EEVblog #1329 - Magnetic Field Shielding DEMONSTRATED

[EEVblog]

A demonstration of near-field magnetic interference and how to shield it.



Near-field vs far-field EMC explained video:

[Axk]

So for in-wall mains wires detection one has to detect either magnetic or electric field because of the wavelength?
I wonder which of the fields is better in such detection and why.

[thm_w]

So for in-wall mains wires detection one has to detect either magnetic or electric field because of the wavelength?
I wonder which of the fields is better in such detection and why.

http://engineering.electrical-equipment.org/electrical-distribution/non-contact-ac-voltage-testers-work.html

[EEVblog]

So for in-wall mains wires detection one has to detect either magnetic or electric field because of the wavelength?
I wonder which of the fields is better in such detection and why.

They work even with no current flowing, so you automatically know it's not magnetic.

[dtmouton]

The copper PCB would not provide significant shielding of the magnetic field at 25 kHz. The skin depth of copper is 0.4123 mm at 25 kHz. The thickness of the copper on the board is 0.034798mm.  So the copper thickness is 1/11.85 of the skin depth!

Rather read up about magnetic shielding and skin depth than watching this video.  The one Ounce  PCB cannot shield magnetics fields with a frequency of 25 kHz.  The problem is probably still due to electric field. The most probable explanation: Without a conductive shield between the aluminum box and the screen, there is capacitive coupling between the box and the CRT screen. This causes a common-mode current to flow through the coaxial cable and into the measurement system. This comment mode current gets converted to a differential mode voltage inside the measurement system. That is what shows up in the measurement. The solution:  Move the box further away from the screen if you want to do accurate measurements.

[mag_therm]

The copper PCB would not provide significant shielding of the magnetic field at 25 kHz. The skin depth of copper is 0.4123 mm at 25 kHz. The thickness of the copper on the board is 0.034798mm.  So the copper thickness is 1/11.85 of the skin depth!

Rather read up about magnetic shielding and skin depth than watching this video.  The one Ounce  PCB cannot shield magnetics fields with a frequency of 25 kHz.  The problem is probably still due to electric field. The most probable explanation: Without a conductive shield between the aluminum box and the screen, there is capacitive coupling between the box and the CRT screen. This causes a common-mode current to flow through the coaxial cable and into the measurement system. This comment mode current gets converted to a differential mode voltage inside the measurement system. That is what shows up in the measurement. The solution:  Move the box further away from the screen if you want to do accurate measurements.

Hi dtmouton,
I think you have read about the solutions to the differential equations for magnetic flux only longitudinal (only parallel) to the surface of a conducting sheet,
and mis-applied them to this demonstration.

Those involved in transformers, RF, or (as I was) induction heating would know that a large sheet of copper will certainly attenuate the H field between a source and a sense coil.

So I set up a source coil on sig gen at 25 kHz and sense coil as shown in the attached photo (a bit rough!).
Rotating the sense ferrite coil shows that the fields in the 3 axis are about the same close to the end of the source soil.
The 'scope was on the lowest, 5 mV/div so I could not measure an accurate attenuation.

The 1 OZ copper pcb, shown in its unopened pack at rear of photo  was placed in between the coils.
The sense coil levels:
No copper pcb : 5 mV
copper pcb in < 0.5 mV
So the attenuation is at least 20 dB
 https://app.box.com/s/0y3mct3yfonwwoqk9ulqjxq278l6asaf

Here is my explanation which is a bit crude.
There will be H field components in all 3 axes, emanating from the coils in the 35660A, in their near field.
I think the strongest components in near field will be along the radii of a sphere centred on the coils.

H Field transverse, (or normal) to copper sheet will induce a J Field. This current density will be uniform through the thickness of the sheet and will
diminish increase out toward the extremities of the sheet. That means, current density does not diminish in the thickness direction. Skin effect does not apply.

A 2D model of AC magnetic solver should verify this, but a lot of mesh would be needed because the copper is so thin.

I did quite a lot of testing of this transverse flux mode of induction heating. It is  only used in certain applications . One application is heating thin strips which have skin effect problems in the ordinary longitudinal coils.

Edit corrected J diminish to J  increase at edges of copper

[dtmouton]

Hi mag_therm,

"I think you have read about the solutions to the differential equations for magnetic flux only longitudinal (only parallel) to the surface of a conducting sheet, and mis-applied them to this demonstration."

The skin depth plays an important role in many other configurations, as well. To quote the book entitled "Electromagnetic shielding" by Celozzi, Araneo and Lovat:

"A fundamental shielding parameter in determining the mechanism of the eddy current cancellation, but also in the flux-shunting mechanism when dealing with ac fields, is the value of the shield thickness compared with the skin depth."

I suggest that you have a look at this book. The skin depth often comes into play and does not only apply to magnetic flux longitudinal to the surface of a conducting sheet. It often makes its appearance when solving Maxwell's equations.

You propose that the current density is uniform throughout the thickness of the sheet. It this were the case, then the magnetic field on both sides of the sheet would be exactly the same, i.e. no magnetic shielding.

One flaw in your and (Dave's) experiment is that you use a CRT tube as your source. It emits both electric and magnetic fields. Rather try a coil as your source. Then we know that we are only dealing with magnetic fields.

I've been asked to leave this forum and will now make my exit.

[bdunham7]

One flaw in your and (Dave's) experiment is that you use a CRT tube as your source. It emits both electric and magnetic fields. Rather try a coil as your source. Then we know that we are only dealing with magnetic fields.

I didn't see a CRT in mag_therm's experiment.  Did you look at the link?

I've been asked to leave this forum and will now make my exit.

Nobody asked you to leave, it was just pointed out that you aren't trapped here.

[mag_therm]

It this were the case, then the magnetic field on both sides of the sheet would be exactly the same, i.e. no magnetic shielding.

Hi dtmouton,  Of course that is correct by Faraday's law. The alternative would be that there would be a different E  ( = -N*dphi / dt) and hence J on each side of the thin copper sheet which is impossible.
The "shielding" occurs because the J eddy currents can be considered to generate an opposing field to the driving field. Hence the nett field at the surfaces of the shield will be the same inside and outside the shield.

I did not use a CRT, I connected the larger coil across some capacitors to get close to resonance then conected to a 50 Ohm pure sinewave , ground isolated from a AF generator at 25000 Hz.
There was only about 1 Volt across the source coil.
The attenuation copper shield : no shield was > 20 dB .

I also think your supposition that the shielding in the demonstration was by capacitance is impossible,  given the low frequency  and the distances shown.
And in my test, the sense coil connected to the 'scope by a shielded cable with only 50mm or so exposed.
So how could capacitance from the source have any effect at that frequency and voltage?

Regards,

[dtmouton]

"Of course that is correct by Faraday's law. The alternative would be that there would be a different E  ( = -N*dphi / dt) and hence J on each side of the thin copper sheet which is impossible.
The "shielding" occurs because the J eddy currents can be considered to generate an opposing field to the driving field. Hence the nett field at the surfaces of the shield will be the same inside and outside the shield."

Hi mag_therm,

Where does N come from in your equation? This is not a wound inductor or transformer?

I think you fundamentally misunderstand the mechanisms that lead to magnetic shielding.

It is well-known that the shielding of the magnetic field is due to the eddy currents. However, there is more to the story. You also need the skin effect. For instance if you use a shielded enclosure then currents that flow on the outside surface of the enclosure must stay on the outside and currents that flow on the inside of the enclosure must stay on the inside. So we need the difference in current density.

My apologies for misunderstanding your setup. Yes. Electric fields will not play a role in your setup.

I suggest that you also have a look at the well-controlled experiments conducted in this video:


He gets 15 dB of attenuation with a copper shield that is approximately equal to the skin depth. It is unclear how you can get 20 dB when your shield thickness is less that 1/10 th of the skin depth?

[Kleinstein]

The setup in the shileding video is a bit different. For one point the shields are a bit small for the coils so part of the signal still getting throung is from field going around the sheet. This is especially obvious with the thick aluminum.
The other difference in the setup is that the shield is very close to the sending coild. Chances are that the driving current in the sending coild will increase with a conductor present, as the inductance of the coil will go down with the shield.

In Daves example the field source is quite a bit away from the source and the shield is larger area than just the coil(s). So it gets much easier to divert the field.

Even if the incident field (without the shield) is perpendicular to the shield, when shilding the magentic field the resulting field needs to go near parallel to the plane. So the case of parallel to plane and thus the skin depth is not that different. It is still not so easy as there is more material around and extra distance between the shield and "detector".

[EEVblog]

I've been asked to leave this forum and will now make my exit.

No one asked you to leave the forum.