Shielding - Conductivity vs Skin Depth

[fonograph]

I read this rule of thumb that 1 skin depth thick conductor = 8db of shielding.So I need 10 skin depths thick metal shield to block 80 db.

This brings me to the main question,copper or nickel? Copper have much higher conductivity but nickel have much smaller skin depth.That means for fixed thickness,nickel will provide more skin depths.Which parameter is more important in shielding? Should I chase minimum skin depth material or highest conductivity material?

[Bud]

I can't recall seeing nickel shielding ever in my life. Nickel is used as plating for cheap RF connectors and it sucks donkey's balls in making good conductive connection.

[T3sl4co1l]

Nickel's also rather more expensive than iron, which has the same properties, and is easily soldered (particularly when provided with tin plating).

Aluminum has the same problem, of course; or worse, depending on what surface treatment it has (anodize is insulating!), and how it's mounted.  Bare or dichromate treated, secured bolts with star washers, probably okay; anodize with smooth washers or none at all, probably pants.

I'd guess they're comparable, copper and iron that is, with iron having more advantage at low frequencies for obvious reasons (magnetic shielding).

Oh, it's probably worth noting, most steel drops off in the MHz, magnetically, so its skin depth is deeper than you'd expect.  This probably gives the advantage to copper.  At 100s MHz+, you don't need much, anyway; you'll have far more leakage from gaps and sneak paths, than through the bulk of any metal that's thick enough to be self-supporting.

Tim

[CopperCone]

you need to coat copper with varnish or it can get nasty. I recommend nickle silver. Solders nicely, looks nice, easy to bend and shape. https://en.wikipedia.org/wiki/Nickel_silver

[mc172]

It's very difficult to coat things in iron.

Nickel doesn't "suck donkeys balls", you can easily achieve 140 dB screening effectiveness with nickel connectors, probably more - that's the noise floor of my gear with an IF BW of 1 Hz.
The only issue with it is that it is slightly magnetic and some people are allergic to it.
What you want is something conductive that doesn't oxidise readily (like aluminium which oxidises extremely quickly), something non-magnetic and something with a galvanic potential reasonably close to the other thing you're mating to.

Gold is often used on cheap connectors but gold is usually put on in thicknesses under 0.2 microns, which doesn't support the skin effect until well over 100 GHz, but it does not oxidise very much which is 99% of the appeal.

Naturally then the choice is silver, which oxidises nicely and has a similar skin depth to gold. The oxide is very conductive, though.

Anyway, I don't think much of this will really help you in the real world. Gold has a massive skin depth but works in the real world. Nickel is magnetic but most RF gear has either nickel plated brass or stainless N connectors sticking out the front.

What frequency are you planning on operating at? Up to a few GHz you are not going to have any material related problems achieving 80 dB, skin effect is miles away and you are relying on pure conductivity, which goes up if you allow the material to oxidise. Your biggest problem is going to be filling up all of the holes in the box.

[fonograph]

Thanks forgreat tips,I really appreciate it but my main question was not answered,I ask again.The rule of thumb I read said 1 skin depth thick shield is 8db of attenuation,nickel have far thinner skin depth.Isnt nickel better shielding material becose in shield made of same thickness it will provide more skin depths hence more attenuation?

Soldering and corrosion resistance is surely important but I am not asking about that.

[xaxaxa]

Thanks forgreat tips,I really appreciate it but my main question was not answered,I ask again.The rule of thumb I read said 1 skin depth thick shield is 8db of attenuation,nickel have far thinner skin depth.Isnt nickel better shielding material becose in shield made of same thickness it will provide more skin depths hence more attenuation?

Soldering and corrosion resistance is surely important but I am not asking about that.

skin depth is a function of permittivity and permeability which are both functions of frequency; magnetic materials usually are only magnetic at lower frequencies (<10MHz) so as frequency goes up the skin depth of nickel increases, and >1GHz I would bet it's worse than copper; I would look for a graph of permeability of nickel vs frequency, and calculating skin depth at the desired frequency.

[fonograph]

I would look for a graph of permeability of nickel vs frequency, and calculating skin depth at the desired frequency.

I did that before making this thread,here is the permeability graph.

But this calculator says the skin depth in nickel is like 10 time shallower at 1 Tera Hertz! https://chemandy.com/calculators/skin-effect-calculator.htm

Also this picture shows nickel having shallowest skin depth.

[T3sl4co1l]

Well yeah, if you use an approximate formula well beyond where it is valid, you will get nonsense results.  That's been true since Charles Babbage's days, to reference a more memorable quote...

Of note, common metals have weird behavior at THz to optical frequencies, like negative permittivity.  The penetration depth needn't be proportional to skin effect anymore.  For example, copper has a clearly apparent change, in the green to blue range: it goes from reflection to absorption!  (Hence, the pink appearance of clean copper metal.)

AFAIK, ferromagnetism drops off in the low GHz, as it's similar to (a group action versus lone phenomenon) electron paramagnetism, which resonates at similar frequencies in modest magnetic fields.  Most materials drop off lower, partly due to physics and partly due to geometry.  So, bulk steel or nickel rolls off way lower (~Hz) than laminations (~kHz), which roll off lower than ferrite (~MHz), which rolls off lower than powder (~100MHz?), which rolls off lower than garnet (~GHz?), and not much works above ~10GHz or thereabouts).

And all this varies with dopants/impurities, handling (work hardening generates dislocations), bias (ferromagnetic materials are nonlinear, full stop), temperature and so on.  Probably not with the phase of Jupiter, although if you want to bring metallic hydrogen into this, who knows...

Tim

[Bud]

Nickel doesn't "suck donkeys balls", you can easily achieve 140 dB screening effectiveness with nickel connectors, probably more - that's the noise floor of my gear with an IF BW of 1 Hz.

It does and you can see it in real time on a VNA, trying to mate nickel plated connectors. The Return Loss will change and jump depending on mechanical force you apply.

[mc172]

It jumps around because you have not applied the correct preload to the connector. I've got one right here. Yes it jumps about a bit if you've not tightened the connectors properly but not enough to worry about. The jumping is not because of the nickel.

Attached is the EMC performance of a pair of nickel plated N connectors. IF bandwidth is 10 Hz. If I had set it to 1 Hz it would be off the bottom of the screen.

[fonograph]

T3sl4co1l  Thank you for avalanche of top quality info.I didnt know about that negative permitivity,also I thought ferrite is highest frequency,what is this powder and garnet stuff? Is it like yttrium aluminum garnet aka YAG,the laser crystal?

[T3sl4co1l]

T3sl4co1l  Thank you for avalanche of top quality info.I didnt know about that negative permitivity,also I thought ferrite is highest frequency,what is this powder and garnet stuff? Is it like yttrium aluminum garnet aka YAG,the laser crystal?

Powder is just powdered iron, usually used in the VHF range.  Small grains of iron (and often other metals), glued together to make solid cores.  Grains small enough that skin effect isn't significant until much higher frequencies than other forms.

Well, to be fair, "ferrite" covers a lot more than just the most common (MnZn) power ferrite materials.  Chemically speaking, YIG is a ferrite, too (i.e., a "salt" of "ferrous acid" H3FeO3 or such, as it were), but really, in practical terms, YIG is just a mixed oxide that happens to crystallize in the garnet family.

"Like YAG"?  Yes, like -- but with Fe(III) replacing Al, which works out pretty nicely in most crystals.

Fe(III), Al, Cr(III), and a number of other ions, fit neatly into each others' spaces, so that suitable crystals can potentially contain any of them in any mixture -- called a solid solution.  So, you get yttrium ferrite.

On a related note, colorful atoms like Cr can give pretty colors to otherwise-boring crystals.  You can get Cr(III) substituting in Al2O3 (corundum), which gives ruby.  Which can be excited with blue to UV light (which it absorbs, hence the red color appearance), and voila, ruby laser!   Or, same thing with Nd:YAG, or...

Conventional ferrites are in the spinel family, by the way -- ordinary spinel is MgAl2O4, but we can substitute Zn, Mn, Ni(II), Fe(II) and others for Mg, and Fe(III) for Al, and the resulting material is usually magnetic.  As it happens, substituting a mixture of Mn and Zn (written (Mn,Zn)Fe2O4) gives a soft magnet (does not retain magnetization) with high permeability and reasonable saturation flux density, while substituting a mixture of Ni and Zn gives a higher frequency limit (usually 10-1000MHz) at the expense of lower permeability and saturation.

Incidentally, note that magnetite Fe3O4 is also Fe(II)Fe(III)2O4, i.e., ferrous ferrite, a naturally occurring ferrite, in the spinel family, that is a hard magnet (retains magnetization).  Typical ferrite permanent magnets are made with strontium or barium, which don't fit very well in place of Mg, so crystallize in a different (hexagonal) form.  And, for whatever reason, that form happens to give a hard magnet.

If chemistry makes your eyes glaze over, suffice it to say, it's a pinch of this and a dash of that, mixed together and heated to make a ceramic (or melted to grow a crystal, or..).  Although exact formulations are rather more precise than that sounds, as it doesn't take much impurity to, say, wreck a mu_r >= 10k mix.

Tim

[fonograph]

To wrap up this thread,the conclusion is that the linked calculator sucks balls and  the result it gives is complete bollocks becose it doesnt account for frequency variable magnetic permeability that falls off hard around 1 MHz.Since the permeability at 500 MHz is 1,same like copper,but its conductivity is much lower,that means copper shits all over nickel past 500 MHz when it comes to blocking anything.So copper for high frequency or electric shielding and nickel for low frequency or magnetic shielding... is that about right?


T3sl4co1l Is that YIG used like  uber fancy core material for some super high frequency transformers? I wonder what is ultimate high frequency transformer core material except air.

[TheUnnamedNewbie]

T3sl4co1l is that YIG used like some uber fancy transformer core for some super high frequency transformers? I wonder what is ultimate high frequency transformer core material except air.

I don't know what you mean with "super high frequency". At the frequencies I work at (>100 GHz, though people do the same starting at a few GHz already) to just ditch a magnetic core and go with whatever dielectric we are working with - usually SiOx since I work with CMOS. This adds other advantages like not having to deal with non-linear behavior of your core. On chip, things get so small anyways that we can just put them so close together the magnetic flux can't leak out. In fact, we run into issues with capacitive coupling - you don't need much capacitance to couple a 150 GHz signal over, and so we are often faced with moving the transformer winding(s) apart further to tune them and make them behave like we want. Keep in mind that these transformers often have 1 winding.

Image source: Han, Jiang-An et al. “CMOS 1:1 Transformer design for millimeter wave application.” 2014 XXXIth URSI General Assembly and Scientific Symposium (URSI GASS) (2014): 1-4.

[fonograph]

That reminds me of that YIG  thing in those old HP spectrum analysers,I believe that one is tunable bandpass filter,and this one you show is transformer?

[TheUnnamedNewbie]

That reminds me of that YIG  thing in those old HP spectrum analysers,I believe that one is tunable bandpass filter,and this one you show is transformer?

YIG is Yttrium iron garnet. It is used for many things in mirowave engineering, such as tunable filters and oscillators. The reason for this is the low loss, and the fact that the porperties of a YIG sphere change with the application of an external magnetic field (don't ask me about the physics that causes this as I don't know it either - not my field). I think they go up to about 10 GHz, but I'm not sure.

The pictures I show are indeed transformers. Keep in mind that these transformers are tiny - like, really, really tiny. Here is a picture with an example. This is from someone in my team a few years ago:



Source: N. Van Thienen, W. Volkaerts and P. Reynaert, "A Multi-Gigabit CPFSK Polymer Microwave Fiber Communication Link in 40 nm CMOS," in IEEE Journal of Solid-State Circuits, vol. 51, no. 8, pp. 1952-1958, Aug. 2016.
doi: 10.1109/JSSC.2016.2580605

To give you a sense of scale: this entire chip is probably just over 1.5x1.5 mm. Starting in the top left you can first see a 3 sets of 2 inductors at low frequencies. The big H shape is an antenna, and you can see a number of transformers (the little circles) in the chain connecting the transformers to transistors (the locations where the two lines from the transformers stop and come together).

[T3sl4co1l]

To wrap up this thread,the conclusion is that the linked calculator sucks balls and  the result it gives is complete bollocks becose it doesnt account for frequency variable magnetic permeability that falls off hard around 1 MHz.Since the permeability at 500 MHz is 1,same like copper,but its conductivity is much lower,that means copper shits all over nickel past 500 MHz when it comes to blocking anything.So copper for high frequency or electric shielding and nickel for low frequency or magnetic shielding... is that about right?

That's one possible interpretation, yes.

You're rather missing the point of what an approximation is, and what's given up in the process, though.

The real world is more subtle than the heavily compartmentalized description you are giving.  The real world is continuous, as is uncertainty about it.

Just because it is not clear, based on given data, how good nickel or copper are at high frequencies, does not mean one shits on the other.
You are welcome -- and encouraged -- to research these missing data yourself.  Can you find permeability of nickel vs. frequency?  Or can you leapfrog that entirely, and find shielding effectiveness for various materials, versus frequency?  (Very likely this is out there, as a scanned article, or on Google Books say, ripe for the finding.)

If you find it challenging to think about continuous phenomena, perhaps you will find it easier to stick with something discrete, like computer science.  The study of ones and zeroes is literally as compartmentalized as you can get. (This is not at all an insult!  Do the things you are best at!  CSci pays quite well these days, indeed you might find a better paying job than I have...)

T3sl4co1l Is that YIG used like  uber fancy core material for some super high frequency transformers? I wonder what is ultimate high frequency transformer core material except air.

Not so much transformers at those frequencies.  AFAIK, it has bulk low frequency properties (permeability and such) just the same as any other core material, but it is usually used as an optical medium (i.e., the speed of light is slower within the material, and also susceptible to Faraday effect -- rotation proportional to magnetization, which allows the creation of isolators and circulators), or for electron resonance (which gives YIG tuned oscillators a wide tuning range with reasonable stability).

FWIW, conventional (Guanella -- matched delay) transmission line transformers are limited, not by the core, or winding length, but by the geometry of the transmission line used (and any error in delay matching).  I suppose some thin semirigid coax wound around nanocrystalline cores could potentially give bandwidth from ~10kHz to many GHz.

Tim

[fonograph]

Can you find permeability of nickel vs. frequency?

I found and posted it here before you made your post,did you not see?

[T3sl4co1l]

Can you find permeability of nickel vs. frequency?

I found and posted it here before you made your post,did you not see?

No, I did not!

Hmm, embedded from a site that doesn't allow remote embed it seems.  Ah, and you got caught by it because your browser caches it, of course.  Ah, I love it when that happens...

Edit: here's what was missing.



It's from Researchgate, a repository of papers (or something like that, I never actually checked what they are, come to think of it), could you link the original paper?

And yes, if these data are true for any thickness of ordinary sheet nickel, then it will indeed be a poor shield at high frequencies, and copper will do better (past about, oh, say, 10MHz I guess?).  Up at 500MHz, you would need an extremely thin foil of either metal to get poor shielding response -- for any practical thickness (say, 0.2mm sheet?), both will be equally good (>100dB?).

Tim