Assuming I need to put my project in a plastic enclosure because its outside may come into contact with other electronics, what would be the best way to shield the enclosure?

One of the solutions I'm considering is a double- or triple-coat spray of MG #841. Other options include lining the inside of the enclosure with copper or aluminum foil. Which one would be most likely to reduce external interference?

Also, whichever of those would work best, would they work better (or worse) if I tied the shielding to ground? For grounding the spray coating, I would glue a #2-56 nut to the inside of the case prior to spraying and cover the top (so the spray can make contact with the sides of the nut), then fasten a wire lug that is attached to the circuit's ground. Grounding the foil would be a good deal simpler, I could just solder a wire directly to the foil.

I realize my shielding ideas are probably overkill, but I'm just curious what method would be best, on a theoretical level. Also, if the spray method would be better than either foil but only marginally so, that would be good to know (in which case I'd probably go with the foil due to ease of application/installation, as well as cost).

 

in the old days, we used to shield the cases of our TRS80 computers with a coat of silver or copper spray on the inside of the case. it helped, but to get the best effect you have to get the rf off all wires and connections coming into the shielded case. otherwise, you just have a bunch of antennas radiating noise in and out of the shielded box.

 

The general rule is, the thicker the shield the better. How thick is good enough depends upon the level of the EMI interference. The best would be a metal box, of course.

The shield should be connected to ground for best shielding. A floating shield can act as a radiator.

You may need to shield the wires into and out of the box also.

 

Do you even need shielding in the first place? Thats the real question..

Personally "IF" I found out that I needed to shield a product a metallic enclosure/proper grounding would be my first choice..

 

Rather than shield a non-conductive box, another approach is to insulate a metal box. We have a MIL-rugged product line that is all aluminum. We send the outside parts out to be Rhino-coated - yup, the stuff they spray on to pickup truck beds. Works.

When it comes to designing a high-shield system, the wires are by far the biggest pain. About the wires, will you have room inside for clamp-on ferrite noise suppressors?

ak

 

This is going to be a small USB dongle, so the only "wires" going into the case are the USB connector and the PCB it's attached to. What would be the safest way to shield those (without interfering with the USB signal)? It's not a Hi-Speed device, so the maximum it will EVER potentially communicate at is 12mbps.

 

For some strange reason Altoid Mint containers seem to be a popular project box.

 

This is going to be a small USB dongle, so the only "wires" going into the case are the USB connector and the PCB it's attached to. What would be the safest way to shield those (without interfering with the USB signal)? It's not a Hi-Speed device, so the maximum it will EVER potentially communicate at is 12mbps.

And you are interested in shielding because you fear stray signals will interfere with your U?sB device, or, you fear the USB device will interfere with other devices?

What frequency, Peak-to-peak voltage and current will you be using? What part of your device could/would act like an antenna? How long is it?

 

I'm looking at building a variant on the XR232 USB for installation inside a computer. Tests by Julien (the designer) showed that the device was somewhat susceptible to EMI, so I'm basically just going for overkill on the shielding (mostly "because I can"). As you can imagine, I'm not keen on having a metal box floating around inside a 1U case. Wwell ideally it shouldn't be "floating around", but I'd really rather not run the risk of it getting jarred and shorting something inside the computer.

 

The device looks like a type of free running relaxation oscillator noise generator so good shielding is needed if you want the results to be 'secure'.

Frequency injection attacks:
http://www.cl.cam.ac.uk/~atm26/papers/markettos-ches2009-inject-trng.pdf

RNG quality: Search for XR232USB
http://www.cacert.at/cgi-bin/rngresults

 

Nice links!!! I've done some TEMPEST designs. Now I have a nice way to explain why.

ak

 

So what defines "good shielding"?

I'm not too concerned about voltage frequency injection attacks, because that would mean my server's hardware has been compromised -- in which case I've got bigger problems than someone fiddling with my RNG device. My main concern is generic environmental noise reducing the level of entropy.

 

So what defines "good shielding"?

I'm not too concerned about voltage frequency injection attacks, because that would mean my server's hardware has been compromised -- in which case I've got bigger problems than someone fiddling with my RNG device. My main concern is generic environmental noise reducing the level of entropy.

IMO most of the examples (simple metal case shields) already given should work fine unless you have something really strange inside the server.

The best way to approach the problem of environmental EMI in a fixed location is the eliminate the source of the EMI or to reduce it to the lowest level practical at it's point of origin first by absorption or shielding then distance the sensitive device away from the source and add shielding to it. Good shielding doesn't eliminate the energy contained in EMI, it transforms the energy into equal and opposite energy to neutralize it so good conductance (remember skin effect for RF shielding) at the target frequency range is a must combined with good grounding to the coupler (less effective >1ghz) and proper apertures or joints for signals and power.

http://www.ti.com/sc/docs/apps/msp/intrface/usb/emitest.pdf
http://ieee.rackoneup.net/rrvs/09/Engineering%20Aspects%20of%20Electromagnetic%20Shielding.pdf

 

Nice links!!! I've done some TEMPEST designs. Now I have a nice way to explain why.

ak

TEMPEST is old school, Emissions Security is the game today.

 

if I were trying to get rid of some emi/rfi, I would make the shielding from what would be proper. at power and audio frequencies soft iron and some special alloys for magnetic shielding. for higher frequencies, copper and aluminum work best. there used to be a brand of radios with excelent shielding, the chassis and all interior shields were coper plated iron. and back in the vey old days, radio stations had to have a monitor on the 600 meter, 500 KHz maritime emergency frequency. just try filtering a 10 kw station out of a sensative reciever just below the station frequency located at the station.

 

So I found a very small aluminum enclosure that I can probably put in the expansion card bay (remember, this is a 1U case -- I don't have a lot of room to work with)... I'd probably end up gluing it to a piece of plastic that will slip into the PCI-E slot and the guide rails so it's physically supported; I'd have to attach it to the internal USB port via cable though. This provides an interesting opportunity for filtering though, I think.

I'm thinking I will use a micro-USB connector with this case, and a cable like this one that has a choke outside of the enclosure, that should hopefully minimize noise on the cable. If I soldered the housing of the micro-USB connector to the aluminum enclosure's panel, I think that would keep everything well-grounded, and have a minimum of openings for any EM/RF leakage.

The XR232USB also has an R-C filter immediately before Vcc on the LM393, and a small choke on the USB V+ line, so I'm thinking that should be pretty adequate to eliminate noise from the power source that could alter output.

Thoughts?

 

Oh, I'm also thinking of using a 4-layer PCB with ground and power in the center; will this help with EMI reduction to any significant degree? Of course 4-layer is more expensive than two layer, but if the protection offered by doing this increases measurably, I would at least consider the extra cost.

 

Oh, I'm also thinking of using a 4-layer PCB with ground and power in the center; will this help with EMI reduction to any significant degree? Of course 4-layer is more expensive than two layer, but if the protection offered by doing this increases measurably, I would at least consider the extra cost.

You might want to design a shield attachment ring for the noise generator chip on the board so a small diy shield can completely surround it like a 100MS.

http://www.ti.com/lit/ds/sbfs013/sbfs013.pdf

 

Question: As I'm starting to put together the parts, I realized that the "choke" shown in the circuit appears to be a two-legged beast, but there seem to be no specifications for it anywhere in the diagram, nor in the parts listing in the PDF (see attached). Searching "choke" on parts stores such as Mouser/etc frequently turn up four-lead varieties.

It does say "ferrite choke" -- so is this basically a piece of wire through a ferrite bead? Would any particular size/shape of bead do better at noise filtering than another (within reason) in this circuit? Could I just use one winding from a four-lead choke and leave the other two disconnected, or connect all four as a common mode noise choke (i.e. one winding from +V to circuit, and one winding from circuit to -V)?

 

http://www.ti.com/sc/docs/apps/msp/intrface/usb/emitest.pdf
http://ieee.rackoneup.net/rrvs/09/Engineering Aspects of Electromagnetic Shielding.pdf

Keep 'em coming, big dog.

ak