To be or not to be EMI shielding...

[Kaithar]

Okie,  I need some advice on a subject I'm not qualified in... so I came here.  Be nice...

I'm doing the feasibility and planning stage of a PC case modding project, part of what I want to achieve is replacing as much of the physical support structure as possible with clear acrylic to build an exhibition case.  The problem is that there's a lot metal involved that supports secondary functions beyond just being a nice bit of solid material.  There are three particular pieces of metal I want a second opinion on:

1/ Motherboard plate
I'm assuming that the reason every case uses a metal plate is to give a base that shouldn't be able deform and is safer against accidental piercing the case and contacting something.  However I know that the plate is grounded... does anyone know if the case is sinking any emissions from the motherboard?  I'm guessing not, since cases with windows in them would leak too much, but I'm asking anyway.

2/ The PSU case
Again, I know the case is very much there for grounding and a little bit for heat dissipation.  I also know there's going to be some potentially noise generating circuits in that, switch mode psu after all.  Is the psu case likely to be shielding any emissions?  I'm on the fence as to if I really want to go as far as swapping the PSU case for an acrylic one though.

3/ The TFT case
This is the big one that inspired a good chunk of the project: Take a monitor and replace all the opaque plastic and metal with clear acrylic so that the guts are visible.  There is a crazy amount of metal in the TFT construction.  There's also TFTs in laptops that have almost no shielding, thus I'm trying to work out exactly what is and isn't shielding.  Breaking it down there are 5 distinct parts that might be radiating:
- the TFT layer,
- the switching lines in the ribbon cables driving the display,
- the driver board itself that's converting the input signal in to instructions to the board,
- the backlight and it's inverter circuit,
- the primary PSU.

The research I've done so far indicates that there is a small amount of emission from the TFT panel, a potentially a noticable amount from the driving ribbons, and a butt load from the backlights.
My plan is to replace the CCFL backlight with an LED based one so I can surgically extract the inverter circuitry as well.  The TFT radiations seem to be low enough that it doesn't get much additional shielding.  The sticking point is the driver board and ribbon cables... There is enough metal in the construction to suggest that monitor manufacturers are shielding them from EMI, but the much lower metal content in laptops suggests to me that it doesn't radiate enough to worry about.

Are these boards liable to be something I need to worry about shielding (say with some metallised tape) or can I safely work on the assumption that it's not going to put out enough junk to be of note?

Also, since we're on the subject, is there a cheap kludgy way I can check the emissions without having to go through the cost of an FCC/CE grade EMC test?  I'm guessing that, with the exception of possibly the panel driver lines, most of the EMI I might be generating is going to be low enough frequency that an RTL-SDR won't pick up much of interest... is an antenna and scope approach going to give me a useful indication of how chattery the hardware is?

[Neilm]

First point - where in the world are you? Requirements vary. If this is a one off thing you are doing and are not going to be selling it don't worry - it will not fall under EMC regulations as far as I am aware. However, if it radiates too much the local enforcement could find you and take action. Exactly what action depends where you are.

If you do want to shield, there are some very fine meshes around that could be used but this will get a bit expensive.

[T3sl4co1l]

Emissions might vary from "not much" to "who knows".  I've heard it said that modern motherboards aren't so big a deal anymore, because the higher clock frequencies give out less radiation.  I don't know what nonsense that is, but who knows, in combination with better PCBs it might be true (multilayer construction, with many traces routed on inner layers, between ground planes).  Differential signaling (PCIe, HDMI, SATA, etc.) should be very quiet, but the signals between CPU, Northbridge and RAM are something like GTL or SSTL1.5 which are single ended and therefore prone to radiation.

What's worst is you need one hell of a spectrum analyzer to be able to determine this.  You usually have to check up to 20 times the highest clock frequency... 30-60GHz, as high as anything goes today.  Good luck renting one of those bastards with pocket change, let alone any of the sensors  or antennas needed to use with it!  (On the plus side, lab time in a proper EMC test facility would be the cheapest part, or they would perform the test entirely, without having to rent equipment.  You're still looking at on the order of $4k though!)

Preferably, yes, everything has to pass EMC.

Practically, if no one complains, no one is going to take action.

So the worst you might achieve is having shitty communication off your own equipment (e.g., stripping shielding from LCD panel comm lines = your display goes batshit when the hard drive cranks or something?) or others (e.g., disrupting nearby Wifi or Bluetooth).

The other thing is protection.  A shielded case carries ESD safely over the electronics within.  Without that, you might zap a USB connector when plugging in a device, and poof, now that channel's dead!  If you start connecting many things together within the enclosure, without having them sheltered by a common ground, ESD on one part could even cause indirect damage on another part -- the energy from the spark has a very short rise time, and tends to rattle around from point to point as it dissipates.  It will go up, through, and inside, any cables it can find.  Not as strong as a primary spark to a sensitive device, but it's only a matter of "strong enough" versus "sensitive enough".

Now, what you're after still isn't *impossible* -- just as TFT screens contain transparent coatings to work at all, you can find transparent "EMC glass" that's conductive, probably thanks to the exact same coatings.  It's not as transparent, and usually has a bit of a tint, but may help a lot.  Obviously, this needs to be joined continuously into the larger enclosure, like with that EMC spring stock gasket stuff and clamping.

And, as mentioned, there are screen or mesh materials that aren't too obtrusive.

To get the full benefits of shielding, remember all connectors penetrating the shield must be grounded to it.  That's why USB and Ethernet and other connectors always have little springy things on them.  Try to do the same, yourself.

Structurally, regarding the motherboard plate: consider the force needed to push against the heatsink during installation, or DIMMs or other cards and connectors.  You don't want warping of the board, because that risks cracking ceramic chip components (resistors, capacitors, ferrite beads).  I doubt the shielding is as important on the bottom side, or at least as long as you ultimately have a shielding window to the outside.  Instead of a plate of sheetmetal, you might get even better results, say, if you went with thick aluminum plate, machined into an open frame shape.

Tim

[calexanian]

EMI or RFI???

[tggzzz]

Metalised tape is principally useful for sealing gaps between metal plates. I suspect it would only be helpful if you use so much that you obscure components - which would rather diminish the point of having a transparent case!

Be aware that many (not all) transparent plastics are relatively brittle and can crack easily. Don't use them to support weight, and be careful if there is a danger someone might hit them hard.

[Kaithar]

First point - where in the world are you? Requirements vary. If this is a one off thing you are doing and are not going to be selling it don't worry - it will not fall under EMC regulations as far as I am aware. However, if it radiates too much the local enforcement could find you and take action. Exactly what action depends where you are.

If you do want to shield, there are some very fine meshes around that could be used but this will get a bit expensive.

Definitely not selling, it's a one off personal project.  I'm mainly concerned about interfering with my own stuff heh.  I'm in the UK, so it'll be CE based emissions standards iirc.

Emissions might vary from "not much" to "who knows".  I've heard it said that modern motherboards aren't so big a deal anymore, because the higher clock frequencies give out less radiation.  I don't know what nonsense that is, but who knows, in combination with better PCBs it might be true (multilayer construction, with many traces routed on inner layers, between ground planes).  Differential signaling (PCIe, HDMI, SATA, etc.) should be very quiet, but the signals between CPU, Northbridge and RAM are something like GTL or SSTL1.5 which are single ended and therefore prone to radiation.

What's worst is you need one hell of a spectrum analyzer to be able to determine this.  You usually have to check up to 20 times the highest clock frequency... 30-60GHz, as high as anything goes today.  Good luck renting one of those bastards with pocket change, let alone any of the sensors  or antennas needed to use with it!  (On the plus side, lab time in a proper EMC test facility would be the cheapest part, or they would perform the test entirely, without having to rent equipment.  You're still looking at on the order of $4k though!)

That's what I was afraid of, yes.  Motherboards are a mess of clocks and buses,  though on the plus side they have to withstand anything they put out heh.

So the worst you might achieve is having shitty communication off your own equipment (e.g., stripping shielding from LCD panel comm lines = your display goes batshit when the hard drive cranks or something?) or others (e.g., disrupting nearby Wifi or Bluetooth).

Heh, a return to the days of knowing your phone was about to ring because your speakers started screaming about it.  I hadn't considered that effect on the driver lines, but you're likely right that it will be less happy.  That's testable at least.

The other thing is protection.  A shielded case carries ESD safely over the electronics within.  Without that, you might zap a USB connector when plugging in a device, and poof, now that channel's dead!  If you start connecting many things together within the enclosure, without having them sheltered by a common ground, ESD on one part could even cause indirect damage on another part -- the energy from the spark has a very short rise time, and tends to rattle around from point to point as it dissipates.  It will go up, through, and inside, any cables it can find.  Not as strong as a primary spark to a sensitive device, but it's only a matter of "strong enough" versus "sensitive enough".

The hope would be that having it enclosed would reduce ESD risks, though being acrylic it's not going to be static dissipative and I can definitely see the benefit of more capacitance from a chunk of metal even without it being grounded.  The USB shield should at least be grounded at both sides.
Hmm. I'm definitely settling on the need to put in a grounded mounting plate, not like there's all that much interesting on the bottom of a motherboard and it would likely be obscured anyway.

Now, what you're after still isn't *impossible* -- just as TFT screens contain transparent coatings to work at all, you can find transparent "EMC glass" that's conductive, probably thanks to the exact same coatings.  It's not as transparent, and usually has a bit of a tint, but may help a lot.  Obviously, this needs to be joined continuously into the larger enclosure, like with that EMC spring stock gasket stuff and clamping.

And, as mentioned, there are screen or mesh materials that aren't too obtrusive.

Hmmm, that's a good point, though using the plastics that's buried mesh based adds some interesting problems to making a laser cut finger joint.
Would I be excessively optimistic in thinking that the kind of plastic used in static dissipative bags would be sufficiently conductive for EMC?  That's reasonably clear and would solve some of the protection issues on things like the TFT lines.  Google says they're typically less conductive than "conductive plastic".
There's some plastic manufacturers in the UK that claim to do EMC compliant instrument windows, so they might be worth nudging for prices.
Mesh would definitely work for some of the less visually interesting parts like the PSU.

Structurally, regarding the motherboard plate: consider the force needed to push against the heatsink during installation, or DIMMs or other cards and connectors.

"Low Insertion Force" my ass. 
My thinking is to rip a mount plate from an existing metal case, I have some junk ones lying around, and make sure it's properly connected to any other metal in the construction.

Instead of a plate of sheetmetal, you might get even better results, say, if you went with thick aluminum plate, machined into an open frame shape.

Hmmm, you mean as a mount plate still, right?  For some reason I have this image of a plate with a motherboard sized hole for the board to sit suspended in.

Metalised tape is principally useful for sealing gaps between metal plates. I suspect it would only be helpful if you use so much that you obscure components - which would rather diminish the point of having a transparent case!

Sadly, yes
It would at least protect things like the driver flexpcb that is probably painfully sensitive.

Be aware that many (not all) transparent plastics are relatively brittle and can crack easily. Don't use them to support weight, and be careful if there is a danger someone might hit them hard.

Assuming you mean more than normal, that would kind of make sense if it's metal as a pigment rather than a buried mesh design.  Enough conductive pigment to make the plastic usefully conductive would likely mess up the chemical structure.  Otherwise, yes, I am aware that they can be a little less than helpful... I'm aiming for 10mm transparent acrylic, should be safe enough for what I have in mind.

[T3sl4co1l]

Heh, a return to the days of knowing your phone was about to ring because your speakers started screaming about it.  I hadn't considered that effect on the driver lines, but you're likely right that it will be less happy.  That's testable at least.

Hey, those days are still around... GSM is just as powerful as ever, and home audio equipment is just as terrible at EMC as ever.  At least a few streams I've watched online, you hear the distinctive bip-dit-beep-bip of a GSM transmission every so often...

The hope would be that having it enclosed would reduce ESD risks, though being acrylic it's not going to be static dissipative and I can definitely see the benefit of more capacitance from a chunk of metal even without it being grounded.  The USB shield should at least be grounded at both sides.

If you have a USB port mounted in a plastic enclosure, then the only ground reference is the cable connecting it back to the motherboard.  So, zapping the port, even if it's on the metallic shield part of it, still carries zappy energy all the way back to the motherboard.  The cable is likely imperfect (the worst part being the transition from cable to header, where there's a full 1-2cm of unshielded connection), and that exposes the USB signals to ESD energy.  Maybe not all of it, because it's kind of a secondary thing (it's seeing what energy is not being shunted by the ground wire), but it's still going to be a large percentage (in the 10s of % range), enough to be a hazard.

If the port is mounted into a big metal shield, that helps spread the energy out, but very quickly, the plate itself becomes charged, and where does the energy go from there?  Well now you have a huge antenna that's inducing energy in everything around it.  Even worse!

ESD testing is typically performed with a signal source on the order of 150pF, so it takes a big plate (i.e., about the capacitance of a human being in free space, oddly enough!) to have any effect on the amplitude.

What's more, ESD testing is performed in two phases: one to check direct effect (zapping connector shields, any exposed metal), and another to check indirect effect: a metal plate is set up near the device and zapped.  Even this non-contact method produces considerable interference, enough to upset communications and cause failures of poorly shielded, sensitive ports (maybe not USB, but cables within the case, like SATA, sure).  The equivalent "real life" scenario is, touching something metallic nearby, like a filing cabinet; the mini EMP from the spark couples into everything nearby.

What should the shielding look like?

Consider standing on the beach.  Your goal is to block a huge wave incoming.

You set up a flat board, vertical, in front of you.  Braced so it doesn't push away.  The wave comes.  It washes right around.  Well, duh, that was pretty stupid, huh?

The physics of E&M aren't quite like water waves (quite a lot simpler, actually... E&M doesn't make turbulence!), but to a general idea, it still works out.

Suppose you get a few more boards, and stick them together, so that you now have a "C" shape around you.  The wave comes, it crashes over the nose of your structure, and goes around it.  The wave eventually washes in behind, but it's much less violent than before.

Add one more board and complete a square.  Now you have a sealed column.  The wave comes, it crashes around the column.  Nothing happens, the water level inside remains constant (or it changes slowly, due to water seeping in through the sand or something).  This is the ideal case, having solid metal shielding all around.

If you must have windows, the best place to put them will be away from the crashing waves.  If the sides of your enclosure are insulated, ESD can't strike there.  If the front, back, top and bottom are shielded, and the ports are only on the front and back (or whatever), the waves will tend to wash over those surfaces.  They'll still wrap around the openings, but it won't be as strong as having a floating plate or something.

You'll still have no contest as far as outgoing radiation or incoming radio susceptibility (a window is a window, at those frequencies), but it's one thing at least.

Hmmm, that's a good point, though using the plastics that's buried mesh based adds some interesting problems to making a laser cut finger joint.
Would I be excessively optimistic in thinking that the kind of plastic used in static dissipative bags would be sufficiently conductive for EMC?  That's reasonably clear and would solve some of the protection issues on things like the TFT lines.  Google says they're typically less conductive than "conductive plastic".

Yeah no, you really need something metallized to do the job..

Hmmm, you mean as a mount plate still, right?  For some reason I have this image of a plate with a motherboard sized hole for the board to sit suspended in.

Nah, I mean, a plate that's more of a web, with nodes at all the screw hole locations.  Could do some artsy forms -- 'bubbles' missing from the plate, or a literal spiderweb style, or straight hard lines, or something more abstract, etc.

Tim