Originally posted by: Stiganator
I need to run an experiment that involves measuring very small signals (pV-mV). The space prevents me from creating a big cage. I'm wondering a couple of things.
How much does the mesh size matter? Is a piece of metal ideal or is a mesh of ~100/inch sufficient? I'm assuming copper will be the best bet.
What about imperfections? If I can only cover 5 sides with just a few holes for cords will I still see a large reduction in noise or does it need to be fully enclosed?
The previous response had things mostly right, though for one thing the general rule is that if the hole size is much *LESS* than a wavelength (e.g. 1/40th wavelength) then it'll be very effective at blocking EM fields of that wavelength.
Really it all depends on how quickly in space (past the hole) you want the E/M fields to be attenuated, and how high the attenuation loss must be. The smaller the hole the less energy passes through it and thus the smaller the re-radiated field on the other side becomes the more quickly with distance from the hole.
Any hole, seam, discontinuity, slot, etc. in the enclosure acts as an antenna source of radiation and it'll be radiating a new E/M field beyond it due to its scattering of the incident radiation and its resonance with the incident E/M fields.
A solid piece of metal is a more ideal shield than a uniformly perforated one, though a uniformly perforated one with holes well less than a wavelength will be OK in many cases especially if the shield is very large and distant relative to the sensitive receiving nodes inside the shield.
Generally faraday cages of non-magnetic metals (e.g. copper, aluminium) are not substantially effective at reducing the flux of static or low frequency magnetic fields at all! To shield magnetic field fluctuations you'll want a good thick shield of some ferrous type of metal like Iron or at least Mu Metal so the magnetic flux will mostly travel through the high permeability shield material and little will remain in the space interior to the shield.
A seam / slot in a shield even if that seam is soldered or pressed together can dramatically worsen the performance of a Faraday cage. If it interrupts current paths (e.g. it is at right angles to the current direction) then it'll act as a large impedance discontinuity at high frequencies and become essentially a radiating antenna.
If it is oriented in parallel to the direction of current flow, however, it will have insignificant effect on the shielding quality even if it is relatively long. Of course for incident fields of random orientations, wide band frequencies you don't know what direction the current will travel in.
Many times some of the worst noise sources of low level measurements, however, are not from E/M fields external to the apparatus, but are due to the noise of the apparatus itself -- e.g. ripple on the power supply lines being converted into noise in the measured signal, crosstalk between high level signals in the apparatus and low level signals, et. al.
Also keep in mind for AC fields of moderate frequencies there's a factor known as "skin depth" that relates to the conductivity of the conductor and the frequency of current. It takes several times one skin depth distance into the interior of the conductor before an AC signal flowing on one side of the conductor diminishes significantly inside the depth of the conductor. So ideally for noise sources in the 10's of kHz up to infinity frequency range your shield thickness and conductivity will be chosen to be at least 5-10 skin depths thick at the noise frequencies of interest. A couple of inches of thickness of copper shield wall is pretty good against even very low frequency signals. However a copper foil of 1/1000" thickness will be at best of modest usefulness against a long wave noise signal in the AM broadcast band, for instance.
In copper, the skin depth at various frequencies is shown below.
60 Hz 8.57 mm
10 kHz 0.66 mm
100 kHz 0.21 mm
1 MHz 66 µm
10 MHz 21 µm
Generally proper engineering of your enclosed circuit and wiring can help to minimize its responsivity to undesired fields which are present in its environment. Proper use of well balanced twisted pair differential circuits, coaxial/twinaxial cables, keeping any sensitive wiring very SHORT, reducing circuit loop enclosed areas to minimize B/H field pickup, reducing curcuit wire lengths to reduce E field pickup, keeping impedances very low if possible, moving sensitive nodes far from noise sources, using distinct low noise power/ground networks, using proper filtration to minimize signal response at unnecessarily low/high frequencies, et. al. all may help a great deal.