BLE Triaxial Earth Field Magnetometer to alert for Aurora Borealis activity and for earth field work
Already have an account? Log in.
To make the experience fit your profile, pick a username and tell us what interests you.
Current State:
Coils (2 designs), Driver, WingKp, Twitter, Aurora Visual Alert: Prototyped
Aurora Visual Alert, WingKp: Complete (need to repurpose and integrate for this project)
Todo: Determine on final coil design and driving method based on thermal and sensitivity testing
Nothing Hidden: This project as completed is fully open source with complete design documentation. Software created for the project is licensed under GPL and everything else under creative commons license CC BY-NC-SA.
Here's a chart recorded on my bench (noisy environment) overnight (9 hours). Chart is for the 28th Aug, time is in UTC. Red Line: Y = Closest Magnetometer I can find (82 miles away), data rate is 1Hz (this is the reference). Blue Line: "Aurora Observatory" is my coil magnetometer, data rate is 1Hz and I've applied at 60 Moving Average just to smooth the data similarly to the reference I'm using.
A few things, Aurora Observatory was roughly aligned to Magnetic East/West. The reference is "Y" so it's aligned to Geographic East, for my location that's about 14 degrees difference. I'm located closer to the Auroral Activity than the reference location, and the auroral activity around that time was present for my location.
Peaks / Valleys consistently match, i.e. it's working well:
Testing metgals cores....
3M Tattle Tape B1 performs best, just one metglas strip, about 1 inch long, dynamic range over +-0.5Gauss (Earths Field) is 199Khz. Difficult to get a hold of, but libraries are sources.
Sensormatic EM Tags perform reasonable well and would be a second choice. 2 strips are needed, dynamic range over +-0.5Gauss is 99.3KHz, with 1 strip, 41KHz, and with 1 strip the same size as B1, 22KHz.
Needed a Helmholtz Coil to test the sensitivity and calibrate the coils. I've designed a 3D printed Helmholtz Coil for this purpose. The best is, it's easy to build, cheap and it does all the math for you and spits out the configuration / sensitivity of the coil using OpenScad:
Here's the first magnetometer plot using the dual wound coil and security tag. Bench mounted in a noisy environment and hooked up via USB. Axis was approximately East/West. X-Axis is "Hours (UTC)", Y-Axis is the frequency delta change of the magnemeter output offset from a chosen point.
The coils are very sensitive, reacting to anything nearby, small magnets, pliers, even my chair and need to investigate if the bumps are temperature related (i.e. Air Conditioning Cycling), or the Aurora (there was activity at the time). This prototype has a Polyester Capacitor, replacing with an X7R.
Comparing Regular fluxgate (1 driving coil and 1 sensing coil = 2) and MagentoInductive (1 coil in oscillation), (using a magnet detection distance from the sensor of around 3 feet), I've come to the conclusion that Magnetoinductive is not as sensitive as fluxgate, which I'm assuming is because the core is not being driven into saturation with MagnetoInductive (it's low current).
Being that the field drops inversely proportionally to the square of the distance, initial testing indicates that MagnetoInductive is approximately 5-10 times less sensitive.
However MagnetoInductive would be fine for a low budget general compass. For the purposes of this project, sensitivity is a requirement.
Now for driving the fluxgate coil. For testing purposes I've been using a transistor oscillator (confusing as MagentoInductive uses an oscillator too), but this transistor oscillator uses both the sense and drive coils. Ultimately I want to migrate away to a microprocessor drive for accuracy reasons, and the output current of a typical microprocessor isn't enough....
To get the cleanest sense signal in a fluxgate, the drive should have it's field reversed on each cycle. I.E. 0V-5V, 5V-0V, 0V-5V, so I need a driver that can switch rails (similar to driving a DC motor forwards and backwards).
So, this evening I've designed and CNC'd (othermill) an H Bridge driver, driven by 1 pin (direction) and is complete with transient suppressors and schmidt triggers (although the transient suppressors I need to order so I'm chancing it for now). All driven by a teensy 3.1 rolling through a frequency spectrum 1KHz - 300KHz for testing purposes.
The PCB off the mill without components is above. The following is the PCB populated:
Testing the drive outputs (unloaded) to check correctness:
Connected to the coil and the Teensy:
Running at 10KHz, notice the flux collapse / recovery in the sense coil is text book:
Schematic for H-Bridge Driver:
3D printed coil formers for 3 test coils now completed. Coils were wrapped fairly haphazardly in a left-right-left-right formation. Windings were stabilized from unwinding with superglue and some hot glue at the ends.
Coil 1: One wire wrapped around small 3D Printed former, approx 35AWG (0.15mm), 2000 wraps, security tag strip cut in half lengthways (1/8 x 1.5 inches) inserted inside former. Resistance: 46.7Ohms
Coil 2: Two wires wrapped together, around small 3D Printer former, approx 35AWG (0.15mm), 500 wraps, security tag strip cut in half lengthways (1/8 x 1.5 inches) inserted inside former. Resistance: 9.6Ohms (per single wrap)
Coil 3: One wire wrapped around large 3D Printer former, approx 35 AWG (0.15mm), N wraps. Core left empty to accept toroids inserted inside. Resistance: 91.8 Ohms
Top to bottom: Coil 3, Coil 1, Coil 2
Showing 3D printed former for the smaller cores, and the finished wound coil showing the placement of the security tag strip (insert fully inside):
Saturday: Further research into the cores for magnetometers has highlighted that the correct core is very important to detecting the earths magnetic field, "metglas" or something similar with high permeability is required.
Um, don't have any, or do I ? Metglas is found in the tags in library books, however I doubt the local library will donate any, but there is something nearly as good in these types of security tags (top rectangular tag):
Remember seeing those sometimes in DVD I buy at Costco. Looking through my collection I've found a black one inside a DVD. The best thing is it's free (excluding the DVD). I won't trash them in future when I find them on stuff.
Removed the top bias strip (attached to the sticky tape) from the tag to obtain the core strip (actually have 2 in there). Removed one and wrapped 500 turns of approx 35 gauge around, not being to fussy about layout (apparently the most important thing is that wrap direction doesn't change when wrapping, otherwise neatness doesn't matter too much).
This is a single wrap, so is tested with the Magentoinductive oscillation technique. Tried 3 different oscillators based on the following oscillation techniques with this coil:
555 Timer: Works well, detects a regular ring magnet about 3 feet away on the scope. Voltage sensitive and temperature sensitive. The 0.005% / C thermal stability of a 555 is too poor for this use, so would need to be temperature compensated. Voltage sensitivity can be solved with double regulation. Also, a little difficult to tune due to the nature of the 555. Otherwise preliminary results suggest it would do the job. Oscillation range can be easily tuned for the best coil saturation. Part count is low and cheap.
OP Amp: Couldn't get to oscillate reliably, tried 2 x ST UA741, and an LM358N. Suspect inductance is the issue and that the coil needs to have more inductance for this to work. Given this, it's likely to be problematic, so abandoning this branch.
Schmidt Trigger Inverter: Same sensitivity as the 555, but oscillation isn't too clean and varies significantly in amplitude and noise which makes it through to the output due to the back emf on the coil despite putting a shunt diode in. Still, would likely be a very simple / cheap solution for a mid range precision compass.
Now the principle is proven to work, it's time to wind better quality coils on 3D printed forms for face off testing for the 2 techniques :
1. Dual parallel wound around a security tag strip (suitable for Fluxgate Technique)
2. Single wound around a security tag strip (2000 turns) - (suitable for Magnetoinductive Technique)
3. A conventional fluxgate design consisting of one of the mystery toroids for the drive and a slip over form for the sense winding, that can be used to test a few different mystery cores.
Saturday: This is my attempt at a 2nd coil. This is a more conventional fluxgate type single core design, consisting of 2 wraps wound together on one of the mystery cores. Put in the oscillator, it is improved but isn't very sensitive. Oscilloscope trace shows a lack of saturation:
Saturday: I've just wound my first coil and built a transistor oscillator and 1 stage clipping amplifier.
The coil is one of my salvaged "mystery coils". The picture below shows the ferrite choke before wrapping (top) and after wrapping. Nowhere near enough winds, but just it's just to test the concept for now. Core is not likely the correct material anyway:
The theory was that it's kind of like a toroid core when conceptualized magnetically, the thinking being that the drive winding is wired through the center of the core.
Here's the breadboarded driver for it:
Tried varying the resistors and although I got oscillation, the results were not stellar. Frequency changes when a large magnet is held very close to the coil, but although it works, it's just not sensitive enough to be useful. Abandoning to more conventional coil designs.
It's the weekend, need to build a sensitive magnetometer in time for Monday...
Here's are the constraints:
1. Can't order anything in for delivery on the weekend (digikey/mouser etc. are out), and our local supply houses are useless, can't get delivery till Tuesday at the earliest (1 day after the deadline)
2. Has to be quick to prototype
3. None of the big guys stock magnetometer sensors that are adequate for the needs anyway, these are specialist sensors and I'd need to wait.
4. Can be breadboarded initially to test, but due to the sensitive nature of the coils, things need to be locked down on a PCB (which can't be ordered in, in time for Monday)
Rapid prototyping MacGyver style, I need to get creative. Found some very old approx 35 gauge (0.15mm) magnet wire and some larger mystery magnet wire in the wire box. I must have purchased it 30 years ago when I first started electronics and was wrapping AM Radio ferrite rods.
I have an othermill, so at least I can prototype PCB's fast.
I've been desoldering everything with any type of ring core or ferrite core in from my stash of kept PCB's from old equipment. Old Motherboards, switched mode PSUs and router boards proved fruitful.
They are all mystery cores, tried investigating online to see if the color indicated the core material, but it doesn't. Some of the coils had numbers on, but most were made in China and old, so none resulted in a data sheet :-( Here's the resulting stash:
ChasinSpin
Ted Yapo
peter jansen
BastelBaus
Pinomelean
Can you provide details of disassembling tattle tape strips?
When I disassembled the strip, I found that there was one about 7" long strip and 4/5 small 7 to 8 mm long strips stuck to this 7" strip.
Kindly guide.