A novel way to measure magnetic fields, and DC current without a shunt?

For a project I am doing I need to measure high currents. I was thinking if I could somehow use core saturation to measure the magnetic field, and thus the current in a close conductor. The idea is to have two the same LC oscillators, and then magnetise the ferrite core of one of those, and mix the output, use the difference frequency in a micro, a PIC to be precise. Either that, and the system should no use a shunt, or use one of those Hall sensors. So I did set up a quick experiment, using the small strong magnet I have as a simulated magnetic field from some turn of high current wire. The result is fascinating, the frequency deviation is *much* bigger then I expected!

Here is the test circuit: --------------------- +5V from USB | | | | |---- d | ------------------------>| BF245 | | | |---- s === || ( === 680p | --- N[]S || ( |_____________ | | 47 uF magnet || ( L1 | | | || ( 10mH === 1500p [ ]1k5 | | | | | /// /// /// ///

I measure 13313 Hz without magnet, and 95071 Hz with magnet close against the inductor! So say 13 kHz to 90 kHz, or more then a factor 7 frequency change! The amplitude does *not* change! By using two of these oscillators, and influencing only one, temperature effects get cancelled, a simple dual gate MOSFET mixer... lowpass, and you have a nice signal you can do all sorts of things with in a micro.

I think I do not have to explain this oscillator type here. As to the coils: One day I asked in the shop for 10 uH coils, the guy gave me 10 mH, these... (I think they are 10 mH, could be 4.7 too, have not measured these, do your sqrt LC if you like), I did not want to call the guy a moron, and though these would be fun to play with, as they are :-)

Here is a picture of the circuit with magnet far away: ftp://panteltje.com/pub/dc_current_sensor/osc_without_magnet_img_1790.jpg

The resulting scope picture: ftp://panteltje.com/pub/dc_current_sensor/freq_without_magnet_img_1794.jpg

Here is the picture with magnet against the ferrite cover of the coil: ftp://panteltje.com/pub/dc_current_sensor/osc_with_magnet_img_1797.jpg

The resulting scope picture: ftp://panteltje.com/pub/dc_current_sensor/freq_with_magnet_img_1795.jpg

The scope is connected to the BF245 source. The frequencies were measured at the gate with freq_pic :-) :

formatting link
This is what it put on my screen via RS232: 00095071 95,071 Hz 00013313 13,313 Hz

Any non linearity will not be an issue, as a lookup table is easily implemented in the PIC.

I love it when an idea works:-)

Copyright (c) Jan Panteltje 2010-always All Rights reserved. Nothing of this can be used without written permission from the Author. Certified Usenet prior art. Beware of the watchdog.

Reply to
Jan Panteltje
Loading thread data ...

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

expected!

from USB

inductor!

effects get cancelled,

do all sorts

sqrt LC if you like),

with, as they are :-)

implemented in the PIC.

Nice work, have you ever looked at Magnetic Amplifiers ? (Mag amps)..

Some what of the same idea I guess.

formatting link

Reply to
Jamie

Well, "saturation" means "no more EMF". A transformer in saturation stops transformering, its EMF goes to zero. So, if you apply a constant voltage until it saturates, then turn off (allowing time for the current to drop to zero), then on a secondary, you'll measure a roughly square bump with area corresponding to the difference in magnetic fields when that pulse started to when it stopped, call them Brem and Bsat (rem = remenance, a material property which should be selected to be fairly small). Apply bias, and the V*s drops -- effectively, Brem becomes Brem + B_due-to-DC-bias, thereby changing the area of the bump.

A ferrite core would be best suited for this, since ferrites usually have moderate remenance and fairly sharp saturation. Gapped and low-permability cores should be avoided, because gap is leakage, which is EMF that isn't due to the nonlinear core behavior we're interested in. Sharper 'switching' means a more repeatable saturation current. (Metglas and permalloy cores are particularly easy to saturate, but they have high remenance.)

Now, to measure an external current, you have to magnetize the core in the same direction as the excitation. Since mu is large, it will saturate easily, and if measuring V*s directly, you'd have to calibrate your integrator, which is silly. Instead, you can use feedback to cancel that out, keeping V*s constant and measuring the A*t required to cancel the sense current.

The same fundamental thing can be done with your oscillator, the difference being an F-to-V converter before the error amp.

The real challenge is, your drive winding causes AC induction in everything going through the core, so the DC you're measuring ain't DC anymore. You can't use a mag amp (DC goes through two cores, excitation is inverse to each so they saturate alternately), because when one saturates, the other is inducing, so you get something that looks like signal 1 XOR signal 2. That's not awful, because with the feedback, the difference should be approximately zero. Still, it's not great. Fix it first, then apply NFB. Maybe the EMF could be picked up with a winding, then shunted into a transformer to null the induction. This transformer would need the same current bias winding to keep it out of saturation, which is no problem.

Not bad, a 48:1 inductance ratio. That inductor must have little gap, I bet it saturates pretty easily, maybe a few 10s of mA. (Cheating with a ~1MA/m magnet doesn't count. :-) )

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
Reply to
Tim Williams

DC or AC? If AC then you can also use a current sensing coil. These are relatively simple to interface without the need to calibrate.

field,

core of one of those,

There can be no such thing as 'the same LC oscillators'. Component variation and aging will kill the circuit.

There are some ferrite types that have a more or less linear saturation. You could measure the variance of the inductance. After calibration this should be quite accurate.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
 Click to see the full signature
Reply to
Nico Coesel

field,

core of one of those,

precise.

sensors.

simulated magnetic field from

expected!

USB

inductor!

effects get cancelled,

do all sorts

What kinds of currents were you measuring? (amps?)

Reply to
PeterD

On a sunny day (Sun, 14 Feb 2010 14:49:47 -0600) it happened "Tim Williams" wrote in :

I think this way: I get a delta f of 82000 Hz for a 0-1.2 Tesla at the front of the inductor. So about 68 Hz per milli Tesla. When using 2 oscillators, and taking the difference frequency, the output would be from about 0 Hz (for 0 Tesla) to say 6800 Hz for 100 milli Tesla. That would be 100 % temperature stable. It is in a range that is easily processed (for example by measuring time between zero crossings) by a simple micro. The resolution would be better then 20 micro Tesla.

The oscillator has really low power, I do not expect it to induce any significant RF (well more LF actually) into a high current circuit like I want to use it for. I like it :-)

Reply to
Jan Panteltje

On a sunny day (Sun, 14 Feb 2010 21:05:02 GMT) it happened snipped-for-privacy@puntnl.niks (Nico Coesel) wrote in :

field,

ferrite core of one of those,

Oh yes there are. I have experience with VFOs :-) Adjust with trimmer for zero offset. Aging? What aging? Good receivers are still on frequency (LC) after 50 years. Use good components.

Reply to
Jan Panteltje

On a sunny day (Sun, 14 Feb 2010 16:09:44 -0500) it happened PeterD wrote in :

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

expected!

from USB

inductor!

effects get cancelled,

do all sorts

It is a secret, this only measures magnetic field (1.2 tesla magnet).

Reply to
Jan Panteltje

On a sunny day (Sun, 14 Feb 2010 15:32:43 -0500) it happened Jamie wrote in :

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

expected!

from USB

inductor!

effects get cancelled,

can do all sorts

sqrt LC if you like),

with, as they are :-)

implemented in the PIC.

Yes, in the sixties I worked in a company that used transductors to make super high power regulators, kVA++. Transformers so big that you needed a ladder to climb on top. Also transductors were used for pincushion correction in color TV, when there were still CRTs used, like in the previous century :-) So I am famuiliar with those and associated circuits.

Reply to
Jan Panteltje

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

expected!

from USB

inductor!

effects get cancelled,

do all sorts

sqrt LC if you like),

with, as they are :-)

implemented in the PIC.

While your double oscillator setup will cancel temperature effects in the oscillators, you cannot easily account for the fact that the material saturation is also temperature sensitive. In the case of ferrite, its also nonlinear with current, temperature, mechanical strain on the core, and past history, which makes a simple linearisation almost impossible.

--
Regards,

Adrian Jansen           adrianjansen at internode dot on dot net
 Click to see the full signature
Reply to
Adrian Jansen

Am 14.02.2010 23:16, schrieb Jan Panteltje: ...

A standard AMR magnetometer (Honeywell makes some) for less than 5¤ has a resolution of 7 mGauss. (Turn it on... Ok, the 48µT of Earth's magnetic field give me a reading of ~560) Some of them (HMC5843, YAS529) have a convenient digital interface.

Falk

-- My english may be bad, but you should hear me shout "Achtung, Fritz, die Autobahn".

Reply to
Falk Willberg

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

I think measuring the magnetic field with a coil is not a new idea:

formatting link

There is even a company who sells compass sensors, made with coils and an ASIC for signal conditioning, but I've forgot the name. The resolution is in the range of micro tesla.

--
Frank Buss, fb@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de
Reply to
Frank Buss

On a sunny day (Sun, 14 Feb 2010 23:42:38 +0100) it happened Falk Willberg wrote in :

Yes, possible, but this did not cost me anything as I have all the stuff... But also the resolution depends on how the micro measures frequency. If it measures time between zero crossing then you can measure very long periods... Not that I want to do that, if I wanted to measure something as weak as the direction of the earth magnetic field I could put a webcam on a compass :-) I think the parts cost of this is still below 5 $, including the PIC frequency meter with RS232 output. The 9 pol RS232 connector is more expensive, especially the screened cap :-)

Reply to
Jan Panteltje

On a sunny day (Sun, 14 Feb 2010 23:52:27 +0100) it happened Frank Buss wrote in :

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

No of course not, but in the end I want to measure a high DC current.

Interesting paper, thank you.

Yes, that circuit in tha tpdf could use a PIC :-)

Reply to
Jan Panteltje

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

Yep, The venerable fluxgate compass, common as dirt in most cars, now-a-days.

I designed a fluxgate control chip for Honeywell Sensors Division, Plymouth, Minnesota, in 2007-2008. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

Jan Panteltje a écrit :

magnetic field,

ferrite core of one of those,

precise.

sensors.

a simulated magnetic field from

Get hold of the HP428A/B DC mAmeter manual and read it.

It is essentially the same flux gate, with the measured current going through the core along with a current compensating winding, all that brought to zero net flux through a feedback loop. That one is, by design, free of all the ferrite non linearities and temperature dependencies.

--
Thanks,
Fred.
Reply to
Fred Bartoli

Hardly a secret. Your little magnet looks to be about 1cm across, so the field lines will be ballpark 0.04 m long. N*I = B*l_m / mu_0, or around

38kA.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
Reply to
Tim Williams

The sensors from Honeywell are good, but last time I tried to order it from Digikey, it was export restricted for my country. But was no problem to get it from an official Honeywell distributor in Germany.

I've found the other company, but I don't think that they are using the same concept, because looks like the coils have only 2 connectors:

formatting link

The specification is impressive, e.g. 15 nT resolution.

--
Frank Buss, fb@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de
Reply to
Frank Buss

The one I worked on, the "sensor" itself was a 14-pin object, three axes. As is usual in these cases I was not privy to the contents, other than specifications of the impedances I would see when driving the "straps"... at ±800mA ;-)

Not unusual, I'm often called in to design "black box"... no prior knowledge of the structure at hand... so the client company can't be accused of "reverse engineering" the competition.

My most famous was a hard-drive controller chip done for SSI (now TI), to compete with a National design. I had no clue how hard-drive controller chips worked, let alone what the schematics of the National part looked like.

So my chip design performed BETTER than the National chip :-P ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

field,

ferrite core of one of those,

precise.

sensors.

simulated magnetic field from

expected!

from USB

inductor!

effects get cancelled,

do all sorts

sqrt LC if you like),

with, as they are :-)

implemented in the PIC.

What about no saturation? Use an "E" core, equal windings on each outer leg as a part of a flux feedback oscillator. Take unknown DC current and thread it thru to couple on one side and see the oscillator get less symmetrical as the current increases; one can even determine DC current direction. Refine with a second external DC wire driven by an opamp so that the oscillator remains symmetrical.

Reply to
Robert Baer

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.