First off, distinguishing between noise (thermal effect in the electronics) and interference (something coming from outside, even if it's noise-like in distribution) the fact that you were able to shield it implies we're talking about the latter. You wouldn't be able to shield out thermal noise inherent in the system.

So, if you're planning to use an optical microscope while you're probing, meaning that you (or a person) will be physically observing the process, but no additional electronics will be added, it should be possible to use a shielded room to isolate from outside interference just like you did with your previous tests.

On the other hand, if you're adding other sources of interference to the measurement system (e.g. an electronic microscope, video camera, etc.) or environment (e.g. leaving your cell phone in your pocket while you look through the microscope) then it won't matter how much you shield from the outside, since you need to eliminate the local interference you're introducing. There are ways to provide some shielding with conductive glass (wire grid or thin conductive coating) so you could potentially encapsulate something under the microscope, but you also have to isolate and filter any conductor penetrating the shield to prevent carrying signals through the shield. I would assume you're needing to manipulate probes at the device you're characterizing, so I' suspect enclosing the device is not an option anyway.

Thank you Michael D. Foegelle ! You made me wonder that in the case of thermal noise, would we have an increased amount of noise in the system if we increase the working temperature?

How would a metal box with a ITO coated glass (as the conductive window) do? we ground the box, have an illumination window, and use coax cables for the readout?

I have seen people use mesh aluminum sheets to make their Faraday cage. Do they really work? If so, I can illuminate my sample through a hole.

And please take a look at this clip:

https://www.youtube.com/watch?v=KQyvyg30S1Q&feature=emb_logo

Its totally open from two sides!

Hi Hanie,

So what he created in the video isn't really a Faraday cage, since as you say, it's still wide open at the ends. Or is it? You'll note that the interference went way down as soon as he connected the ground wire, which should tell you something. In fact, in a true Faraday cage there would be no need to ground the device under test (DUT) at all. The fully closed shield prevents the fields from reaching the DUT. So why did the "tube" still have an effect even though it was open on the ends?

Another important part to this equation is the frequencies we're dealing with, and in that respect you could consider this a cage in that a wave is highly attenuated when it tries to go through an opening that's smaller than half a wavelength. This is what we refer to as the cutoff frequency and where we get shield penetrations referred to as "waveguide below cutoff" that will work fine so long as you don't try to run a wire through them. Water, air, fiber optics, etc. are ok, but as soon as you run a wire it's now a coaxial line and that wave can ride just fine between a center and outer conductor (i.e. your cable). So when he closes the rectangular cavity he's created, it's really a waveguide that the wavelength can't travel down easily.

If you look at the audio spectrum I attached, on the left is before closing the waveguide, and the right is after. All of the frequencies we can hear top out at a few kilohertz. At 10 kHz, the RF wavelength is 30 km, so a LOT larger than the size of the tube there, and most of the frequency content is lower than that. If you toss his waveguide dimensions of 5.5x2.5" and 10 kHz into a waveguide attenuation calculator you'll find it's about 60 dB/ft (assuming aluminum as the worst conductor in the list), so after an inch or two, the signal has dropped drastically, which is what you see happen when he closes it.

However, the real magic happens when he attaches the ground lead. Any EMC engineer worth their salt will tell you, "there's no such thing as ground," the way people think of it as a zero DC potential. At higher frequencies, that ground wire wouldn't be doing anything. However, in this case, the frequency is so low that what he's really doing is changing the coupling mechanism to the internal PCB by ensuring that all of the currents being induced by the external EMI are not coupling into the circuit that is amplified to drive the speaker.

So, back to your problem. Why/how is the interfering signal coupling to your DUT? If it's a low frequency signal, then proper grounding of your DUT and probe and avoiding magnetic loops that can couple interference into your measurement should eliminate the problem. However, if we're talking about signals where the wavelength is getting relatively small compared to the geometry, then there's no such thing as ground and you have to find another approach to block the EMI.

Hope that helps,

Michael

Hi Michael D. Foegelle , thank you SO much for the detailed answers!

Do you think if we submerge the electronic parts in liquid or better to say, have 'salty water bath' on top, would help with cancelation of the interference?

If so, then I can still use the illumination from the microscope while not having to deal with a perfect Faraday cage.

Hi Hanie,

I'd think the contamination of circuitry due to the immersion would be a problem, and chances are that approach would also alter your attempted measurement.

So in that vein, can you provide more information on your typical measurement setup and process? E.g. what test equipment are you using, what is the actual measured quantity or curve, what's the actual DUT, how are you hooking things up, etc.

Thanks,

Michael

Hi Michael D. Foegelle ,

Since the current is low, I used an amplifier and a source measure unit to measure the output current. The graphs are from the photodiode in a typical enclosed Faraday cage and dry. However, if I want to add illumination or simultaneous readout from the microscope, I don't know how things might work. By the end, the PD will have to work in an aqueous solution.

Thanks!

So the measurement you're performing is just a DC measurement and you're measuring over a long time period, correct? Generally this is just a common electromagnetic compatibility (EMC) problem and it's typically better to try to figure out how the interference is coupling into your measurement and prevent that rather than just trying a brute force mechanism (e.g. shielding) to keep it out. Given the time scales on your plots, it would think it would be possible to just apply some low pass filtering to eliminate any noise that might be coupling in. However, grounding/bonding your work-space and circuit to the test equipment is important to avoid picking signals up on ground loops. If you don't couple the signal in, you don't need to filter it out. At any rate, unless the measurement is a lot more complicated than it would appear in your post, I'd think you could solve the problem without resorting to a complicated and messy immersion bath (that probably won't help much anyway).

Thank you Michael D. Foegelle !