Close
0%
0%

A new high accuracy tilt sensor

This project aims to build a tilt sensor that is cheap, very accurate and has a wide measuring range (up to 360 degrees).

Similar projects worth following
The tilt sensor that I built and am developing has a totally new structure: it's basically a three-coil transformer and ferrofluid functioning as the coils' "iron core". It's very accurate (even more than the ones used in industry) despite that basically everyone can build one. It's also cheap, since all you have to build one is copper wire for the coils, and a few drops of ferrofluid. I've got two functioning models right now: one can measure tilt in a 40 degrees and the other in a 360 degrees interval. The 360 degree measuring range is quite good, because there aren't any tilt sensors out there that have as wide measuring range as this, or if there are, as the measuring range is higher, the accuracy is lower.


2016 Hackaday Prize entry video:


There are two functioning models right now:
  • The linear sensor (hereinafter referred to as Linear FCDT) that has the proper geometry (Length/Radius) ratio can measure tilt in a 40 degrees interval and has a relatively good resolution.
  • The circle-like sensor (hereinafter to as Toroid FCDT) that has the proper geometry can measure tilt in a 360 degrees interval.

There is a little demonstration video of the circle-like sensor in working:

(In this video, the sensors' signal conditioners' (Analog Devices AD598) output is measured with a multimeter, and the measured values are sent to a laptop via RS-232 serial port. Then a LabVIEW program calculates the tilt based on the sensors' characteristic, and indicates it.)


The idea

First, let's imagine three coils placed next to each other. The middle coil is called the primary coil, and the two on the sides are called the secondary coils. Let's place an iron core into the coils' air gap, and excite the primary coil with an AC signal (normally with a sine wave). Now, what will happen? Well, it's basically functioning like a transformer, so depending on the iron cores' position different voltages are being induced in the secondary coils. Then, if I move the iron core, the induced voltages in the secondary coils are changing. With the proper signal conditioning circuit, I made a linear displacement sensor! (By the way, these machines are called LVDTs - Linear Variable Differential Transformers, and are widely used in industry.) There's an awesome website where there is a more awesome simulation about the working of an LVDT: http://www.rdpe.com/ex/hiw-lvdt.htm

Cool. Now, you'd probably heard about what ferrofluid is (if don't: https://www.youtube.com/watch?v=5APHa7vscoI). Ferrofluid is a ferromagnetic fluid which contains nanoparticles of metal, that's why it's being attracted by magnets, and basically functioning like "liquid metal".

The basic idea is that what would happen if you'd remove the LVDT's iron core and replace it with a ferrofluid filled glass tube. Somehow like this:

Well, the ferrofluid would act like the replaced iron core, and as you tilt the sensor, the induced voltages in the secondary coils would change depending on the tilt, so you'd make a tilt sensor!


Linear FCDT

And that's it. That's the basic concept behind the sensor. I call this arrangement the Linear FCDT

(Linear - the coils placed "linearly next to each other; FCDT - Ferrofluid Core Differential Transformer)

Our next task is to make experiments with it. First, we'd like to know what is the maximum value of tilt it can measure. Based on experiments I made - as I mentioned earlier - it's a +/- 20 degrees interval. Why? Well, if you reach a specific state of tilt, the ferrofluid runs out from the air gap of one of the secondary coils, like this:

Thus, there won't be generated any voltage in that secondary coil (or the generated voltage would be insignificant), so the sensor's output voltage will deteriorate.

Our next thing to do is to make a sensor which can measure in a larger range, but first, let's measure the accuracy of this arrangement! After hours and hours of making an extremely precise measurement, I came up with this diagram:

(How the precise resolution measurements were made: https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46066-how-the-precise-resolution-measurements-were-made)

Theoretical background of the Linear FCDT: https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46957-theoretical-background-of-the-linear-fcdt


Toroid FCDT

So, if we would like to have a greater measuring range than 40 degrees, we'd have to redesign the whole sensor. The problem with the Linear FCDT was that the ferrofluid ran out from one of it's secondary coils after reaching + or - 20 degrees of tilt. What if we bend a glass tube like a circle, fill it half with ferrofluid, and turn the coils around it toroid-like (hence Toroid FCDT) with the secondaries...

Read more »

LinearFCDT_Moon_MeasuringProgram.vi

The LabVIEW program for measuring the output voltage of a Linear FCDT, and sending it to an email address in every 12th hour.

vi - 38.57 kB - 10/09/2016 at 11:10

Download

SignalConditionerBoardEagle.zip

Eagle CAD files for the first sensor signal conditioner circuit.

Zip Archive - 70.76 kB - 09/24/2016 at 16:41

Download

PresentationalBoxBoardEagle.zip

Eagle CAD files for the presentation box. The board consist of a 230V to +/-15V power supply and a carrier circuit for the AD598 signal conditioner.

Zip Archive - 66.02 kB - 09/24/2016 at 16:41

Download

3D_ToroidFCDT-3.123dx

Autodesk 123D Design files for the Toroid FCDT.

123dx - 1.44 MB - 09/24/2016 at 16:28

Download

3D_ToroidFCDT-2.123dx

Autodesk 123D Design files for the Toroid FCDT.

123dx - 1.72 MB - 09/24/2016 at 16:28

Download

View all 12 files

View all 8 components

  • Our article has been published!

    Aron Molnar12/24/2016 at 21:51 1 comment

    Our article about the project has just been published in the Journal of Magnetism and Magnetic Materials!

    Thank you for your support!

    http://www.sciencedirect.com/science/article/pii/S0304885316322703


  • Detecting the Moon!

    Aron Molnar10/09/2016 at 01:36 1 comment

    If the sensor is truly can reach that high resolution and sensitivity, it's theoretically capable of detecting the gravitational effect of the Moon.


    To test the theorem I made a Linear FCDT especially for this experiment. This sensor should have a big L/R ratio (L: length of one coil, R: radius of the cell) (whys explained in this log: https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46957-theoretical-background-of-the-linear-fcdt), and extremely symmetrically winded secondary coils. I made a sensor that has an L/R ratio of 20 or so: it's 196 mm long, the cell's inner diameter is 6 mm, has 5 layers of 0.25 mm coil wire in each coil, and looks something like this:To get reliable data from the sensor, we have to take it to a very massive and vibration-free place and make measurements with it for at least 3 days in a row. We found the proper place at the University of Pannonia on top of a high-tech CNC machine. To make it convenient, I made a little program in LabVIEW that sends the measured data to my email address at every 12th hour for 3 days (the program's VI can be found in 'Files' section).

    The measurement was started on 7th of Oct at 11:47 a.m. (GMT+2:00) and we're planning to continue it for 1,5 week.


    We got some anomalies in the data, but there is one encouraging fact: we detected a little sine wave signal that got a periode of 24 hours.

    This means that we almost detected the gravitational effect of our Moon!

  • Project is about to be published in a scientific journal!

    Aron Molnar10/08/2016 at 17:10 0 comments

    That's right! We wrote an article about the sensor and we are about to publish it in a scientific article - it's under review!

    Here is a little section of the article:


  • Deduction of the Linear FCDT volume formula

    Aron Molnar10/08/2016 at 13:29 0 comments

    In a previous log (https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46957-theoretical-background-of-the-linear-fcdt) I showed that in case oftilt degrees, the volume of ferrofluid found in the (left) secondary coil is:And in case of tilt degrees, the volume of ferrofluid found in the (left) secondary coil is:In this log I'm going to show you where this formula comes from.


    First, we need to slice the sensor at a point, and calculate the area of the ferrofluid at that point. This area should only depend on values we already know, such as the alfa tilt degree, the L coil length, the R radius, and the x variable. If got the area at that point, we can integrate it along x, and we get the volume of the ferrofluid.

    According to the picture above, the area of the fluid is:

    Next, we have to express theta with respect to R and h, since we only know the value of R (and we will know the value of h after the next step). We can do it using the simple angle-dependent cosine: With this, the formula of the area looks like this: Now there is only one variable we don't know: h. Let's determine it!

    At a chosen x, h equals to:

    And from the picture above, h0 is:Substituting h0 to h's equation:
    And that's it! The area of a slice looks like this:In the formula, the area depend on the R radius, the L coil length, the alfa tilt degree, and the x variable. Our last thing to do is to integrate it along x, and we get the volume of ferrofluid in a coil. But note that in cases of

    we integrate from 0 to L, and in cases ofwe integrate from 0 to


  • Experimental and theoretical curves of the Linear FCDT

    Aron Molnar10/04/2016 at 22:00 0 comments

    Previous post: Theoretical background of the Linear FCDT (https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46957-theoretical-background-of-the-linear-fcdt)

    In the previous post I showed how we can calculate Vsec1, and Vsec2 with respect to the tilt of the sensor. I also pointed to that the measuring range and the sensitivity of the Linear FCDT is strongly depend on the L/R ratio. (Where L and R are shown in the previous post"s first picture.)

    In this post I'm going to compare the theoretical and experimental curves of our Linear FCDT.


    Our Linear FCDT was tested with the AD598 signal conditioner IC, which is used with LVDTs. The coil length of the sensor was L = 18 mm; the radius of the inner cell was R = 3 mm; the amplitude of the excitation voltage was A = 6 V; the frequency was f = 11 kHz; the M cell constant was determined using the already known dimensions of the cell, and the extreme values of Uout:The theoretical curve was determined with the knowledge of the L/R = 6 rate, and the M cell constant:

    The experimental data and the theoretical curve of the Linear FCDT

  • Theoretical background of the Linear FCDT

    Aron Molnar10/04/2016 at 21:54 0 comments

    The Linear FCDT has three coils, which take place next to each other. The primary coil is usually excited with a frequency of 1 to 20 kHz and an excitation voltage range of 1 to 24 V rms.

    The structure of the Linear FCDT

    The cell surrounded by the three coils is exactly half-filled with ferrofluid, so there is a volume of


    ferrofluid in the cell. At any tilt of the sensor, there is always a volume of


    in the primary coil. The spare volume of


    is divided amongst the secondary coils depending on the tilt of the Linear FCDT.

    Let's check out the left secondary coil on the picture above. Starting from α = 0 °, in cases of positive α tilt, the Vsec1 amount of ferrofluid will decrease.

    • In the case of tilt degrees, the liquid level can reach the outer wall, so the amount of ferrofluid in the left coil is:
    • In the case of tilt degrees, the liquid level can’t reach the outer wall, so the amount of ferrofluid in the left coil is:

    Deduction of the formula can be found here: https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/47221-deduction-of-the-linear-fcdt-volume-formula

    If we plot the values of Uout normalized to 1 in cases of disparate L and R rates between -90 ° and +90 ° it is noticable that the measuring range, and the sensitivity inside that, strongly depends on the L/R rate. In cases of low rate we get a relatively great measuring range, in cases of high rate we get a smaller measuring range, but higher sensitivity.

    The effect of L/R rate to the characteristics of the Linear FCDT


    Next log: Experimental and theoretical curves of the Linear FCDT (https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46959-experimental-and-theoretical-curves-of-the-linear-fcdt)


  • Experimental and theoretical curves of the Toroid FCDT

    Aron Molnar10/04/2016 at 21:47 0 comments

    Previous post: Theoretical background of the Toroid FCDT (https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46947-theoretical-background-of-the-toroid-fcdt)

    Now that we know from the previous post that how to calculate Vsec1, and Vsec2, we can move on to calculate Uout of the sensor with respect to the tilt, and compare our results with the experimental data.

    In this post I'm going to compare the theoretical and experimental curves of our Toroid FCDT.


    The output voltage of the toroidal sensor is given in a form of The M cell constant was determined using the known dimensions of the cell, and the extreme values of Uout,

    The theoretical and the experimental curves of the sensor can be seen on the graph below:

  • Theoretical background of the Toroid FCDT

    Aron Molnar10/04/2016 at 19:07 0 comments

    The Toroidal FCDT is consist of three coils just as the Linear FCDT, but the coils were winded toroidal around a glass tube bent in an O-form. Each secondary coil overlays the half of the torus, and they has around 800-800 turns. The primary coil overlays the the whole torus:

    The Toroidal FCDT and the mounting

    The external radius of the Toroidal FCDT is R = 80 mm, the internal radius is r = 3 mm. The inner cell is precisely half-filled with ferrofluid, thus the volume of the fluid in the cell is And the volume of ferrofluid is divided amongst the two secondary coils depending on the tilt of the sensor. It is obvious that when rotating the sensor, the amount of ferrofluid in the secondary coils is changing linearly with the tilt. In cases of α = 0 ° and α = 180 ° there is an equal volume of ferrofluid in the secondary coils. In cases of α = 90 ° and α = 270 ° the ferrofluid is always in just one secondary coil. It can be seen on the diagram below:

    The amount of ferrofluid in the secondary coils depending on the tilt in the Toroid FCDT


    Next post: Experimental and theoretical curves of the Toroid FCDT (https://hackaday.io/project/11225-a-new-high-accuracy-tilt-sensor/log/46956-experimental-and-theoretical-curves-of-the-toroid-fcdt)

  • 3D models

    Aron Molnar09/24/2016 at 16:10 0 comments

    We made some 3D models about the sensors and the resolution measurement test bench just to visualize them better. Here they are:

    Linear FCDT:

    Toroid FCDT:

    Resolution measurement test bench:

  • How the precise resolution measurements were made

    Aron Molnar09/24/2016 at 13:22 0 comments

    I've been saying that the Linear FCDT is capable of measuring tilt with a really high resolution, but I didn't say a word about from where do I know that. I just only said that „After hours and hours of making an extremely precise measurements...”. It’s time to say a few words about those extremely precise measurements.


    The main idea is that get a laser pointer, attach it to the test bench (picture below), adjust the bench a little bit, see how many millimeters the laser point went from the relative horizontal level on the wall (the wall-test bench distance is about 10 meters), write down the output voltage of the signal conditioner IC, compute the tilt, and do it again at least 50 times.

    This is the test bench with the Linear FCDT (sadly, I didn’t take photos with the laser attached to it when we made the measurements):

    The schematic of the measurement looks something like this:

    So after we tilt the sensor a little bit, the position of the laser point on the wall will change significantly, since L is great. Then we measure x, and compute the tilt with this single equation:

    Next, we inspect how the output voltage of the signal conditioning circuit changed, and write down these two number next to each other: ouput of sig conditioner IC - tilt.

    With this method, we can do extremely precise measurements, because L is great, so if we tilt the senor just a little bit, the chage of x will be great enough to measure that conveniently.

View all 11 project logs

  • 1
    Step 1

    In these instructions I'm going to show you how to make the Linear FCDT we used to detect the Moon with:

    This sensor has a great L/R ratio. It means that it has a really high sensitivity, but it can only measure tilt in a small interval.

    (Making a Toroidal FCDT is a little bit more tricky and you'd probably need a glass technician who would bend the glass tube for you.)

    The Linear FCDT consist of three coils (2 secondary, and 1 primary), a cell half-filled with ferrofluid, and a signal conditioner circuit.

  • 2
    Step 2

    First, let's prepare the cell. It will be a 196 mm long glass tube, inner diameter 6 mm, outer diameter 10 mm.

    • Cut the glass tube to size.
    • Coat it with a shrinkable, so the coils won't slip when winding them.
    • Lathe 4 separator disks to separate the coils from each other; place them to equal distances from each other on the shrinkabled glass tube, and strengthen them to the tube with superglue.
  • 3
    Step 3

    It's time to coil! We need 5 layers of 0.25 copper coil for each coil. To make your life easier you can DIY a simple coiling machine like this:

    Or, if you've got access to a lathe, you can use that too. Just strengthen the previously prepared cell to a threaded rod with M6 screw nuts, strengthen the end of the rod to the lathe's chuck, turn on the lathe on small speed, and start winding.

    Note that when winding the secondary coils, you have to wind them symmetrically. In the figure below I illustrated the winding direction of the secondary coils:

    After finished with coiling, it's optional to check the resistance of the coils. The secondary coils' resistances' difference should not exceed 0.5 Ohm! My secondary coils had a resistance of 27 Ohm, and the primary had 28 Ohm.

View all 5 instructions

Enjoy this project?

Share

Discussions

RichardCollins wrote 04/06/2019 at 23:38 point

Aron, 

If you have data from your Moon tracking experiment, I can compare it to the sun moon tidal acceleration signal I use to calibrate superconducting gravimeters (SGs) and three axis broadband seismometers.  The sun moon signal is extremely complex and geometry dependent.  It is nearly impossible to game.  If you get it right, you only need a linear regression on each axis to calibrate your instrument as a sun moon gravimeter. 

I will try to put some software on GitHub to generate the signal at a location.  If you have breaks in the data, noise, earthquakes, it doesn't much affect the calibration.  I have experimented with continuous random calibration that can be added to a small computer connected to the device.  It probably can run on a Raspberry Pi or a clone.

About 95% of the signal at an SG station is the sun moon tidal acceleration.  After you subtract the sun moon signal, that removes most of the earth tides but leaves all the rest of earth-based data, including the seismic data. And its associated gravitational early warning signals.

The SGs are so precise. But only vertical data.  If they were three axis, you can solve for the position and orientation.  An alternative to GPS, based on gravity.  I had to learn to solve for orientation because portable instruments are not always set aligned to the earth's reference frame.  It only works for permanent stations and long term observatories.

I suggest decimating the signal to one minute. That will give you a boost in precision, and it is easier to do a quick round calibration. That is 1440 readings per day, and quite satisfactory for an initial look over a few weeks of data.  I have run one second data from the SG and seismometer stations, so if you have it, that is pretty easy to handle, and is the common standard for many geophysical networks.  For early warning, you have to use arrays of correlated sensors to build up an image of the source. Starting with three axis from the beginning allows each instrument to determine direction.  That might be enough to give single station warning of magnitude and direction.  If gravity is not refracted at high frequencies, then it should be geometric line of sight to the disturbance.  LIGO has some people who are very good with these kinds of things, but their heads are somewhere out there in the universe right now.

What you have done is something I need to study before I can see how it compares to others, and to see how it might be optimized for specific markets and tasks.  I am just getting familiar with HackaDay.IO, but something(s) seem to be missing.

Richard Collins, The Internet Foundation

  Are you sure? yes | no

Lee Cook wrote 03/30/2017 at 17:17 point

Hi,

I came across this:

http://liquiglide.com/

Coating the inside of the sensor tube with something like that could get around some of the problems you mentioned.

  Are you sure? yes | no

Peter Walsh wrote 11/07/2016 at 23:05 point

I E-mailed about your project and that you won to a friend of mine who works for Linear.

Hopefully something might come from it.

  Are you sure? yes | no

Aron Molnar wrote 11/08/2016 at 06:02 point

Cool, thank you!

  Are you sure? yes | no

Lee Cook wrote 11/07/2016 at 12:33 point

Congratulations!!  A very cool idea!

Is it be possible to increase sensitivity (potentially reduce backlash effects) by having a graduated winding arrangement?

  Are you sure? yes | no

louis.haeb wrote 11/06/2016 at 18:47 point

Hi Aron,
congratulations for winning the $10k prize! Your FCDT tilt sensor is really clever and should be useful for a lot of applications. If I can offer some suggestions :


-- For the material of the tube (instead of glass) : if you used a polymer with very low surface energy, such as PTFE or acetal, the ferrofluid bead would not adhere at all to the tube wall, and this would reduce "backlash" of the ferrofluid level.


-- You mentioned that the AD598 chip for signal conditioning is very expensive. Have you considered using the AD698APZ? It's the next chip in the series, has all the same specs as AD598 except better gain drift and offset drift figures, and only costs $34 each on Digikey. In addition, it's a SMT part, so you could make your sensor more compact.

Awesome project, I can't wait to see what you build next!

  Are you sure? yes | no

Aron Molnar wrote 11/07/2016 at 14:34 point

Hi Louis,

Thank you for the awesome suggestions, we'll consider them.

As a matter of fact, I found an AD598 for less than $30 on eBay - however that was a one-time offer.

  Are you sure? yes | no

Peter Walsh wrote 10/10/2016 at 03:04 point

Just so you know, your project doesn't have a contest video.

If you have a video, you need to edit the project, find the section "contest entry videos", and put a link to your video there. Also, the video should be tagged "2016HackadayPrize" on the site where it lives (probably YouTube).

This is a change from last year, you might not have received the E-mail talking about it.

  Are you sure? yes | no

Aron Molnar wrote 10/10/2016 at 12:17 point

Done!:)

  Are you sure? yes | no

Jack Laidlaw wrote 05/10/2016 at 20:06 point

Really cool project! It would be cool to attach this to a really low rpm motor and use it to convert the motor into an stepper motor.

  Are you sure? yes | no

Aron Molnar wrote 05/11/2016 at 11:14 point

Thanks! Hm, that's not a bad idea. 

  Are you sure? yes | no

amiramon wrote 05/09/2016 at 00:34 point

Awesome, and very well explained in few words.

  Are you sure? yes | no

Alan Campbell wrote 05/07/2016 at 19:07 point
Great project!  Doyou have any information on the long term (years) stability of the fluid? Do all the little metal particles settle out and you end up with sludge in your
tube with a fixed offset in your measurement?  If this is a problem, one
could compensate for this when the level was not in use by exercising the fluid
when the gage is idle but one would need to be aware of the settle rate.

  Are you sure? yes | no

Tibor Medvegy wrote 05/08/2016 at 08:33 point

Ferrofluid won’t separate over a long time, the Brownian motion won’t let the nanoparticles to settle.

Ferrofuid is used in many application, if you don’t let it dry out or leave it in a strong permanent magnetic field for a long time, you won’t have any poblem.

  Are you sure? yes | no

Peter Walsh wrote 05/06/2016 at 05:34 point

Hey, check it out! You're top article on the Hackaday blog!

  Are you sure? yes | no

Aron Molnar wrote 05/06/2016 at 11:35 point

Wao, awesome! 

  Are you sure? yes | no

Legrange wrote 05/06/2016 at 14:18 point

Well deserved. This is a really cool project, have a like and a follow.

  Are you sure? yes | no

Aron Molnar wrote 05/08/2016 at 17:34 point

Thank you!

  Are you sure? yes | no

O4karitO wrote 05/05/2016 at 01:00 point

Won't it be easier and cheaper to fill the tube with timy metal balls? You will(please, correct me if i'm wrong) get the same result  and won't have to deal with ferrofluid.

  Are you sure? yes | no

Aron Molnar wrote 05/05/2016 at 17:29 point

Well, doing that would definitely kill the accuracy. Those little metal balls would never be in the same position, so at an exact tilt, the sensor's output would never be the same.

Also, you'd have to worry about abrasion: those tiny balls would hit each other from time to time, and after a while, the whole system would be a mess. 

By the way, ferrofluid actually consist of a huge amount of metal nanoparticles in a liquid, so basically ferrofluid is just a lot of tiny metal balls.

  Are you sure? yes | no

Aron Molnar wrote 05/05/2016 at 17:32 point

And also, you don't need much ferrofluid to build one of the sensors, you'd just need nearly a few drops of it. So for instance the Linear FCDT wouldn't cost more that a few dollars, max 10. 

  Are you sure? yes | no

O4karitO wrote 05/05/2016 at 17:53 point

yeah, I was just thinking about a way to avoid ferrofluid viscosity. 
Isn't the sensor's possible measurement rate quite low with ferrofluid?

  Are you sure? yes | no

Aron Molnar wrote 05/05/2016 at 19:11 point

Yes it is. Actually, the next thing I want to do is to reduce the measuring rate. Right now, it takes about 20-25 secs to fully reach the final output voltage. Basically, it could be reduced by using lower viscosity ferrofluid, but that would also mean less sensitivity because in a less viscous ferrofluid there are less metal nanoparticles. The main problem is that the ferrofluid sticks to the wall of the glass tube, and it takes time for the fluid to get off from the wall. The solution is to surround the glass tube with a compound that doesn't let the fluid stick to it.

  Are you sure? yes | no

Peter Walsh wrote 05/03/2016 at 17:22 point

As a science type I'm always interested in new types of metrology, and the sorts of experiments you can do.

If you can measure 1/10,000 of a degree, a quick back of the envelope calculation shows that the gravitational attraction of the moon will result in +/- 2.1 on that scale. When the moon is on one horizon, the system will read 2.1/10,000 of a degree higher, and when it's on the other horizon it will read 2.1/10,000 of a degree lower. The mass of ferrofluid divides out, so it should be true for any size detector you build.

That's a pretty low number, but easily measurable. If you set up one of your detectors on a concrete floor in an unused corner of your basement, for instance, and could show a sin wave signal that corresponds to the position of the moon over time, that would be a good-proof-of-concept to readers.

And if you average many measurements over time, you might get more resolution that the 1/10,000 of a degree. You're probably already doing this. If averaging gives you much more resolution, you could conceivably measure the gravitational attraction of the sun.

If I have some time I'll do the calculation for magnetic fields, to see how sensitive your system is to that. Might be interesting if you can detect a truck driving by on the highway or something.

Also, it occurs to me that this might make a sensitive accelerometer. A quick scan of the ADXL320 datasheet shows a resolution of 2mg acceleration force (at 60 hz). With the accuracy you are citing, you could probably get *much* more accurate acceleration measurements, and if you could get this accuracy faster than 60 hz it would have lots of applications in inertial guidance systems. Maybe not quad copters per-se (due to the size), but lots of other things might benefit from high-accuracy IGS, such as cars when they go into a tunnel, or tanks when GPS is jammed in a war zone.

Just some thoughts. Looking forward to seeing the project progress.

  Are you sure? yes | no

Aron Molnar wrote 05/04/2016 at 15:47 point

Your ideas are brilliant! I've never tought I can use this device for things like that. Detecting the orbit of the Moon, or even the Sun would be a very interesting experiment and I'll definitely try it out as soon as I can, and post all the results. 

I can't wait to see the results of your calculations about the sensor's sensitivity to magnetic fields. 

You've got so great ideas I can't even respond to them as smart as I'd like to, so I'm just saying a big thank you! 

  Are you sure? yes | no

Peter Walsh wrote 05/06/2016 at 05:43 point

Sensitivity to magnetic fields is turning out to be much more difficult and complicated than I thought. It's related to many things dependent on the actual magnet used, for instance, that I can't directly measure.

Maybe you can grab a rare-earth magnet and just try it? See how far the magnet has to be before you don't see any difference.

Also, fluxgate magnetometers are sensitive enough to measure the Earth's magnetic field (half a gauss, very weak), so your system *probably* won't have advantages over that method.

Your system could be used to measure acceleration. If one of your rings is in a car, for instance, as the car goes around a curve the ferrofluid should be attracted to the outside of the curve.

Accelerometers exist already, but I believe they've got limited resolution. Your system might directly provide 14 bits of resolution, which would be quite a lot for this application.

Just some thoughts. If you take one of these on a car ride, I'd be interested to hear the results.

  Are you sure? yes | no

Aron Molnar wrote 05/06/2016 at 21:11 point

Actually, I tried to model the magnetic field of the sensor using the magnetic field modelling program FEMM (Finite Element Method Magnetics). I didn't succeed, because I couldn't adjust the program to deal with induced alternating voltages. At first, I modeled a simple LVDT with a moving iron core in it. I ran the simulation several times, each time changed the place of the iron core, and made a GIF. The simulation isn't accurate, but it's perfect to illustrate how the magnetic filed would look like around an LVDT:

 

Using the Toroid FCDT as an accelerometer that way is really a good idea.

  Are you sure? yes | no

Peter Walsh wrote 05/03/2016 at 04:28 point

Using a ferrofluid as a an inductive core - opens up a lot of interesting possibilities!

Okay, that's awesome! One "like" to you, good sir!

  Are you sure? yes | no

Aron Molnar wrote 05/03/2016 at 14:47 point

Thank you!

  Are you sure? yes | no

Aron Molnar wrote 04/25/2016 at 07:24 point

Thanks!

Sorry about the video, updated so you can watch now. 

  Are you sure? yes | no

J. M. Hopkins wrote 04/23/2016 at 02:04 point

Any more specifics on construction? I'd be interested in accuracy and bandwidth of the sensor

  Are you sure? yes | no

Aron Molnar wrote 04/23/2016 at 09:57 point

I'll update more information about the project soon.

  Are you sure? yes | no

J. M. Hopkins wrote 04/25/2016 at 03:38 point

Awesome job! The video says private at this time though

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates