Screening effectiveness of boxes made from PCB

[Alan_G3XAQ]

This is a two-part question about screened enclosures.

If I need to screen a circuit from external RF my first choice might be to enclose it in a tinplate box, such as

https://www.tekoenclosures.com/en/products/family/RF

A fairly standard approach seems to be to solder the ground plane of the enclosed board to the box. I do wonder if the lower conductivity of the tin (15% that of copper) makes much difference when soldering feedthrough capacitors and coax sockets to the box walls? One might guess that the "ground resistance" could muck up things like the stopband attenuation of filters.

So, on to the second part. How about using seam-soldered PCB material to make an enclosure? This is handy for circuits that are an inconvenient size for an off the shelf tin box. The downside here is that the copper on standard (1oz) board is only 35 microns thick, which is about half a skin depth at 1MHz and not much more than a skin depth at 10MHz. So that's not going to be much cop, is it? How about using double sided board? Will this act as one 70 micron layer or (more likely) two coupled layers? More importantly, will double sided be just a bit better (sqrt 2) than single sides or a lot better or no better at all?

[Craig Sawyers]

I'm currently developing a product that needs high sensitivity portions to be screened. I'm using a product that I have used before, manufactured by Perancea http://www.perancea.com/prod-fr.htm . A limited range is available from RS and Farnell, although Perancea will sell the full range at a £100 minimum order, and also do specials to supplied drawings.

I'm using PCB mounted ones (75x75 mm x 25 height; RS at around £12), although I've used freestanding ones too.

These are tin plated steel, and are excellent for magnetic fields in particular. An overall enclosure in aluminium deals with electric fields.

However, there are other ways of skinning the cat. The late great Jim Williams constructed a lot of his stuff using soldered together double sided PBC, and dead bug construction.

Yet another approach, used by an old Whafedale colleague (Richard Lee), who constructed his exceptionally low noise moving coil head amp inside a Duraglit tin; in fact he called the design "Duraglit" for that reason.

Craig

[marty_ell]

Originally Posted by Alan_G3XAQ

If I need to screen a circuit from external RF my first choice might be to enclose it in a tinplate box, such as

https://www.tekoenclosures.com/en/products/family/RF

A fairly standard approach seems to be to solder the ground plane of the enclosed board to the box. I do wonder if the lower conductivity of the tin (15% that of copper) makes much difference when soldering feedthrough capacitors and coax sockets to the box walls? One might guess that the "ground resistance" could muck up things like the stopband attenuation of filters.

For all on-the-bench practical purposes, the material of the enclosure makes little difference. If all the box's seams are continuously soldered then only that which goes through surfaces need be considered - in this case via the feedthroughs/sockets - that is where any practical leakage occurs. Sockets are the worst case; bolting is never as good as soldering and, even if solidly soldered, such as SO239 are never as good as eg. SMA. Assuming you pick your feedthroughs/sockets for the band of interest then tin plate is fine.

If you need ventilation you may use an array of holes, ensuring the diameter and separation is far below the wavelength of interest . Alternatively use mesh (which must be continuous) whose apertures must be small. Mesh material is more critical; anything worse than brass (eg. zinc or steel) is problematic at VHF upwards.

Don't forget that the dimensions of the box will affect VHF/UHF/microwave filters unless the signal is properly constrained by microstrip practices (unless you are deliberately using the box as a resonator). HF filters generally ignore the enclosure as long as open inductors are far enough away from box surfaces.

Originally Posted by Alan_G3XAQ

So, on to the second part. How about using seam-soldered PCB material to make an enclosure? This is handy for circuits that are an inconvenient size for an off the shelf tin box. The downside here is that the copper on standard (1oz) board is only 35 microns thick, which is about half a skin depth at 1MHz and not much more than a skin depth at 10MHz. So that's not going to be much cop, is it? How about using double sided board? Will this act as one 70 micron layer or (more likely) two coupled layers? More importantly, will double sided be just a bit better (sqrt 2) than single sides or a lot better or no better at all?

Again, the principal requirement is total enclosure. PCB material has been used to cobble together high attenuation boxes for decades; see any RSGB/ARRL handbook since the 60's. The field strength of what little signal gets through the box's skin depth falls essentially to zero within a fraction of the wavelength. 1oz copper is fine and cheap !

As an example I put together a UHF signal source in such a box with one switch, one feedthrough (for power) and a tiny hole through which an adjustable probe would protrude. When commissioning with the probe absent I though there must be something wrong with my VNA, it could detect nothing above -140dbM, the noise floor of the VNA at UHF!

Single sided material is best to remove all worries about possible magnetic field coupling from inside the box to the outside copper. Double-sided should be seam soldered inside and out and feedthroughs/sockets need double-sided soldering.

Hope that helps, Marty

[emeritus]

We used to use die cast Eddystone boxes at Plessey in the 1970's. We did have problems with signal leakage that were traced to the use of BNC connectors, established by soldering them up solid in desperation. Solved by only using type N or SMA screw connectors. The frequencies involved were in the UHF band, 400 to 600MHz. RF gaskets were found to be unnecessary.

[Radio Wrangler]

Pick a material, any material. Show it to my beautiful assistant, whisper your frequency of interest, and she will look up its conductivity, look up the right equation for skin depth, and then calculate the skin depth at your lowest frequency of interest.

So long as your material is at least a few skin depths thick, it will do.

For metals and RF, you're probably OK

Works for metals, submerged submarines in seawater, the lot.

Where screening fails is usually due to gaps where lids don't make full contact around their periphery. View such an insulated slot as an Alford slot antenna from inside to outside. Coupling peaks where the length hits a resonant frequency.

So you need positive fixings << Lambda apart.

Connectors are easy leaks. Use screened connectors, screened cables and you want solid grounding where they go through your screening skin.

Get it wrong and you have a loop antenna inside the box connected straight out to a whip antenna. You might as well forget the box.

David

[G6Tanuki]

I guess the answer is 'it depends'.

I've used double-sided PCB material to make up ad-hoc screened boxes for things lile direct-conversion receivers [where a big problem is the local-oscillator signal, which can be several hundred Milliwatts in the case of a switching-diode mixer, 'leaking' and getting into the front-end with who-can-guess-what phase-relationship to the LO signal you're feeding to the mixer's LO port].
A double-sided-PCB box proved remarkably effective - I found it 'a'good idea' to do what I could to keep the inner and outer foils with separate 'identities' at RF - so you essentially had two concentric boxes separated by a dielectric.

Just one inadvertently-placed feed-through could get you an additional 10dB of leakage.

Equally, beware of inadvertent circulating-currents and slice the PCB where this will limit their ability-to-travel. Remember the first RACAL RA17 receivers which had 'strange' cuts in their chassis to manage circulating currents!

As another example: the Pye Westminster W15FM power-output strip has a piece of single-sided PCB as its cover. Once, I had a Wessie whose cover had been mislaid and I cut a bit of aluminium as a replacement. Try as I could it just wouldn't tune-up properly. Then I grabbed another radio from the pile and looked at its screen. Yes, the copper on the PCB was divided into several distinct areas. Make up a PCB-screen to the same spec as the original and it tuned-up perfectly!

Lesson Learned.

Also, there are instances where a 'lossy screen' can be what you want: some gear I worked on in the 80s had plastic cabinets internally sprayed with a coating consisting of large aluminium flakes and very fine carbon-powder. It was resistive, so attenuating the RF from within.

Good idea - but after a few years the coating started to flake off and the cooling-fans happily spread the flakes into areas where there was 10 Kilovolts of EHT. The resulting 'fizzle! crackle!' was not a good thing to have around CMOS memory-chips.

[Alan_G3XAQ]

>Pick a material, any material. Show it to my beautiful assistant,

Eh? Which of the neddies is this one?

> whisper your frequency of interest, and she will look up its conductivity, look up the right equation for skin depth, and then calculate the skin depth at your lowest frequency of interest.

Yep done that.

> So long as your material is at least a few skin depths thick, it will do.

Sadly PCB falls at the first hurdle (fence?). MIT says you need 4 skin depths to contain 98% of the the current. Pick an online skin depth calculator and see that 35 micron "1 oz" PCB doesn't meet that spec until your reach 55MHz. S'ok for your VHF types but for us near-DC folks it seems like it's a dud. At least it is if the unwanted signal is primarily coupling via a magnetic field.

> For metals and RF, you're probably OK

Hmm, or maybe not on TopBand or MW.

> So you need positive fixings << Lambda apart.

How about the commercial tinplate boxes with "finger spring" lids? Are you going to make me solder the lids down?

>Get it wrong and you have a loop antenna inside the box connected straight out to a whip antenna. You might as well forget the box.

Yes, I can see that soldered feedthroughs and coax sockets will be de rigeur

Sometimes lifting these stones makes me wish I'd left well alone.

Alan

[Alan_G3XAQ]

>Just one inadvertently-placed feed-through could get you an additional 10dB of leakage.

I've not come across this before. Where are the right and wrong places to place feedthroughts?

Alan

[G6Tanuki]

Originally Posted by Alan_G3XAQ

>Just one inadvertently-placed feed-through could get you an additional 10dB of leakage.

I've not come across this before. Where are the right and wrong places to place feedthroughts?

Alan

The situation I had was essentially two concentric screened-boxes separated by the PCB as dielectric.

Feedthroughs were by default grounded to the inner bo [and decoupled further when they emerged]. One misplaced feedthrough was grounded to both inner and outer boxes - so providing a DC-to-light short between the inner and outer boxes and negating the whole idea of two-boxes.

[Alan_G3XAQ]

Originally Posted by G6Tanuki

Feedthroughs were by default grounded to the inner bo [and decoupled further when they emerged]. One misplaced feedthrough was grounded to both inner and outer boxes - so providing a DC-to-light short between the inner and outer boxes and negating the whole idea of two-boxes.

OK, I get it now. I think that means that in your case the undesired coupling to the outside world was capacitive rather than inductive. I wonder where the balance between the two modes lies most of the time in practice?

Thanks, Alan

[Steve G4WCS]

those "barista style " coffee tins are a good source of material. easily worked and very solderable once the paint/lacquer is sanded off.