Flux gate question

Discussion in 'Electronic Design' started by George Herold, Feb 8, 2013.

1. George HeroldGuest

Any fluxgate experts? I recall Jan posted some thing a while back.

A colleague has made a flux gate by sticking to inductors side by side
(with the axes in the same direction), but wired so the fields are in
opposite directions. There’s then a pickup coil that is wrapped
around the pair (all coils with a common axis).
I’m trying to understand the ‘scope trace.
First off the inductors are Bourns 5800-392-RC, L= 3.9mH, R= 8 ohms
Ioperate=0.1A, Isat=0.2amp The L/R ‘frequency’ is about 300Hz. I’m
driving them at 200 Hz and monitoring the drive voltage. So there is
a bit of a phase shift between the displayed voltage and the current
(which is what I care about.)
I’ll show a current waveform below. I’ve got a max current of ~0.6A
about 3x the 10% saturation current.

Here’s a ‘scope shot with B=0
http://bayimg.com/caJbbAAeg

So I interpret this trace as follows:
The two inductors have different saturation currents one inductor
saturates first, and the pickup coil then sees an increasing B field.
As the second inductor saturates there is no longer any change in the
B field. And I’m left with a spike. As the current decreases the
same thing happens, but in reverse. And I get another spike with the
opposite sign.
My question is why do these two pulses have different widths/
amplitudes. The one for increasing current is always sharper and
higher. (I tried another pair of inductors same type of behavior.)
I wondered if it might be the opamp driving the inductors (OPA544)?

I also measured the current (0.5 ohms in series)
Here’s the signal with a sine wave
http://bayimg.com/MaJiEAAEG
There a bit of weirdness near the pulses.

I also did it with a triangle wave drive here,
http://bayimg.com/mAjigAaEG

For completeness here are pictures with a B field applied in one
direction
http://bayimg.com/CAJBKaAEG
and then reversed
http://bayimg.com/eajbeaaeG

Is there a ‘physics’ reason for the asymmetry? or just electronics?
(I was thinking I could try an RC Zobel network in parallel with the
coil?)

Thanks
George H.

2. George HeroldGuest

Interesting... I'll have to think about it.
I'm working on the synchronous 2f detection.

Our 'baby' lockin has no 2f function. I'm thinking of a bridge
rectifier and a little transformer.

Does anyone else have this problem? I'd be happy to start using some
other image hosting site. I liked this one 'cause it's easy for me.
No registration or other mumbo jumbo.. just upload an image and copy
the URL.

George H.

3. JoergGuest

Nope. Blazingly fast here, loads whambam style as if they were on my
harddrive.

Here Jim is always razzing me about my computation machine and now _his_
machine is behaving like a slow-poke, tsk, tsk, tsk ...

4. Spehro PefhanyGuest

Flickr is about a 35msec ping, less than 1.5 seconds to load.

How long does it take from your hard drive? ;-)

5. JoergGuest

Must be something wrong on your side. Or at your ISP but that's not very
likely.

No idea. Normally doesn't matter because you wouldn't even feel 200msec
latency if the server was at the South Pole, provided that the data
streams fast enough. I get the same speedy reaction from image servers
in Europe, when participating in NGs over there. Well, except those that
are over-fluffified, script-laden or when my firewall stomps on the
brakes because of nasty stuff in the data.

6. JoergGuest

Way under a second load time here. Maybe the electrons are freezing a
bit on the way to Canada. Ping is 191msec, so must be electrically
farther away from me.

Flickr is 67msec ping.

No idea but first I have to find it in the directories and that takes
time

7. JoergGuest

Nah, ain't happening here. As I said, something must be wrong with the
computation gear on your side ... <snicker, snicker>

8. George HeroldGuest

Enough chit chat about bayimg. :^)
Does an inductor behave differently going into saturation versus
coming out of it? Maybe I could just wind a pickup coil around one
inductor and see how it behaves.... I stuck an RC zobel across the
inductors R=16 ohms C = 20uF (C = L/R^2)
(OK should have been ~30uF) This did absolutely nothing to the pickup
signal.

I guess the difference doesn't really affect the signal, I'd just like
to understand it.

George H.

9. JeroenGuest

When the coils go into saturation, the voltage across them is quite high
and so is dI/dt. When they come out of saturation, the voltage is much
closer near zero, and dI/dt is consequently much smaller. At a guess,
I'd think that the symmetry should be better at 1kHz. Of course, you'd
need to increase the drive voltage by five times too.

Jeroen Belleman

10. George HeroldGuest

Oh, you mean because of the inductance!
(I've been trying to draw B/H curves with different curvature going
back and forth, but no hysterisis... didn't work to well)

At low frequency (~20Hz) they do become more similar.

* At a guess,
* I'd think that the symmetry should be better at 1kHz. Of course,
you'd
* need to increase the drive voltage by five times too.

Hey there's a new 140V opamp... but not much drive current.

Thanks Jeroen

11. George HeroldGuest

Thanks Jim, I don't think you need hysteresis to explain it.
Jereon nailed it for me (I think.)
It's just the inductance... going into saturation it's pushing me
along,
and coming out it's sorta holding me back... That snaps up the one
side and stretches out the other. (excuse my anthropomorphizing)

George H.

12. JoergGuest

I had only three programs that crash. They got voted off the island and
replaced with similar software of adequate quality. Problem solved. Now
there is only one (Skype) which I have to keep in order to communicate
with one engineer. No other crashes whatsoever. It's simple: There is
software of good quality and then there is software of lesser quality.

13. George HeroldGuest

I was talking with someone who said the difference between German and
US companies, is that US companies will ship stuff with software that
sometimes doesn't work right.

George H.

imgur.com

Tim

15. JeroenGuest

It's actually the DC resistance of the coils being near the
average inductive reactance that causes this asymmetry.

The spikes on the winding common to the two coils occur
when the current sweeps through some small interval on
either side of zero, in which one of the coils is deeper
in saturation than the other. (The region where the B-H
curve changes slope.)

At very low frequency, the coil resistance dominates. The
curve traced out on an I-U plane is a straight line, crossing
the critical current at equal speed in either direction. The
spikes are symmetric w.r.t. to peak voltage.

At high frequency, the inductive reactance dominates, the
curve on the I-U plane is a (distorted) circle and we're still
crossing the critical current at equal speed in either
direction, but now the spikes are symmetric around zero
voltage.

In the intermediate frequency case, the critical current
crossing is unequal in the two directions and one of the
spikes sits at a lower voltage than the other, taking more
time to cross the critical current, and yielding a lower,
longer spike.

So far for intuition. Next is mathematics or simulation...

Jeroen Belleman

16. George HeroldGuest

Grin when he said it I thought of Eagle.
It may have issues, but it doesn't crash.
With all previous pcb software my motto was 'save early and often'.
(mind you I still run version 4.15 from years ago.)

George H.

17. JoergGuest

Same here, version 4.16r2, ten years and counting. Anything after that
wasn't really worth the upgrade for me because they didn't include the
most predominant missing feature, a hierarchy. But I rather forego a
hierachy than tolerate crashes.

Lately there was some talk about a link with LTSpice. No word from their
engineers about it so I'll wait. If that ever works really seamlessly or
if hierarchy comes I'll send them a check and upgrade.

[...]

18. Robert BaerGuest

Take the case of triangle drive (TEK0018.BMP) and note the linear
rise to about the peak (large, first spike).
Then carefully note the partly exponential current increase as the
core saturates.
Fast dropoff at saturation, goes to zero.
Hysteresis happening here; drive is at slower rate so see slower,
lower amplitude pulse.
See same thing with reverse polarity drive.

Nothing fancy..

19. John SGuest

I did a ping on bayimg.com and got 192.121.86.164

A DNS lookup said it is in Stockholm, I think. See
http://www.dnsstuff.com/tools#ipInformation/type=domain&&value=192.121.86.164&&
(watch the wrap)

John

20. George HeroldGuest

Nothing fancy..

Thanks Robert, It may be 'nothing fancy' for you, but I'm still
trying to get my head around it. I'm 'mostly clueless', when it comes
to magnetic materials.
So let me try this as a hand wavy explanation,
If I just had one inductor, driven by a voltage source then I'd write
down the voltage as,

V = d/dt(L*I) +R*I
= I*dL/dt + L*dI/dt + R*I

So as I go into saturation dL/dt is negative. The first term looks a
bit like a negative resistance. More current is sucked from my voltage
source. And conversely coming out of saturation the first term is
positive.

I think I'll have to do the one inductor case.
There's nothing like data to help guide my thinking.
(I happy it's a 'real physics' thing and not a circuit screw up :^)

George H.
- Hide quoted text -