Author Topic: suggestions for high-resolution tiltmeter (inclinometer) sensor?  (Read 51382 times)

0 Members and 1 Guest are viewing this topic.

Offline branadic

  • Super Contributor
  • ***
  • Posts: 2378
  • Country: de
  • Sounds like noise
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #25 on: May 14, 2018, 07:08:35 pm »
Please read the specifications carefully:

Angular range:    1600 seconds (±600 seconds calibrated) *
Best Accuracy:    0.2 seconds
System Accuracy: ±1 seconds over central 100 seconds, ±6 seconds over ±600 seconds

Where is best accuracy achived, at zero point?
System accuracy is ±1" (±4.85µm/m or ±0.00028°) over ±50" (±242,41µm/m or ±0,01389°) that means a five times worse accuracy near zero point and ±6" (±29.09µm/m or ±0,00167°) over ±600" (2908,87µm/m or 0.16667°). That is not as amazing as it seems at first glance. If you compare that with a Wyler Zerotronic ±0.5° the Wyler is the winner.

-branadic-
Computers exist to solve problems that we wouldn't have without them. AI exists to answer questions, we wouldn't ask without it.
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #26 on: May 14, 2018, 08:10:01 pm »
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.
Mine is a Talyvel 3 or 4, I forget the model.  My sensing unit is slightly bigger overall.  The pendulum might be about 3 cm long, and hangs about 2-3 cm from the support.

Jon
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #27 on: May 14, 2018, 08:11:19 pm »
Hmm.  Might be a good project for a class 9 bearing from an old hard drive.
Nope, the bearings all have some sticktion.  All of these sensitive inclinometers have no bearing or other friction.

Jon
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #28 on: May 14, 2018, 08:20:33 pm »
Please read the specifications carefully:

Angular range:    1600 seconds (±600 seconds calibrated) *
Best Accuracy:    0.2 seconds
System Accuracy: ±1 seconds over central 100 seconds, ±6 seconds over ±600 seconds

Where is best accuracy achived, at zero point?
System accuracy is ±1" (±4.85µm/m or ±0.00028°) over ±50" (±242,41µm/m or ±0,01389°) that means a five times worse accuracy near zero point and ±6" (±29.09µm/m or ±0,00167°) over ±600" (2908,87µm/m or 0.16667°). That is not as amazing as it seems at first glance. If you compare that with a Wyler Zerotronic ±0.5° the Wyler is the winner.

-branadic-
You have to know the conditions these specs were made at.  Is this to retain this level of calibration over the full temperature range?  What about time?  Is it supposed to keep the zero calibration for a year, until sent back to the factory for re-certification?  I think these conditions may apply.

I can only speak for how well my unit works, which is phenomenal.  If I want to do a real zero to the earth, I set up my surface plate and zero the plate in by flipping the sensor back and forth, until the reading does not change in 4 90 degree positions.
Then, it will stay in cal for the whole day.  I know that by placing the sensor on my 3500 Lb lathe and walking to the opposite end of the bed, the level shifts by one arc second.  That is my body weight shifting the concrete floor slab.

Jon
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #29 on: May 14, 2018, 09:48:52 pm »
Hmm.  Might be a good project for a class 9 bearing from an old hard drive.
Nope, the bearings all have some sticktion.  All of these sensitive inclinometers have no bearing or other friction.

Jon

I'd certainly agree that friction is a significant issue with the HD bearings, probably a major problem. But if there is a pendulum, there is a bearing.  So long as one is careful, a razor blade on a flat surface will work.  It's just fragile.  Of course, these are not things expected to survive being dropped on the floor.

I'm clearly going to need to play with a pair of RF oscillators tuned by capacitor plates on a pendulum feeding a mixer and frequency counter.
 

Offline JBealeTopic starter

  • Frequent Contributor
  • **
  • Posts: 298
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #30 on: May 14, 2018, 10:39:01 pm »
Making a geometrically stable and low-friction pivot may be non-trivial, if you want very good performance. The knife-edge idea is simple, but I gather not the best available solution. Here is a brief survey from the related field of horizontal seismographs:
http://www.myeclectic.info/SeismoPivots/seismopivots.htm
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #31 on: May 14, 2018, 11:14:24 pm »
Interesting link, thanks.

A seismograph is different in that the mass being supported is much larger.  As noted in the link, chemical balances use polished agate equilateral triangles.

I'm quite fascinated by the resolution of the Talyvels.

The crossed rods suspension seems particularly easy to implement using ground drill blanks.  If I were interested in geophysical tiltmeters, I'd probably still go with the bent tubing,  But I have a lathe and mill and have application for something like the Talyvel.  My machinist's level is actually good enough, but I'm a bit of a general metrology nut.  A sub-second of arc electronic level would be a cool project.

Please post followups of what you do.  I was hopelessly corrupted by Claire Stong's collection of "Amateur Scientist" columns which came out in 1962. "For less than the average price of a set of golf clubs, you can smash atoms in your backyard"
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #32 on: May 15, 2018, 03:15:40 am »
But if there is a pendulum, there is a bearing.  So long as one is careful, a razor blade on a flat surface will work.  It's just fragile.
That's why the Talyvel uses 5 hair-thin wires to suspend the pendulum, no friction there.  And, i have dropped it about 18" from a lathe bed into the chip pan, and it seemed to suffer no harm - to my great relief!  This is a $4000 instrument, although I was lucky enough to get it for less than 20% of that on eBay, due to a really bad description.  But, I knew what I was looking at, so went with the photos, not the words.

Jon
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #33 on: May 15, 2018, 03:22:00 am »
  But I have a lathe and mill and have application for something like the Talyvel.
Yes, I have a machine shop at home, too.  And, I got a fairly BIG lathe about 16 years ago:
http://pico-systems.com/sheldon.html

And the bed needed to be resurfaced.  I started out with a straight edge, but but realized a 25" straight edge cannot properly test a 72" lathe bed.
I built a mini-carriage that I could mount dial indicators or the Talyvel on and check slope at 3" positions along the bed, and then correct the hills
and valleys.  Then, I could use the straightedge to test for smaller scale variations.  It all worked out REALLY well, the lathe is amazing to use.

I occasionally drag out the Talyvel to check things, but don't use it often.

Jon
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #34 on: May 15, 2018, 05:10:21 am »
Nice machine.  I've just got a Clausing 10 x 20 with only 54 feeds.  But I like how you reground yours.  Mine could stand some attention.
 

Offline branadic

  • Super Contributor
  • ***
  • Posts: 2378
  • Country: de
  • Sounds like noise
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #35 on: May 17, 2018, 04:38:14 pm »
Did a very quick test today, some 34R 0603 resistor as a heater on a flexible circuit board, 1k NTC 0402 on both sides of the resistors. No optimization and no special package to prevent air currents. The NTCs were measured with some digitizing prototype board. It worked pretty well. Tilting the board between +/-90° showed big changes in differential resistance. So chances are that by optimizing and special packaging such a discret thermal sensor will deliver quite good results.

-branadic-
Computers exist to solve problems that we wouldn't have without them. AI exists to answer questions, we wouldn't ask without it.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #36 on: May 17, 2018, 06:34:50 pm »

Slide the thinnest feeler gauge you have under one end and tell us what the change is.
 

Offline jmelson

  • Super Contributor
  • ***
  • Posts: 2758
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #37 on: May 17, 2018, 06:46:13 pm »

Slide the thinnest feeler gauge you have under one end and tell us what the change is.
Slide a single HAIR under one end and see if you can see the reading change.  If not, it is in the carpenter's level class.
A master precision level will go off scale with one hair under the end, over a 10" length.

Jon
 

Offline metrologist

  • Super Contributor
  • ***
  • Posts: 2199
  • Country: 00
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #38 on: May 17, 2018, 07:46:57 pm »
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.   

How are you doing that if your $50 level is graduated in 500 millionths over 10"? I really do not like the looks of the Shars level, but I may buy the 8". The Starretts are too expensive for me...

Also, when I wanted to measure my surface plate, I tried a front surface mirror on the plate and used a laser about 10m away to reflect the beam back onto grid paper for ~1asec/1mm resolution.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #39 on: May 17, 2018, 08:06:44 pm »
You set up the surface plate so it's level by reversing the level.  You then setup a pair of stacks of gauge blocks say 1.00000" and 1.000050". 

The statement was in reference to fitting the level vial for capacitive sensing and calibrating it.  You'd need a large set of grade AA blocks rather than the B blocks I have to setup the example above.  My blocks have an uncertainty of 50 millionths.  I don't recall the smallest incremental difference I can create. But I can make measurements at least to 0.0005" increments and linearly interpolate.

 I have a partially completed interferometer project for checking my gauge blocks I hope to be able to work on again soon.  I've got the HeNe laser, splitters and mirrors (from old CD & DVD drives).  All I need is time and precision rails.

I suggested the short Shar's level because the OP wanted to measure geophysical tilt.  So all he needs is the vial. An 8" or 10" is much more satisfactory for setting  up machine tools.  The larger the body of the level, the more work making the body flat and hence the higher the cost.
 

Offline metrologist

  • Super Contributor
  • ***
  • Posts: 2199
  • Country: 00
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #40 on: May 17, 2018, 09:40:54 pm »
We are just about the same, I suspect. I think my blocks are 50 millionths too, and I have another nicer Polish set with a somewhat recent cal sticker (I don't run a commercial shop). I picked up a AA Rohn 24x48x6 surface plate that I would like to check/verify the characterization and cal, it looks too good...
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #41 on: May 17, 2018, 10:20:18 pm »
If you figure out a way to verify surface plates to a tenth without a Talyvel or an autocollimater please post to the metrology group.  I've wrestled with how to do this for a long time. The frustrating part is Edmund sold WW II surplus autocollimaters for almost nothing in the 60's. 

It seems to me that some copper foil strips in the appropriate locations on a vial combined with a trimmer capacitor would produce a very sensitive indicator.  After all IIRC the 0.0005/10"  is 1/8" from the center point, at least on my level.

I don't have any qualms about the Chinese levels.  You might need to scrape them, but I learned to do that 20 years ago.  As long as I don't have to do it too much it's rather pleasant.

I very much want a 24" x 36 AA plate.  I haven't bought one because I want a zero ledge which is a little hard to find cheap and shipping is as much as the plate.  And at 450 lbs, a real headache to move.  Rather than ledges, I'd prefer to drill holes in the sides and epoxy threaded inserts to hold things  in place.

John Meshna sold  core frames from the MIT test machine for $6.  Unfortunately, a lot of money if you're 12 and it was over 40 years before I understood the historical significance of the core frames and where they came from.  But I do have a Singer card with core memory and an uncut IC wafer as well as an early IBM logic card bought from Meshna for the transistors.  I'd pulled them, but miraculously found them.  So I plan to replace them and try to cleanup the mess I made 50 years ago.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #42 on: May 20, 2018, 05:01:52 pm »
I was just looking in a metrology text which showed an older (analog meter) version of the Talyvel.  The pendulum unit appeared to be no more than 3-4" high.  It was stated in the text that it uses an LVDT sensor and was sensitive to 1 arc second and that it takes about 1 second to settle.

A piece of 0.010" music wire and watch jewels would seem to be a pretty robust and sensitive pivot.  Buying suitable jewels is not cheap ($24 each in the instance I found).  However, it might be possible to salvage them from a thrift shop mechanical watch.  Music wire is available as fine a 0.003". Guitar strings are more readily available, but only go down to 0.008" that I have seen. 

LVDTs are pricey, but their construction is  quite simple.  It's 3 coils with a movable core. The center coil has a sine wave applied and the measurement is made by comparing the amplitude of the signals in the other coils. So making one should not be very difficult.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #43 on: May 24, 2018, 02:25:01 pm »
The Talyvel really piqued my interest so this morning I considered every possibility for sensing a pendulum to one arc second.

I've come up with an interesting possibility.  A pair of photodiode reflective object detectors with a pair of first surface mirror on a pendulum.

Take a look at the current vs distance plot on page 5 of this:

https://www.fairchildsemi.com/datasheets/QR/QRE1113.pdf

The response is nearly linear from 0 to 0.5 mm and they are $1.01  quantity 1.  With an MCU with a 12 bit ADC, that gives a 0.125 micron resolution.  The tangent of one arc second is 5e-6.  So with a 2.5 cm pendulum you'd have  one second of arc resolution.  A 10 cm pendulum would give 1/4 arc second resolution which should suffice for an arc second level. 

For the pendulum support 0.0005 or 0.001 feeler gauge stock should work fine.  Because the level will be reversed for calibration, any residual bend will be cancelled out.  Starrett 0.005" stock is $15/ft on Amazon, but 0.001" stock is $2.80/ft from Victor Machinery Exchange.  With suitable stops on pendulum movement it should also be quite robust.

One can scavenge suitable mirrors from old CD & DVD drives.

 

Offline JBealeTopic starter

  • Frequent Contributor
  • **
  • Posts: 298
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #44 on: May 24, 2018, 09:21:08 pm »
That might be the simplest way, although if you're going to do optical displacement sensing, I think you can do better by using a collimated beam (eg laser pointer) reflecting from a mirror on your pendulum onto a split photodiode (position sensor).  That's how they measured the deflection of AFM tips (atomic force microscope) down to the nanometer level. 

Although it is angle-sensing rather than distance-sensing, so for a rigid pendulum you could make it as short as you want, what matters is the optical lever arm between the mirror and the detector.
« Last Edit: May 24, 2018, 09:34:32 pm by JBeale »
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #45 on: May 24, 2018, 11:17:29 pm »
I was looking for a  photodiode array for doing angular sensing when I found the device I referenced.  Did you look at the datasheet? With two of those 0.25 mm from a target the performance should be there.  The mirrors are somewhat gilding the lily as just painting the pendulum flat white would give  almost the same result.

There's a whole lot more to an AFM than just the sensor.
 

Offline JBealeTopic starter

  • Frequent Contributor
  • **
  • Posts: 298
simple circuit + bubble level resolves 0.05 arc-sec (?)
« Reply #46 on: May 25, 2018, 02:28:01 pm »
This appears relatively simple- a traditional glass spirit bubble level, with three foil electrodes stuck on the outside, and a simple circuit to detect differential capacitance change as the bubble moves. The alcohol fluid has a dielectric constant around 80 and the air bubble near 1, so the capacitance change signal is good enough to resolve 0.05 arc seconds, so they claim.  Published as ARC-11344 by Michael G. Dix et.al., Ames Research Center in Mtn.View back in 1981.

"A Simple Tiltmeter" in NASA Tech Briefs Fall/Winter 1981, p. 319
Single page description: https://www.docdroid.net/enKb2ed/nasaames-simpletiltmeter.pdf
Original PDF incl. 150+ other pages: https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20100027513.pdf

It seems almost too good to be true. If this was reliable, I don't know why anyone would use the more expensive electrolytic sensors, which do not claim much better resolution.  The high-sensitivity bubble levels are themselves not super cheap, although it is apparently possible to make them yourself: http://fer3.com/arc/m2.aspx/Making-your-own-spirit-level-vialswithout-attached-pictures-Morrow-jan-2012-g17728
photos illustrating above: http://fer3.com/arc/m2.aspx/Making-your-own-spirit-level-vialsanother-attempt-send-Morrow-jan-2012-g17731

Or for $50 + shipping you can buy such a high-sensitivity bubble capsule:
https://www.leveldevelopments.com/products/vials/ground-vials/5955101-ground-vial-15-x-96mm-sensitivity-0-01mmm/
« Last Edit: May 25, 2018, 04:08:57 pm by JBeale »
 
The following users thanked this post: oilburner, RoGeorge

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1928
  • Country: us
    • The Messy Basement
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #47 on: May 25, 2018, 04:21:51 pm »
This is all really interesting stuff- thanks everybody for the links!

Some time ago I had to build a micro torque meter. It was basically a pendulum off a shaft. As the shaft was rotated and the pendulum raised to the 90 degree position, the torque was easily calculated. I used a Gurley optical encoder good to about 0.001 degree. The trick was the bearing- I used a Newway 1/4" air bushing. Zero friction and perfect location. https://www.newwayairbearings.com/catalog/air-bushings/
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14072
  • Country: de
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #48 on: May 25, 2018, 04:33:00 pm »
Those bubble levels can be sensitive to defects or contamination on the inner surface. So the bubble can see some "friction" / sticking. How much depends on the quality.
Besides getting the level sensitive, one also needs a way to calibrate / find the zero position.  Depending on the stability this might need a way to repeat the zero adjust just before use - one might want to do that anyway if the surface to measure is not that smooth / flat. So measuring both ways around would be just a version of averaging over a few positions.
 

Offline rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: suggestions for high-resolution tiltmeter (inclinometer) sensor?
« Reply #49 on: May 25, 2018, 05:50:23 pm »
Those bubble levels can be sensitive to defects or contamination on the inner surface. So the bubble can see some "friction" / sticking. How much depends on the quality.
Besides getting the level sensitive, one also needs a way to calibrate / find the zero position.  Depending on the stability this might need a way to repeat the zero adjust just before use - one might want to do that anyway if the surface to measure is not that smooth / flat. So measuring both ways around would be just a version of averaging over a few positions.

Clearly you've never used a 10 second level.  If the level is not true from use to use, there is a serious problem with warping of the frame.  Time to rescrape it true.

A level is the easiest instrument to calibrate, bar none.  The center is the midpoint between 180 degree reversals.  Proper vials are ground internally rather than bent so that cross slope doesn't effect them.

The biggest issue in use is thermal expansion.  That's why there are insulated pads spaced away from the body of the level so that it will not warp from handling.

For an unattended instrument such as the OP wants, capacitance sensing is excellent.  I'm not sure about how well that works in a shop environment as I've not tried it.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf