Multilayer magnetic shielding

Can you get deep enough into the Murata power module to stick a loop of wire on top of the toroid to counter-act the single turn formed by the embedded progressively wound toroid? It won't do a perfect job. but it could help quite a bit.

Ripping out the toroid, ripping off half the offending turns and putting them back on to make a non-progressive winding would be a lot more work but might be even more effective.

It's probably a silly question, but not all that silly.

I use the Murata converters on the bottom of my boards, with the jitter-sensitive parts on top. One ground plane and three power pours between. But I really use the Muratas for their very low output common-mode capacitance, which couples a fraction of the switching current of other converters.

Copper fights fields with eddy currents. PCB thickness copper is transparent at 60 Hz, basically no help at all. It's hard to shield 60 Hz fields.

Murata is doing radical things with PC boards, like that converter with PCB windings and the embedded toroid.

A similar(?) effect is using higher order LC filters: to a point, less energy storage is needed (less total L and C) for a given attenuation at some frequency -- because Fc can be closer. Also may have better stopband attenuation due to more stages of parasitics dividing down (and also using smaller individual components, that are better behaved).

Still, nothing should beat that aluminum plate? I wonder if E field was uncontrolled in your test -- what would've happened if it was grounded to the coax probe, or stomp box?

The coax loop probe as described, isn't perfectly shielded because it's still sensitive to 2nd order effects: E-field coupling into the shield, dropping a voltage across its inductance (so, this way is aggressively more significant at HF) and thus reading a signal off the loop. Or for the best common case, shield split opposite the cable end: perpendicular field won't couple due to symmetry, but a sideways component still can.

But due to that cutoff asymptote, this probably shouldn't explain fundamental getting through.

Hmm, what were the "other things" at -40dB? You say the plate was about equivalent to five layers, but you say four layers stops working? How did you figure this out then?

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website:

Agreed. Two picofarads is really slick.

Well, the skin depth formula has a 1/sqrt(mu) in it, so a bit of magnetic stainless (mu ~ 1500) or low-carbon steel (mu ~4000) can help a lot at low frequency. (Mu metal is generally disappointing because if you look at it crossways it has to be hydrogen-annealed all over again.)

But with nonmagnetic shields at least, multilayer does a massively better job per gram of metal--it's the magnetic equivalent of superinsulation in a vacuum chamber.

It's sort of like the Dick Garwin big-box effect that Win talks about--if the shield is really continuous it works roughly like a metal sheet with thickness

t_eff ~ sqrt(t_shield * D_box),

where D_box is a typical dimension of the box (e.g. its length). That's quite a startling improvement with thin metal.

If they'd just use a non-progressive winding, birds would really chirp.

Cheers

Phil Hobbs

Works amazingly well below 10 MHz, for sure. The loop is only 7x12 mm, which isn't a very large fraction of 30 metres. Also you're neglecting the skin effect--ambient HF E fields will induce current on the outside of the shield but not the inside, so it won't couple to the centre conductor.

I admit that hardline would probably work better--a millimetre of solid copper is pretty impermeable to RF--but this gizmo I built out of a dead patch cord is pretty slick, I think.

The main test is that I get a really deep null by orienting the loop correctly. A monopole E-field probe doesn't do that.

The aluminum plate was slightly larger, and it influenced the (small) coupling from the input and output inductors.

The main line of evidence is that the skin depth in copper at 90 kHz is

So it ain't skin effect that's doing the heavy lifting at 90 kHz. (The high harmonics die out much more rapidly, so skin effect is having more of an impact out there, as you'd expect.)

Cheers

Phil Hobbs

Could you test with grafoil? Graphite sheets are a 2-D conductor, might have a multilayer effect that goes into the nanoscale, and it's cheap.

Possibly. They don't bend or solder as well as copper, though. ;)

Cheers

Phil Hobbs

As you say later, the closed box of conductor effect is pretty useful. It works both ways.. I put a copper tape box over a three terminal SM regulator, and interference took a big drop when I tacked the copper tape to the ground plane.

Your new thing with conductive planes sounds a bit different. Are the dimensions such that the plane looks ~ infinite to the source and detector. (I always need pictures.)

George H.

Your wish, etc.

Cheers

Phil Hobbs

That should be free and easy on the PCB layout.

Youtube has x-rays, with music.

Am 14.02.21 um 20:39 schrieb Phil Hobbs:

This one is broken?

Gerhard

Fixed, thanks.

Cheers

Phil Hobbs

-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510

The steel is actually 10 mils, i.e. 0.25 mm, not 0.1 mm.

Cheers

Phil Hobbs

Thanks... GH

Phil Hobbs needs to learn to trim.

=================================

Mu-Metal is up to 100 times better.

Millions of audio transformer casings and scope tube shields would disagree with your pessimism

..... Phil

Have you ever looked at a B-H curve for those steels? You're quite right that mu = dB/dH changes, and its maximum often isn't at H=0, but H_0 is around 100 for carbon steel and 1500 for good magnetic stainless.

The formula is correct, so it does in fact reduce the skin depth accordingly.

Well, you folks seem to think that your own primitive plumbing is somehow a lethal insult to Americans, so who knows?

The shielding effectiveness of mu metal gets trashed if you bend it, and is seriously reduced by just dropping it on the floor. The only cure is re-annealing.

If you have a high volume application and are very careful, mu metal is terrific.

Cheers

Phil Hobbs

Phil Hobbs wrote: Phil Hobbs needs to learn to trim. =================================

Go try some carbon steel or magnetic stainless shielding on a scope tube and see how wrong you are.

Fuckwit.

..... Phil

I bet you say that to all the boys. ;)

Cheers

Phil Hobbs

Phil Hobbs is an Evil POS narcissist:

========================

All of it apparently too real for the Hobbs prick to answer.

..... Phil