suggestions for high-resolution tiltmeter (inclinometer) sensor?

[rhb]

What's the sensitivity of the WW II vial?

A similar setup using a 4"/2mm vial mounted to a concrete pier might even detect the moon deforming the earths crust.

Alfred Loomis, the financier who retired to Tuxedo Park to develop radar in the late 20's,  bought a couple of the best pendulum clocks made.  They were several thousand dollars each.  Later he was involved with developing ovenized crystal oscillators. Comparisons soon showed the effect of the moon on the pendulum clocks.

[JBeale]

Still uncalibrated, but progress. I made a smaller diode bridge with two SOT23 packages each being a diode series pair, and they are low-capacitance parts with <1.0 pF at Vf = 0: On Semi MMBD352LT1G.  Used short wires to the electrodes. Seemed safer to solder the wire first, and only then apply the copper tape to the glass vial, which worked but was tricky to handle.

Placed the sensor on my concrete garage slab since the floor of the house was not stable enough. I'm convinced I see a clear tilt signal when standing to one side or the other of the sensor as my weight locally tilts the big concrete slab, probably by a few arcseconds.  On top of that, a continuing background drift, and a low level of random noise.

[RoGeorge]

Very impressive results, I would've never believe such sensitivity is possible! 

What is the cause for the steady increase in voltage reading (in the last chart), and why the tilt variations are all almost rectangular? If you move slowly from left to right, will you see a triangle shaped variation instead of the square shaped one?

Is there any hysteresis in the sensitivity of the tilt detector?

[JBeale]

It is neat to see the supposedly solid concrete floor bend (slightly) under my weight. I only just got this new version of the tiltmeter working today. I may be able to say more about its properties after running it a few days. The steady background trend could be a real physical tilt (anything from differential heating, hydration of soil, etc.) or could be some electronic artifact. I have it sitting on a smooth "surface plate" which is in turn resting on the concrete slab, so I can reverse it 180 degrees and see if the trend also reverses. To do that cleanly I should have the vial precisely level with the housing, which it is currently not. I will try reversing it after acquiring some more data.

I'm only sampling once every 2 seconds, so the rectangular shape of the offset when I stood still nearby, is because I moved into and out of position on either side of the sensor quickly enough that it appears nearly instantaneous on the plot.  If I had instead moved very slowly past it, I would have expected to see a smooth curve, instead of the rectangular pulse.

[StillTrying]

When you say you stood to the left and right of the sensor, is that the sides or the ends of the sensor, I think it should show very little change at the sides and max at the ends.

When you've run it for a few days, I think there's a good chance you'll see the 12hr 25min and 24hr 50min Moon/tides cycle.

[rhb]

That thing will be a hell of a chore to calibrate. Level is easy, that can be done electronically by reversal. So long as you stand in the same place.

The only way I can think of would be to use a 10 ft piece of heavy 2" square tubing w/ 3/8" walls.  put the sensor in the center and use a surface plate and gauge blocks in 0.0001" increments at one end.  And that would only get you 0.17 arc seconds.  Using B grade blocks the uncertainty would be about 0.085 arc seconds.  And that completely ignores air currents and temperature differentials in the tube or heating of the gauge blocks.

It's truly amazing.

[JBeale]

Yes, calibration would be- interesting. I meant I stood near the left end and right end of the capsule. I was trying to make a detectable signal, so I stood where any deflection due to my weight would have the largest response.  As you suggest, in the orthogonal direction there is very little sensitivity.

The data covering 11 hours last night looks very strange. Starts out with a nearly linear drift but then gets more complicated. The plot shows two "sudden" step-changes during the night (taking place over a few minutes, and most of the shift within 30 seconds). Looks very odd and nothing obvious happened inside or outside at those times. I also have a seismometer in the garage and that didn't see anything either. Now I'm starting to want another sensor, to see if this is some kind of sensor artifact, or something real.

[RoGeorge]

The step-changes might be some discharges. Somewhere in the whole apparatus, I suspect there is a capacitor acting like an integrator. That constant increase in voltage shows there is an integrator somewhere.

Capacitors with small charges can be suddenly discharged by energetic particle, like e.g. cosmic rays. Not sure if this is the cause of discharges. It might be as well caused by the measuring instrument doing an auto-zero calibration.

Anyway, you got yourself a hell of an intriguingly sensitive instrument! 

I want one, too, just for the fun of playing with it. 

[StillTrying]

"The plot shows two "sudden" step-changes during the night (taking place over a few minutes, and most of the shift within 30 seconds)."

Different expansions with temperature changes correcting themselves ? You've got glass glued to metal.

[rhb]

The data covering 11 hours last night looks very strange. Starts out with a nearly linear drift but then gets more complicated. The plot shows two "sudden" step-changes during the night (taking place over a few minutes, and most of the shift within 30 seconds). Looks very odd and nothing obvious happened inside or outside at those times. I also have a seismometer in the garage and that didn't see anything either. Now I'm starting to want another sensor, to see if this is some kind of sensor artifact, or something real.

 

You're embarrassing me.  I'm a geologist (igneous petrology) who then spent his career in geophysics and I don't have a tiltmeter or a seismograph :-(

But I *did* get four polarizing microscopes which I am refurbishing.  Unfortunately, almost 40 years later I have only a very rudimentary notion of how to use one despite having spent 9 months using one all day, every day.

I studied the calibration problem a bit.  The only viable method looks to me to be a Michelson interferometer, a 100:1 lever and a micrometer head with a 10:1 belt drive.  And even then you might need to multiply the movement by repeated bounces in an optical flat to get adequate resolution from the interferometer.

I live a little over a mile from a lake, so I should be able to detect the rise and fall of the water in the lake with one of these.

I knew that these could be very sensitive, but I'm still stunned at how simple it is.  Naturally I shall need at least two mounted orthogonally on bedrock.

I'd like to see details about your seismograph.  I read all the articles in C. L.  Stong's "The Amateur Scientist" when I was a kid.  And even though I am not an earthquake guy, I'm quite familiar with the design problems.

[JBeale]

I did build one of the swinging-door type Lehman horizontal seismographs a while back, but the current one I simply purchased. It is vertical-only, uses a 5 Hz geophone which was extended with some kind of active electronics to have useful sensitivity down to below 0.5 Hz.  It's the basic model from https://raspberryshake.org/products/raspberry-shake-1d/  It plugs into your home network and the "Raspberry Shake" company archives your seismic data 24/7 for access from anywhere, which is quite convenient.

At the moment (12:40 pm PDT) the tiltmeter drift rate has been picking up the past few hours, with maybe with two more step-changes in the opposite direction as the ones last night. I definitely need a second sensor to tell if these are just an instrument artifact, but I'll have to wait until the new capsules come in from aliexpress to build another one.  I'd give it about an even chance to be caused by something wierd about my (quickly constructed) sensor circuit. 

If you zoom in to samples 25441 - 26122 (0.8 to 0.9 mV), you can detect a tiny step up and then down, which exactly matches the timing of me leaving and then returning in a car that was parked on the driveway outside.

[rhb]

You should consider a MEMS sensor for the seismograph.  The orthogonality of those is not very good, but you've got a sensor that will easily allow calibrating out any errors.

I spent quite a lot of time working on this subject 5-10 years ago. for multicomponent seismic recording.   Never did anything with it, but I did work out how to correct for the sensor errors with a very simple setup that could be built from a scrap hard drive (for the sake of the class 9 bearing) and an optical prism.

[branadic]

This is capacitive sensing right? Have you thought about using a CDC (capacitance-to-digital converter)?
Have you measured capacitance? What size is it?
What if you paint the electrodes directly on the vial using some silver ink? Your copper tape include some adhesive, that can cause your drift issue due to relaxing effects or humidity, beside any mechanical relaxing by mounting the vial.

-branadic-

[JBeale]

All I can say with the gear I have handy is that the total capacitance change is less than 1 pf.  A CTD converter might be good to try. I don't know if those also measure a difference of two external capacitors eg. Cx - Cy, or ratio Cx / Cy, or an absolute reading? The real signal is the differential capacitance, which the simple diode bridge measures directly.

Yes, I am also wondering about that sticky adhesive-backed copper tape and possible drift. If I could make solid electrical contact to silver paint that might be good. Or even very thin plain metal foil which is superglued on and/or lacquered over, might be more mechanically stable.

[Conrad Hoffman]

Might be useful to download a copy of the General Radio 1615-A capacitance bridge and the 1605-A Impedance Comparator. Both use a "simple" tapped transformer to make differential measurements and achieve great sensitivity and stability. You don't need 80% of the circuitry, just a transformer wound on a toroid and a few other things. Probably both on the IET site or BAMA.

My surface grinder is on a cement floor. I use an accelerometer to balance the wheels. There's an obvious change in amplitudes as I walk around near the machine. That baffled me at first, because how could that possibly change balance, then I realized the the coupling to the 4 leveling feet was changing, so the vibrational amplitudes through the frame changed. Cement floors, unless unusually thick, are quite flexible on the scales we're talking about.

Somewhere I've got a book about building DIY weather instruments and I think they covered seismographs too. No matter what you're building, a copy of Building Scientific Apparatus by Moore, Davis & Coplan should be an essential item on the shelf. It covers little of what we're talking about specifically, yet is applicable beyond measure.

[JBeale]

For reference, here is the Type 1615-A Capacitance Bridge, from p.60 of https://www.ietlabs.com/pdf/Manuals/1620_im.pdf
I guess that's one way to do it, but my existing circuit is also simple (it doesn't use a transformer, anyway). I will need to investigate what the factors affecting the stability are.  Without measuring anything, I suspect I'm resolving differences of femto-farads just with the simple diode bridge.

[rhb]

No matter what you're building, a copy of Building Scientific Apparatus by Moore, Davis & Coplan should be an essential item on the shelf. It covers little of what we're talking about specifically, yet is applicable beyond measure.

I've got  three editions and a duplicate of one.  But I'm shocked to learn there is a 4th and I don't have it.  Worse yet, it's been available for 9 years.  So time to buy another book.

[branadic]

All I can say with the gear I have handy is that the total capacitance change is less than 1 pf.  A CTD converter might be good to try. I don't know if those also measure a difference of two external capacitors eg. Cx - Cy, or ratio Cx / Cy, or an absolute reading? The real signal is the differential capacitance, which the simple diode bridge measures directly.

Without measuring anything, I suspect I'm resolving differences of femto-farads just with the simple diode bridge.

Take a look at AD7746 (+/-8pF) and with additional guard AD7747 (+/-4pF) both with 24bit resolution, you get down to aF. You can try PCap01AD made by ACAM even though somewhat more complicated to get it run but with aF resolution as well. There are other CDC's available but I wouldn't recommend them.
All three can measure differential capacitance. PCap01 measures with the third electrode either grounded or floating while AD products do have an excitation signal on the third electrode. You can choose what you prefer.

-branadic-

[RoGeorge]

TI has a few capacitance to digital converters, too. They are cheaper than their AD counterparts.

http://www.ti.com/sensing-products/capacitive-sensing/capacitance-to-digital-converters/products.html

[branadic]

Cheaper yes, but not easy to handle and you need a lot of external components. The only interesting part from them is maybe FDC1004. Easy to go is AD774x, but more performance gives PCap01. PCap02 and PCap04 are not of interest here, because you want low noise and high resolution.
If I had a vial I could paint electrodes with silver ink on it, add either PCap01 or AD7746 board to it and reproduce the results. Any recommendation where to get an equal glas vial?

-branadic-

Edit: Found your vial used is still available at ebay, I ordered a few and hope to reproduce your results soon.

[JBeale]

Thanks for finding the vial on ebay. Sure enough, that looks like the exact part. It is 1.5" long and it comes in that box. I wonder what the actual sensitivity of the "Type L 1 A31314" vial is.

[metrologist]

I spent 5 mins on google and could not find the spec.

Also, I wonder if there is a fancy way you could wind magnet wire around it. Perhaps have half coils on top that loop around the bottom half coils so you can reverse direction?

[branadic]

I think it is some custom specific precision vial. Further more it is not clear to me if it's a Top Ground Vial, an Ordnance Vial or a Reversion Vial. However, doesn't make any difference at the moment. Hopefully it doesn't take to long for them to arrive, to get a first impression of what resolution can be reached. I have a Leica Nivel 230 and a Wyler Zerotronic to compare with on our test bench.

-branadic-

[JBeale]

Based on the photos I believe it is an "Ordinance Vial".  I'm very interested to hear what your findings are, given you have some good measurement tools to compare with.

Knowing I would be reassembling it later, I just used hot glue to temporarily hold it against the bottom of the diecast box housing. I wanted the glue to hold the glass flat against the metal, but I have read that cheap hot glues are 50% wax by volume, with a horrendous (!) thermal expansion coefficient. Now that I know the circuit works, I should try for a more kinematic mount, maybe spring wire holding the vial against a v-groove pillow block at each end.

[metrologist]

I'm trying to invent some kind of interesting mounting for my vials too... plan to sink this into the roughed-out base here, but need an adjusting mechanism.