolute encoder so that start-up is a nonissue.

 

[blcpro]

Check the listings for "tilt sensor" on

http://www.motionnet.com

You may find some leads there.

 

[MacGyverS2000]

VE1, weights won't work in this situation as more than one axis is required for measurement (and the object can shift in all 3 dimensions). As the plane of the weight shifts towards the horizontal, measurement will no longer be possible.


blc, I'll take a look...

 

[VEBill]

So you need to measure both pitch and roll. It is probably a 2-D problem (unless you also need yaw - a different problem as yaw doesn't usually include a gravity reference).

So you could use two pots or encoders. To prevent jam up due to tilt in the other axis, it would be trivial to include pivots on the weighted arms. The pivots would allow them to swing in the other axis so they won't jam up.

You could also use an analog joystick with the same weighted arm concept. An analog joystick is typically a pair of pots (often 100k) connected to the handle by a very simple mechanism (those pivots and a slotted internal lever) that allows each axis to move independantly.

One limiting factor might be that the range for a typical analog joystick is probably at least +/- 45 degrees, but certainly less than +/- 90 degrees.

Such analog joysticks were used on the old Radio Shack Color Computer (the 'CoCo'). They're probably available surplus all over the place.

A source of cheap digital encoders can be an old PC mouse (non optical, with the ball). It would be a relatively simple project to convert an old PC mouse into a pitch and roll sensor by linking the encoders to a weighted arm (or arms).

 

[sreid]

Analog Devices has 2 Axis MEMs accelerometers that make good tilt sensors. Perhaps the ADXL310.

http://www.analog.com/en/subCat/0,2879,764%5F800%5F0%5F%5F0%5F,00.html

 

[hoytclagwell]

Hello:

I have worked quite a bit with Fredricks & AOSI vials...they are not cheap (at least the good ones). These need to be driven with an AC signal...even a quick hookup to an ohmmeter will very quickly ruin a precision vial I worked at Spectra-Physics designing construction leveling laser electronics for 10 years.

One low-cost method is to shine a led through a low cost bubble vial & detect the position with a photodiode...this gives a fairly good range, depending on the parts. A quad photodiode can be used with a round "pancake" bubble level to readout in both X&Y.

An off-beat method is to use an old analog meter movement (like an old VU meter or tuning indicator meter guts). A very small weight is hung on the end of the needle. This weight also goes through an opto interrupter. An opamp drives the meter to keep the weight centered in the interrupter. The amount of drive current needed to keep the weight centered in the interrupter is related to the tilt of the assembly. Since this approach is closed loop, the sensor reacts very quickly & does not have a weight swinging randomly around & generating "noise" like the "weight on a pot" approach. These force balance sensors are available commercially for thousands. I built one once for fun for about $2.

 

[mgopalan]

Based on

your price 5$ total (this is always the hard one)
Your range -90 to 90 Degrees
assuming you want Inclination and NOT Pitch and Roll
assuming that you have small quantities ~50 units
Assuming that you need about a degree in resolution
Assuming you want better that 1 degree stability
Assuming you want better that 0.5 degree repeatability
Assuming you want DC or quasi static frequency response

and stating that I know nothing about electrolytic sensors

I would say

Get a SINGLE axis MEMS Accel with a +/-1 g range (or close) and be done with it. They are ridiculously easy to buy, integrate and use. They are also unbelievably rugged, much more so that ANY alectrolytic sensor could possibly be.

I think you have three options out there.

The ADXL103 is probably the best single axis MEMS accel solution out there right now in my opinion. It is about 15$ for the chip. But it is 5 mGee stable over its entire temperature range without calibration or correction which will give you a 0.3 degree stable measurement until the end of time with nothing more than 5VDc and three tiny 0805 ceramic caps. In large quantities >5000 the chip costs 7.50$. which gets close to what you need

The MemsIC MXA2500A is new and is probably ok. It is their "ultra low cost" offering and is about 11$. I have no direct experience with this chip, but prior work says that they are alright.

The ADXL311 is your low cost alternative. It is 8.50$ in low quantities and is about $4 in large volumes. IT IS A DUAL AXIS CHIP. so you could either use in as a dual axis inclinometer (and get only 0-90 Degree range but get pitch and Roll) or just waste an axis and just use it as a single axis accel. The problem with this chip is that the sensitivity is at 167mV/Gee and it has a 300uGee noise floor. It should however get you 0.5 degree or so resolution without any trouble . With about a 1 Hz Bandwidth on this chip, you COULD get better than a 1mGee noise floor if you have sub millivolt stable power supplies, a/D's etc etc. The only problem I see is that the data sheet just plain does not give bias instability numbers. It might move 25 or 50 mGee on you over time and use. That might be acceptable and nothing more than a simple recalibration issue. Usually in my expoerience the bias shifts are more temperature and shock related. If this is going in a benign environment, then you should be fine for a long time.

I would go with the ADXL311. It should give you your 1 degree measurement needed with the added flexibility of giving you pitch and roll and with the hope of maybe even being in your price range.

By the way, this only covers the cost of the sensor itself. The Power, Analog stuff, A/D processor and whatever else is on top of this....

Just as an fun FYI. The last inclinometer I built measured to 0.1 Degree inclination from -90 to 90 degrees. Its did this over 25C to 150C (300F) continuous use in a 20Grms 50-500 Hz vibration environment with 1000Gee shocks While sitting inside a pressure shell rated to 20000psi. It cost 30K to build and calibrate.

MG