DIY Self Oscillating Fluxgate

[Altra]

Over the years there has been a lot of interest in the Speake magnetometer
sensors and how to diy the circuits. I know some of the other guys here have made working units but never posted much info. The mu metal cores have been a obsticle to building these since it is hard to find in small quanities. But actually mu metal is all around us. I discovered a while back that anti-theft tags have a strip of Metglas 2705M mu metal. It has an amazing permeability of 1,000,000u. These are the plastic rectangles you find attached to higher priced items.If you cut one open you will find two sheets of thin metal with a mylar seperator.One of the metal strips is some kind of soft iron and the other is mu metal.The metglas is very flexable and shines on one side. I suspect at the check out isle, the clerk magnetized/demagnetized(?) the soft iron sheet so that the mu metal vibrated at a different frequency when radiated with emf at the front door.
For ease of experimenting you can use a pre-wound coil from a reed relay.
The best one I found was a Tyco OMR 5v 250ohm spst. I found that using higher resistance coils (>250ohm) limits the dynamic range. The most sensitive configuration is in a L/R oscillator. Will also work in a LC configuration, but not optimal.
You remove the glass reed switch which leaves you with a nice hollow core. Cut a thin strip and place inside the core. See attached pdf for circuit.

Precision Navigations, Inc patent US4,851,775

OMR 5 volt reed relay _ No longer in production

Radio Shack still has them
Compact 5VDC/1A SPST Reed Relay
Model: 275-232 | Catalog #: 275-232

Ebay US
search "5V SPST Reed Relay" for $3.99, from Circuit Specialist

Ebay UK
search "5V SPST Reed Relay" for 2.66 pounds

[Davor]

You beat me to it
Kudos!

[Altra]

Davor,

Could you look at the flip flop circuit (fig 10) in my pdf. I could never get it to oscillate.
Tried different resistor values without result. Maybe you could simulate it.

Thanks

[Davor]

OK, I'll see about it. Point with self oscillating fluxgates is that metglas saturates, and inductance changes rapidly past the saturation limit, hence the spikes in oscillograms. As saturation is offseted by external field, e.g. Earth field, so does saturation happen sooner or later, thus changing the frequency.

I did not simulate much saturable cores yet. It will be interesting.

[Davor]

I don't have the issue of N&V with that circuit, and you did not indicate which transistors were used. However, I made it oscillate. But I'm not very happy with it's performance.

I assumed the coil inductance to be ~20mH (without core). Guess that would be about right for a reed relay driver at 5V.

Oscillation can not start spontaneously. It needs a kick to start. In simulation I fixed that by checking the "Skip initial operating point solution" but that does not mean it would work well in real life.
Most probably the original circuit used some deflection transistors with some serious Cceo. I used BC337 initially, but I made it oscillate only after connecting some 100pF in parallel with collector and emitter. Such oscillation produces high voltage spikes. There is a model of 2N3019 which comes close to deflection transistors, and I put a 33V zener in parallel so it works happily as is.

I needed to replace 1k in Q2 collector with 47k.

It is a bad circuit anyway. Too many things to go bad, and it can fail oscillating if there is no kick at the beginning to jump-start oscillation. It is very assimmetrical though, so perhaps it would provide good earth field to frequency conversion linearity.

[Altra]

Thanks for giving it a go. It's mostly academic, since LM393s and cmos gates are cheap.

The circuit came from the Q&A section of the magazine. Someone asked for a simple
magnetometer circuit and the editor posted this circuit with little details. Maybe missing a capacitor?

Probably from a 1001 circuit book.

[Davor]

Could be. There are many nowadays forgotten circuits that made a whole era tick. I wonder how many people still know what deflection oscillator does.

Anyway, here is a circuit that works happily on it's own ... for academic purposes (the presented voltage is across the coil, and 250ohm is coil resistance). I did not check any of the emitter coupled circuits, but as you said, comparators are cheap.

[Altra]

A dead TV, the first place you checked was the horizontal deflection oscillator or driver.
R.I.P. Deflection Oscillator. But even with all the mpu's and digital processing there is still a place for
these little gems.

I'll bread board your circuit and let you know

[Davor]

While you are at it, could you please check the inductance of the reed relay coil with and without the metglas insert?! I'm still at loss how to calculate the nasty bits of saturation criteria for this setup, and most feasible approach seem to be empirical one, but in order to make a model I simply need some basic data.

[Altra]

The coil inductance

Air core 13.30mH/250ohm - Same on both meters

With Metglas 94mH/250ohm - B&K 875B

With Metglas 55mH/250ohm - Extech 380193 test freq 120hz

With Metglas 77mH/250ohm - Extech 380193 test freq 1khz


The meters must use different test frequencies. I think the the Extech LCR meter uses 1khz
If it's important to your model I can scope them.

[Altra]

Your circuit works. I used 2N2222 for the npn's they have about the same beta as the BC-337's.
The scope shot looked very close to your sim.

It oscillates at 48khz - air core.

With the coil in a north-south orientation and metglas core

9.4khz, with coil phase reversed 19.3khz. You get the same result if you rotate the coil 180 degree n-s

Not bad, almost 100% frequency change ~ 55hz/Deg

[Davor]

Happy to hear that

Theory is fine, but only as a base of practical implementation.

BTW, the difference of results with coil testers could be due to different signal level. Metglas is easily saturated, so more juice would result in less effective inductance.

[Davor]

I'm just wondering, there is an experiment where a magnet is let fall through a copper pipe, and as a result of eddy currents breaking it's fall you have to wait for it for quite some time. I know mounting sensors such as FGM into aluminium tube is recommended, and I guess it attenuates every fast changing field and provides some level of shielding.

How about using endfeed copper fittings with end stops on both ends and have a sensor shielded capsule-like in such contraption?

[ApBerg]

I'm just wondering, there is an experiment where a magnet is let fall through a copper pipe, and as a result of eddy currents breaking it's fall you have to wait for it for quite some time. I know mounting sensors such as FGM into aluminium tube is recommended, and I guess it attenuates every fast changing field and provides some level of shielding.

How about using endfeed copper fittings with end stops on both ends and have a sensor shielded capsule-like in such contraption?

Think the same.. alu-tube can give problems ?! I have made a smaller pcb.smd version, fits in a polyester-walking stick, still needs some testing... can take some time

Best regards.

Ap

[Davor]

My reasoning is that copper comes in it's pure form and annealed, hence with best possible conductivity. On the other hand, aluminium comes either as cold worked (hardened) or as an alloy, and in both cases it's conductivity is not at it's best. On the plus, you'll find end stops (caps) for copper pipes that are quite friendly to work with.