Coil and ground plane

nce of the

ype

the pcb

e,

th

to be

r than

s a

ker

ies

gy

rting

mal

he

t that

hort

and

eplace

e.

?).

se

une

Interesting thanks,

George H.

ort on

e)

ause

=A0Q is

e. =A0Q

tches

falls

t -

Thanks Phil, that's pretty much my model for understanding the frequency shift. (Less B field in some parts of space, for the same driving current.) That part of the problem never bothered me.

George H.

I could be wrong, because I'm not doing the math. But I could see how the speed of light could go down dramatically in a conductive material (as well as losses going up dramatically); this would cause a _real_ resonance, as long as the losses were low enough.

It's interesting that you can use it to measure thickness, though.

Thinking like that will get you to wrong conclusions. There is NO change in electrical speed in the material that causes the 'wave' pattern rather an interference pattern from a 'standing' wave. As the B field goes into the material it is attenuated and rotated. Skin depth is 1/e, but is also a phase shift.

If the material goes to infinity, the eddy currents just keep going down into the material as a function of depth, not time, we're talking steady state here. You can map currents at any depth as magnitude

Again, get a copy of free femm 4.2, use the axisymmetric model for a coil facing a solid plate, plot eddy current density as a function of depth and watch the eddy current's phase rotate as you go below the surface. Then nibble away at the plate's thickness and watch how the fields reflecting back up to the surface alternately reinforce and cancel the surface field, affecting inductance by alternately increasing/decreasing inductance, from what it would be had the material extended to infinity. And it 'looks' like a standing wave.

You can also kind of think in terms of superposition, one is the eddy currents in infinitely thick material and the other is what is necessary to make all the currents go to zero out in the air when you run of material THOSE two subtract back at the surfacecausing the apparent standing waves, in other words, the field back at the surface [and its effect on inductance] as a function of thickness.

Sorry, not clear, I've had my face in this stuff since 1989, and have not had to transfer to others, so never developed the vocabulary or concepts to describe what, for me, is a second hand nature image of what's happening.

of the

pcb

an

at

ce

ttdesign.com- Hide quoted text -

Hi Tim, Thanks again for the nice explanation... It always takes me a while for something to sink in.

So I think I've got the shorting plugs.... If I wrap a bit of, say, 24 gauge buss wire about my coil I'll shift the frequency and spoil the Q. (somewhere near Q=3D2 when I tried it.) But my sheet of 30 mil copper with a hole in the middle for the inductor to pass through, should have a nice shift and minimal Q loss.

I'll put a hole near the corner and report results...

George H.

Huh?

By definition, leakage inductance doesn't couple anywhere.

I think maybe the idea was the near by plane that acts as magnetic coupling in part of the physical coil that comes to a short, you are in effect doing a transformer linkage. The (K) maybe on the low side but the part of the coil that is not near one of these planes can then be considered as the leakage inductance.

Also, I do think with weak coupling like this it could also effectively lower the Q of the resonator.

I remember years ago back in my youth playing around with near by ground plates on a radiating coil, by adjusting the position of this plate I could not only move the resonant frequency but it would also effect the Q of the tank.

Jamie

I disagree. If it couples anywhere, that "anywhere" would be reflected into the primary circuit. That's not leakage inductance, that's unwanted coupling.

Leakage inductance can be thought of as a simple lumped inductance in series with the primary circuit, with no associated K or M.

No, it just doesn't couple as intended.

RL

Re-reading this thread,

Robert this is great! Thanks. It's the Kramers-Kronig relations again. If there's loss (or gain) there's gotta be a phase shift.

We use this trick to get a 90 degree phase shift in a Michelson interferometer. With a lossy beam splitter you get a shift between the 'output' beam and the beam reflected back towards the source. (With no loss these are naturally 180 out of phase.)

If you've got enough signal you might play this game for several wavelenghts. Did the device change the frequency too?

George H.

"Looks like", meaning, the circuit model equivalent. Leakage is the series inductance bridging the primary and secondary equivalent circuits. When k =

Tim

-- Deep Friar: a very philos>

No it can't, or rather, it can ONLY for very small values of leakage (k ~=

In nonlinear transformers, this is a valid model, because the core model component implements the dependent self inductance. Self inductance in saturation is then due entirely to leakage flux, and the coupling ratio in saturation appears as an inductance divider (actually a Y network) between primary leakage, core inductance (which is small but nonzero, depending on degree of saturation) and secondary leakage.

When the core is not saturated, leakage inductance remains constant (as it should, because it depends on geometry alone, not core state), but coupling factor is dramatically higher, because the core's effective inductance is mu_r times higher and the Y network divider is hardly loaded by it.

Tim

how

Not so bad really. I could follow it a little. It is a "Know the concepts, know the math, but i don't have a spoken language for it yet.", type deal. BTDT. BTW using femm 4.2 to visualize it a bit can really improve understanding, just got to get those boundary conditions right.

?-)

ow

At least with 4.2 if you forget to set the boundary conditions it defaults to A=3D0 which always gives you weird flux lines to alert you.

Just draw a circle around it and set the boundary to mixed at

There is another way to set the boundary to infinity, using two sections of space but it eludes me

how

I did not know of the Kramers-Kronig relations.

I don't think so. Wasn't looking for any shift. Thinking back...system was capable of 10-5 to 10-6 stability. A 20kHz tone shifted slightly would have appeared as coherent noise. Noise floors were plotted down into the 1 ppm, so thinking back, don't think there was any shift.

There however was a strange effect, the 'apparent' beam into metal was NOT like a flashlight peering around, but more like a water hose spraying around. The delay between manual position and illumination was THAT noticeable.

If you want to see sample image peering into 1 inch thick aluminum and seeing 10mil crack around a rivet, and see the 0.1 mil separation between the two 1/2 inch slabs, give me an email address and I'll send it to you. Not visible in the image, but the data shows it, is the 'tilt' to the rivet, not centered in the hole. You can see which side has more contact etc.

Amazing imagery when you consider the signal levels involved here. We're talking about how the conductive material slightly changes the inductance of the coil! It's like 1% of 1% of 1% of 1%....by the time you get there the signal is below 10ppm

how

I just looked up Kramers-Kronig relations ! The mere mention of causality makes me bristle! The mere mention of an inviable rule makes you immediately ask, "What if..." Then again, causality is another condition to be satisfied, ...under certain conditions.

So with a complete change of subject..What IFcausality were NOT a condition? Well, for one, you'd have spatial FFT's.Spatial analyses are non-causal, aren't they? Perhaps, not: the f(x) could be considered causal, and the f(-x) could be considered causal, after applying the KK relations to both THEN recombine to get back, food for thought, need more coffee.

Apply to imaging and crispning up a defocused image, by deconvolving a known spreading function (loss of contrast) with the resultant image, thereby getting back an original crisp image. For what it's worth, I still haven't been able to do that, ....yet. But, perhaps doing separately, as in f(x) and f(-x), THEN combining and the re-combining using additional restrictions, we'd be able to deconvolve! Ah, the illusive grail.

The thought of success at deconvolving is so tantalizing. just looking at the smoothed edges of a limited focus system, it seems one should be able to undo that 'spreading' damage and make the edges more visible. Are you doing work in that area?

e how

ial

the

Oh not a shift in frequency... but if you stimulated the system with different frequencies you'd see different skin depths. Different wavelegths.. (if I can use that term) perhaps better to say different thicknesses of material that gave the first 180 phase shift of the returned signal.

Sure, the email listed, gherold at teachspin dot com works.

Is there some link/ site describing how the system works. (I'm still a bit confused.)

e how

ial

the

Hi Robert, My understanding of the K-K relation is most likely flawed.* I was thinking about it in the shower this morning, and realized, I don't get lots of the subtleties. I didn=92t find the wiki article very helpful this was a bit better.

At it=92s basics it just says there is a relation between the real and imaginary part of a function. An LC resonance has both a phase shift and dip (or peak), the two are related. If you measure one you know all about the other. (in theory)

No, I'm only half following you. But would more detectors (coils) help? You've got blurring because of one big detector averaging the phase information. Would four smaller ones make it better. (See I told you I only half understood :^)

George H.

Hide quoted text -

Considering that we're not talking about an intentionally coupled structure here, just a discrete inductor with an 'open' field, references to leakage may be a little off-base.

Leakage inductance is the inductance of the winding structure with all other materials (other windings, shields , cores etc) removed. The effect of the core shape on this measurement is the core's limitations on the shape of the winding, not on it's ability to capture all flux lines.

The changes in L and Q measured are actually changes in coupling to external parts that represent 'core' and secondary winding material/coupling variations, not leakage.

RL