suggestions for high-resolution tiltmeter (inclinometer) sensor?

[JBeale]

I'm interested in this thing: wikipedia, https://en.wikipedia.org/wiki/Tiltmeter  You can use a simple pendulum, or a water-filled U-tube, but it seems the better ones work like a bubble level (spirit level) in a carpenter's tool. They use some electrolytic fluid instead of alcohol, and internal electrodes to read out the bubble position accurately. I'm interested in resolving for example one arc second (1/3600 degree). 

The concept is simple enough, but I guess it's a small market. So far it looks like any connectorized tiltmeter module with better than 0.001 degree resolution costs over $1k just for the module.  Anyone know if you could buy a bare electrolytic bubble capsule itself somewhere, if I wanted to roll my own? 

"Spectron SH50058-A-003" http://www.spectronsensors.com/datasheets/SDS-116-3106.pdf is an example of the kind of thing I have in mind, but I believe that particular one is not intended or priced for the individual experimenter.

[StillTrying]

Are you trying to detect the movement of the Moon.

If you made a pendulum 1m long, you'd have to detect a movement of the bob of 0.0048 mm.

http://www.seika.de/english/html_e/FAQ_e.htm

[Conrad Hoffman]

This sounds similar to the recent thread looking for a cheap arcsecond rotary encoder. An arcsecond is difficult in a small space. The Starrett 199 Master Level is on this order, but no electrical connections. I wouldn't trust those types of connections anyway and would probably try something like photo sensors to detect the bubble. Precision levels usually have the curvature ground into the ID of the tube. No idea how they do it. You can buy the tubes, but not cheap for the good ones. Looking at vial makers, none seem to list better than arcminutes. Glass is pretty flexible, and you don't need much curve, so I'd try supporting the ends of a moderately long tube and just warp it slightly by pressing up on the middle.

If you want to try something different and have some space, there's a trick using two mirrors. You drill a hole in one and put a contrasting ring around it. View another mirror through the hole from a distance away, and you get the infinite reflection typical of such things. It's extremely sensitive, though I'm not sure how to instrument it and not sure if it gets you to arcseconds. You can also build a primitive interferometer with two mirrors having less than 100% reflectivity. Shine a laser through them. A lens can get you rings rather than a spot, reducing the sensitivity to whatever you might need.

[BrianHG]

In the earlier thread, the cheapest solution was 2 load cells vertically tied together, one 90 degree rotated, with a weight suspended on the bottom cell and the top one tied to your vertical frame of reference.  The 2 load cells were tied to precision ADCs and with a little math, you got the tile along the x and y axis as the load cells reported bend off axis.  Note that due to the positioning of once axis load cell detector being above the other, you need to mathematically correct for it's position.

Obviously, if you only need to measure bend/tilt along 1 axis, life becomes easier.

[JBeale]

Thanks for the responses guys, these are interesting ideas which I had not thought of before. I have some experience with load cells and 24-bit ADCs, under load I know they are sensitive to any vibration (acting as a seismometer).  The cheap ones will also creep.
The optical mirror idea is interesting. Could use a split-photocell as a spot-position detector.  I had a setup like that before; I think my resolution was limited by light scattering from random dust passing through the air (it was a good dust / air quality monitor!) but the idea does work.

The neat thing about the electrolytic bubble sensor with electronic readout by AC bridge is it is so compact, mechanically simple, and less-sensitive to air drafts, etc. than a big pendulum, but still has quite good resolution. Just the sensor itself is a bit specialized and pricey.  Maybe I should get ambitious and try to make one.  There is a low-resolution tiltmeter for 2 axes that is a simple vial half-full of electrolyte and five wires sticking up inside, one central wire and four corners, those get you maybe 0.01 degree. The high-res ones are drawn as slightly-curved glass tubes with inside electrodes that may be plated on; I don't know how they do that.

[ez24]

I bought a special bubble level years ago and it did not work for me.   I threw it away so I cannot get the name.  I think it was 1 degree max.  I called the company and the reason it did not work was the "friction" needed to get the fluid to move (or ball inside the fluid).  It was not cheap.  A bubble level is not the way to get high resolution.

I got it to measure the tilt of a building.  Electronic levels are much better.

[BrianHG]

Thanks for the responses guys, these are interesting ideas which I had not thought of before. I have some experience with load cells and 24-bit ADCs, under load I know they are sensitive to any vibration (acting as a seismometer).

Vibration due to to the mass of your pendulum is a fixed frequency like a clock pendulum and you can completely eliminate it by software (ie tuned notch filter).  As for other seismic noise, to make it operate with the same response as any other direction of gravity sensor, it is nothing more than a low pass filter.  Basically averaging many samples.  Though the device isn't tiny as optically looking at a bubble, it will still be really small as you would use load cells similar to that in scales which measure fractions of a milligram and your pendulum mass would be about the weight 10-50mg.

Use only 5v or less on the excitation voltage to minimize temp build up and drift in the resistive elements.

https://www.aliexpress.com/item/0-100g-200g-300g-750g-Electronic-Scale-Aluminum-Alloy-Weighing-Sensor-Load-Cell-Weight-Discount-50/32670225988.html?spm=2114.search0104.3.229.d9834197VZ03X8&ws_ab_test=searchweb0_0,searchweb201602_3_10152_10151_10065_10344_10068_10130_5722815_10342_10547_10343_10340_5722915_10548_10341_5722615_10696_10084_10083_10618_10139_10307_5722715_5711215_10059_308_100031_10103_10624_10623_10622_5711315_5722515_10621_10620,searchweb201603_25,ppcSwitch_5&algo_expid=bc25c045-4366-4798-b9fa-00bca22a66a4-37&algo_pvid=bc25c045-4366-4798-b9fa-00bca22a66a4&transAbTest=ae803_1&priceBeautifyAB=0

https://www.aliexpress.com/item/100g-small-range-high-precision-weighing-sensor-load-cell-weight-sensor/32776835270.html?spm=2114.search0104.3.250.d9834197VZ03X8&ws_ab_test=searchweb0_0,searchweb201602_3_10152_10151_10065_10344_10068_10130_5722815_10342_10547_10343_10340_5722915_10548_10341_5722615_10696_10084_10083_10618_10139_10307_5722715_5711215_10059_308_100031_10103_10624_10623_10622_5711315_5722515_10621_10620,searchweb201603_25,ppcSwitch_5&algo_expid=bc25c045-4366-4798-b9fa-00bca22a66a4-40&algo_pvid=bc25c045-4366-4798-b9fa-00bca22a66a4&transAbTest=ae803_1&priceBeautifyAB=0

[RoGeorge]

This one is pretty clever and claims to measure three hundred and sixty degrees of tilt to an accuracy of one ten-thousandth (0.0001) of a degree.
https://hackaday.com/2016/05/05/clever-and-elegant-tilt-sensors-from-ferrofluid/

[Marco]

Nice. But if you can get that resolution with inductance, why couldn't you do it with the capacitive sensors? Which are simpler to construct.

So how would you calibrate something like that? Try to turn it smoothly with some intense gear reduction at 1 turn per day?

[tecman]

I designed some angle sensors a few years ago.  I can tell you that you will have problems.  Electrolytic sensors do not have the accuracy you need, and suffer from temperature effects as well.  They also share a problem with pendulum type sensors as well, which is damping. Bandwidth vs damping is a big issue that can plague you.  I ended up with a mems accelerometer which was good to about 1.5 minutes.  For one second you will need something exotic, likely optical.  A sensor based on an interferrometer would be a good bet, but will be relatively big and expensive.aul

[JBeale]

Bandwidth is less of an issue because I'm measuring geophysical creep, for example a 10 minute time constant is OK.

So the claim of <0.0001 arc degree resolution and <0.0008 arc degree null repeatability for SH50055-A-009 on http://www.spectronsensors.com/datasheets/SDS-116-3106.pdf is just a lie?   Wouldn't be the first one on a datasheet, and it's a small company but it seems they've been around for ~80 years. There are other companies that claim arcsecond resolution also.

[branadic]

Just a few hints on something in the league of 0.00006° or 1µm/m:

capacitive pendulum
https://www.wylerag.com/en/products/inclination-measuring-sensors/digital-inclination-sensors/zerotronic-inclination-sensor/#specifications

optical fluid detection
http://metrology.leica-geosystems.com/en/NIVEL230_1680.htm

capacitive level
https://www.hahn-schickard.de/en/services/sensors/inclination/electronic-spirit-level/

-branadic-

[JBeale]

The capacitive sensor looks interesting. A bubble in dielectric fluid changing relative capacitance on a normal PCB with some pads forming an electrode pattern, seems like a pretty simple thing to try out.

[Marco]

To me the Seika one seems easier to build than the bubble approach, to get precise movement of the bubble you need an extremely precise curve for the bubble to move along.

[JBeale]

Precise curves may be needed if you need accuracy over a large angular range. I am looking for very high resolution tiltmeter, but I do not need it linear or accurate over a large range. Ultimately I only care about very small deviations, much less than 1 degree from a nominal setpoint that would be an initial differential null point.  I think it may be good enough to simply have some concavity of large but unknown radius, and then calibrate the sensor externally. As the saying goes, everything is linear to first order :-)

[branadic]

The capacitive sensor looks interesting. A bubble in dielectric fluid changing relative capacitance on a normal PCB with some pads forming an electrode pattern, seems like a pretty simple thing to try out.

Well, it's not as easy as it seems, the dome needs precise machining and some surface finishing. The pcb design was a challenge too and the sensor mounting wasn't trivial. A well understood model to calibrate the sensor is necessary as well as the equipment to calibrate it

To me the Seika one seems easier to build than the bubble approach, to get precise movement of the bubble you need an extremely precise curve for the bubble to move along.

The Seika sensor is at least a bubble approach too and only a copy of the 1-axis sensor Hahn-Schickard developed and transfered to:

https://www.2e-mechatronic.de/en/category/sensor-technology/

Unless you can't make this type of sensor very big you will never get the required resolution. By the way, the dielectric fluid is one of the main secrets inside those sensors and the amount of fluid needs to be very precise.

-branadic-

[ez24]

Bandwidth is less of an issue because I'm measuring geophysical creep, for example a 10 minute time constant is OK.

So the claim of <0.0001 arc degree resolution and <0.0008 arc degree null repeatability for SH50055-A-009 on http://www.spectronsensors.com/datasheets/SDS-116-3106.pdf is just a lie?   Wouldn't be the first one on a datasheet, and it's a small company but it seems they've been around for ~80 years. There are other companies that claim arcsecond resolution also.

70 not 80  same as me  ha ha

I also want to measure geophysical creep so I called them but they were closed (3pm on Friday) but Kevin answered and told me the SSY0271 costs $130  (just a random pick because they said this was "low cost").  He was not sure if this would be right and the engineer would be back Monday

My creep is I built a 10x12 building on a land fill on the side of a hill.  The building is strong enough that it could roll down the hill without damage (I sure hope not).  Over the years I jacked it up and put fill under it and leveled it.  It seems to stop moving the last 5 years (which is good because I got old in the last 5 years).  But I would like a tilt recording so I am interested in what you are doing.  Can you tell more about your creep?  Do you want to record the data?  If so what format?  My guess is CVS (spreadsheet) format is best so it could be plotted.

Also how would you power it?  ie  AC or batteries.

Monday I will talk to the company.  I would trust their specs but you can talk to them about them.  Seems a good find.   

[Marco]

The Seika sensor is at least a bubble approach too

I was talking about the NG4i from the first link to Seika.

[branadic]

I was talking about the NG4i from the first link to Seika.

I know, you detect the covering of a dielectric fluid with the electrodes so half of the content is air, thus a bubble.

-branadic-

[rhb]

For an accurate low cost tiltmeter I suggest a Chinese machinists level bubble vial.  The machinist's levels are good to 10 seconds when read visually.  So optical sensing the bubble position with a photodiode array  should get you to a fraction of an arc second. 

http://www.shars.com/4-master-precisions-level-1

The vials for machinist's levels are ground internally to a very large radius curve.  The price is dependent upon how long the body is, so a short one saves money for your application.

However, as you are more concerned with change than accuracy, you could simply bend precision glass tube to make your own vials. The downside of that is that tilt in the cross direction leads to an error.  So you would need to mount two on orthogonal axes.  Use a slab of scrap stone kitchen counter top.  Drill holes through the stone for clamps, lay the tubes flat on the stone and clamp the ends loosely.  Slip a piece of of the thinnest shim stock you can find under the center of the tubes and tighten the clamps.  The tube will deform to a uniform curve of large radius.  You'll need precision shim stock of two different thicknesses which are as close to each other.  Ideally you want 0.xxx0" and o.xxx5" material.

Connect the shim stock as one terminal in a capacitive bridge with two pieces of copper foil for the other terminal of the capacitors.  Use something like a 5-10 MHz xtal clock chip as the signal for the capacitive bridge and attach a JFET input instrumentation amplifier to the copper foil pads.  Adjust the position of the slab so that the amp outputs are close to zero.  You'll need to hermetically seal the whole thing with a bag of silica gel and preferably a shot of dry nitrogen gas before closing it up.  Or pull a hard vacuum on it.

You can, of course, also use optical sensing for homemade vials.

Have fun and let us know more about the project.  I'm a geophysicist.  I've never done any tiltmeter work, but am well aware of it.  There are companies that can measure submillimeter deformations over miles using satellites or aircraft and laser or radar  interferometry.  It's pretty mind boggling stuff.

BTW another approach  would be to use a differential pressure sensor with a liquid filled tube on each port.  That's likely to be fragile though.

[branadic]

A standard 200mm machine level delivers 20µm/m (4,13sec or 0,00115°). Optical detection is one possible solution. But forget about the word low cost, as long as you have no possibility to calibrate the sensor on an already existing Acutronic or similar gear.
There is information on Wyler website on how their Zerotronic works, maybe you can copy that capacitive readout pendulum system or simply spend 3k€ on it.

What you could try is a thermal inclination sensor. Use a flex pcb, place a heater (resistor) in the middle and two NTC next to the heater (one on its left and right side). Put this setup in an evacuated package, this way you create a simple 1-axis thermal inclination sensor. For sure not as effective like a similar MEMS device, but worth a try on a low cost scale.

Another approach could be the mentioned vial together with a CCD from a flatbed scanner plus some microcontroller... many ways to go.

-branadic-

[rhb]

The specs on high end Starretts are 0.0005"/10" or 0.0001"/10" which correspond respectively to 10 arc seconds and 2 arc seconds.  The Chinese ones such as I linked are good to 0.0005"/10" visually.

Verifying and calibrating a level with gauge blocks on a surface plate is not rocket science.  I can do 50 millionths over 8" with my $50 level and set of blocks.  That's about 1.3 arc seconds.   For $300-400 you can do 1-2 millionths over the length of the level.   It will take a long time to do though, as you'll have to wait for the blocks and level  to cool off each time you change the stack.  And you have to have very stable room temperature.  You'd probably have to leave the room if you wanted to 0.02 arc second accuracy (1e-6"/10").  Otherwise radiated body heat would disturb things.

In the case of a tiltmeter, one is rarely, if ever, interested in the actual angle.  That's really pretty meaningless as you do not know how long a distance it applies to unless you're using long baseline fluid level tiltmeters such as shown in the wikipedia article.  Those have severe issues with thermal gradients.  In most cases all one is interested in is the fact that the earth has deformed. 

All in all, electronically sensing a bent piece of glass tubing is probably the most sensitive option for a low cost device.  The geophysical application is not like other angular measurements.

However, an NXP MPXV5004DP  0 - 3.92 kPa differential sensor is cheap.  I'll leave as an exercise what change in water level is full scale for that sensor.  But with a 24 bit strain gauge ADC from eBay it will be staggeringly sensitive and quite precise.  Albeit, accuracy will be very tedious to obtain.

[jmelson]

I have a Taylor-Hobson Talyvel, a very sensitive electronic level, used for machine alignment.  When in factory calibration, it is supposed to be accurate to 0.2 arc seconds.  I certainly believe mine to 1 arc second.

It has a small aluminum pendulum suspended by 5 TINY wires.  It has 2 steel discs on the ends of the pendulum that are very close to a pair of inductive proximity sensors.  I THINK these sensors are just solenoid coils, but don't know for sure.  There is also a pair of magnets that span an aluminum part on the pendulum, acting as an eddy current brake.  You can pick the sensing head up and place it on a surface, and it will settle to full accuracy in less than 3 seconds.  (A master precision level usually takes at least 30 seconds for the bubble to stop moving.)

Just for anyone who is not aware how small an arc second is, this level will be completely off-scale with a human hair under one end.

Jon

[JBeale]

If the Talyvel can resolve 0.2 arc-seconds (1 micro-radian) that is impressive. I haven't seen a dimensioned drawing, but from the photos of this online http://www.spectrum-metrology.co.uk/electro-optical-metrology/clinometers.php the pendulum is not larger than 10 cm, meaning the end of the pendulum is resolved to better than 100 nm.  Capacitive sensors can do that; apparently magnetic ones can as well.

[rhb]

Hmm.  Might be a good project for a class 9 bearing from an old hard drive.