How to build a Faraday cage to keep a WiFi signal inside?

[bttd]

I'm working on a wifi device testing box and made the first prototype of the Faraday cage. I need holes because of cooling. For this reason, the box has 3mm diameter holes on the top. The box made from steel, and have 2mm wall thickness.

The wavelength around 12,5 cm for 2,4 GHz and 6 cm for 5Ghz. The 3mm holes need to be small enough to block the signal (smaller than 1/10 of the wavelength). But the cage does not block wifi, but even GSM signals. The signal strength going lower but not disappear.

Is there anything I'm missing? I also tried to gound the box, but it won't help. The 2mm thick steel not enough?

 

[anorlunda]

[Moderator: Even though this is school work, it it not homework. I hope that it will get better answers in the EE forum.]

 

[berkeman]

@bttd -- Welcome to the PF.

There are a number of things that go into making a good RF shielded enclosure with good RF attenuation between the interior and exterior. All feedthroughs (signal, power, etc.) generally need to be filtered, and cooling air needs to flow through baffles. You also need to pay close attention to the seams in the enclosure -- closed seams need to be welded or soldered, and door seams need RF gasketing to make good contact.

Can you post a few pictures of your enclosure so far? What-all do you have for feed-throughs?

 

[berkeman]

Here is a good baseline for RF shielded enclosures:

https://jretest.com/product/jre-2525/

and this: http://www.ets-lindgren.com/products/test-enclosures

 

[eq1]

But the cage does not block wifi, but even GSM signals. The signal strength going lower but not disappear.
Is there anything I'm missing?

Faraday Cages are high pass filters. They can attenuate a signal but not eliminate it entirely so I think your setup is probably working.

Your goal should be to attenuate the signal so much that the hardware can no longer detect it. This can be hard to do as that hardware is specifically designed to pick up very weak signals and it's very good at it.

If the box must have holes (you can't use solid walls and make them thick) probably you're going to have to do something like put a cage in a cage in a cage and make sure the holes are offset in space.

If the holes are for cooling maybe you could just put the device in a fridge. :)

 

[sophiecentaur]

The 2mm thick steel not enough?

The thickness of the metal is the least of your worries. Every joint or seam is a possible way in or out for the signals. If you could weld the seams it might help. If you can screw the door / lid down or use spring fingers around the join. If you can filter every input / output, including the power. etc. etc. Remember that there will always be a 'weak link' so you need to work on a very broad front, attacking all the points I have mentioned in turn and note which one makes most difference.
You need to specify what sort of performance you are actually after. No design can ever exclude everything.

To avoid the leaking inputs and outputs, you may consider operating everything inside and use batteries - then screw the lid down every few cm.

 

[qutron]

Why re-invent the wheel?
A microwave oven operates near to 2.4 GHz and you can watch the food as it cooks through the perforated metal door window. Human eyes are very sensitive to microwave energy and the high power of the microwaves must be reduced to near nothing. We are talking about from 700 watts up to 1.5KW of microwave energy for home units.
So as a test just place your wifi receiver inside a microwave oven and see if the oven cavity will attenuate the wifi signal.
Note how small the holes are in the microwave's perforated metal door, you may have to use many similar small holes like that for your required cooling.
5 GHz may require even smaller diameter holes.
Needles to say, Do not turn on the microwave.

 

[Asymptotic]

So as a test just place your wifi receiver inside a microwave oven and see if the oven cavity will attenuate the wifi signal.

Worth trying. This article and Physics Girl video is on-point and, well, interesting.
https://interestingengineering.com/...u-know-if-your-microwave-has-a-radiation-leak

Maximum allowable microwave leakage is 5 mW/cm2, and (based on a cursory Google search) is typically 0.2 mW or less. For an oven power of 1500W this works out to -54.8 dBm to -68.8 dBm attenuation.

The following resource gives wifi signal strengths of -67 dBm to -70 dBm (for high speed streaming to basic connectivity), -80 dBm (unusable, but still connecting), and -90 dBm as a total wipe-out.
https://www.metageek.com/training/resources/wifi-signal-strength-basics.html

I need holes because of cooling.

Another angle on the general idea in post #5 is seal the RF enclosure as completely as practical (no cooling holes), and mount a couple of thermoelectric cooler modules on the outside. Interior air could be controlled at a constant temperature by adding suitable temperature sensing and TEC power control circuitry.

 

[sophiecentaur]

A microwave oven operates near to 2.4 GHz and you can watch the food as it cooks through the perforated metal door window.

The door seal on a microwave oven is very clever and it works by using two slots which are λ/4 deep. It produces a very good 'short' across the gap but only at its design frequency. Probably good enough near 2.4GHz but no good at 5GHz. That gap would be very leaky at most frequencies, I'm afraid. Worth a try - just to prove me wrong perhaps.

 

[DaveE]

As others have said, it is the seams, corners, doors, etc. that matter most. You want to think in terms of creating a very low impedance path to allow circulating currents to cancel the EM field that is trying to exit. So if the electrons can't circulate easily in any particular orientation, then the field that would otherwise induce that particular current won't be attenuated well. It isn't good enough to have a bunch of metal walls, they MUST be connected to allow electrons to move around the enclosed surface freely in all directions.
As far as holes in the enclosure, you can search for information about "wave-guide below cut-off frequency" to understand how hole size, material thickness and frequency interact. Basically small holes are ok, big ones aren't. For example: https://www.evaluationengineering.c...85/shielding-a-look-at-waveguide-penetrations. There are many other sites that also talk about this.

 

[Anand Sivaram]

Why re-invent the wheel?
A microwave oven operates near to 2.4 GHz and you can watch the food as it cooks through the perforated metal door window. Human eyes are very sensitive to microwave energy and the high power of the microwaves must be reduced to near nothing. We are talking about from 700 watts up to 1.5KW of microwave energy for home units.
So as a test just place your wifi receiver inside a microwave oven and see if the oven cavity will attenuate the wifi signal.
Note how small the holes are in the microwave's perforated metal door, you may have to use many similar small holes like that for your required cooling.
5 GHz may require even smaller diameter holes.
Needles to say, Do not turn on the microwave.

I just did a crude experiment. My 802.11n, 2.4GHz Accesspoint is around 2 meters away directly from the microwave oven but one brick wall between them, that is between the drawing room and the kitchen.

I tried with my Moto G5 Android mobile with a WiFi analyzer App. I could see the mobile screen though the mesh door of the microwave oven.

Observation:
Just outside Microwave : -45 to -50 dBm
Inside microwave: -70 to -75 dBm

 

[sophiecentaur]

if 25dB isolation is good enough then it's good enough. I would hope that the isolation is a bit better at the oven Magnetron frequency! Like I said earlier, the door seal is for one specific frequency. If you had to unscrew the oven door to take out your noodles, they would be cold before you could eat them.
But I am surprised the screening is as narrow band as the 50MHz in 2.4GHz would suggest. Interesting measurement though.

 

[sophiecentaur]

I did a quick Google search about leakage from microwave ovens and, among all the waffle, I could see nothing better than 30dB of screening seems to be available. That corresponds to 1W escaping for a 1kW oven. If that were concentrated on an eyeball or a bit of the brain, I guess it could be relevant but you'd have to try hard.
However, the results of @qutron are pretty much in agreement, which is always pleasing - even it it means that a microwave oven is no good as an aid to measuring low signal levels.

 

[Anand Sivaram]

I did a quick Google search about leakage from microwave ovens and, among all the waffle, I could see nothing better than 30dB of screening seems to be available. That corresponds to 1W escaping for a 1kW oven.

That is 10 time more power than a typical WiFi access point.

 

[bobob]

Why not just buy some copper screen and make an enclosure from that? Then, make sure you have it solidly tied to (earth) ground. That's the standard way of making a faraday cage and it solves your cooling issues. If you don't have it solidly grounded, you'll just make an antenna.

 

[sophiecentaur]

That is 10 time more power than a typical WiFi access point.

I wouldn't be surprised if a WiFi unit was affected by operating it bang next to a running oven. More than a metre or so away would even the levels up enough for the WiFi to operate ok.

 

[essenmein]

I'm working on a wifi device testing box and made the first prototype of the Faraday cage. I need holes because of cooling. For this reason, the box has 3mm diameter holes on the top. The box made from steel, and have 2mm wall thickness.

The wavelength around 12,5 cm for 2,4 GHz and 6 cm for 5Ghz. The 3mm holes need to be small enough to block the signal (smaller than 1/10 of the wavelength). But the cage does not block wifi, but even GSM signals. The signal strength going lower but not disappear.

Is there anything I'm missing? I also tried to gound the box, but it won't help. The 2mm thick steel not enough?

Could be that the steel is ferro magnetic, so while RF hits it, due to large self inductance presents too high an impedance to attenuate incident fields.

I'm not 100% certain this is the case, but the two diy Faraday cages we have here for EMI work are all non magnetic materials (Al, Cu).

Edit:

Because science just ran down to see if our cage blocks my phone (LTE) and I go from full bars to one bar and can still receive calls. lol.

Mesh is ~1.5mm square, Alum wire, alum sheet bottom, copper tape on edges, capacitive pass throughs for power and data.

 

[berkeman]

Unfortunately the OP @bttd has not been back to the PF since they posted their question. Hopefully they will drop back in and read the good advice that is being given.

As a follow-up to my post #4 where I showed typical RF shielded enclosures that I work with, I should have mentioned that it takes two of them to perform good, accurate link budget testing at WiFi frequencies. We use two of these enclosures connected by coax with variable RF attenuation between them to see how good our link budget is with our WiFi products (how much RF attenuation does it take before your RSSI and SNR are unacceptably low).

We found we could not just use one box with a device inside and another device outside the box, because the one box setup only provides about 60-70dB of attenuation. Our devices work down to around -85dB, so the second box to house the second device is necessary.

 

[Anand Sivaram]

I wouldn't be surprised if a WiFi unit was affected by operating it bang next to a running oven. More than a metre or so away would even the levels up enough for the WiFi to operate ok.

It used to be more of a problem during 802.11b times because of DSSS. I heard that 11b WiFi used to get disconnected when microwave oven is used. Now, with 802.11n etc. it uses OFDM-MIMO which has much better noise resilience.

 

[sophiecentaur]

Mesh is ~1.5mm square, Alum wire, alum sheet bottom

A mesh of crossed aluminium wires could have a significant resistance compared with the same thickness of sheet with a punched hole mesh. The surface of aluminium oxidises as soon as its in the air so the contacts (non linear of course) could be poor. Notice the mesh in the microwave oven door is holes in a sheet.

 

[darth boozer]

https://www.physicsforums.com/threads/microwave-faraday-cage.650830/