Low Cost Magnetic + Electrical Shielding Enclosure

[coppercone2]

Lol, this could have something to do with the connector type. there is a expensive special non magnetic special BNC connector. NO nickel diffusion barrier, some kind of other plating method, perhaps other measures against diffusion. Might be as simple as superfinishing or something to reduce solid state diffusion, or just left as pure copper or brass on the surface (not silver)

However, normal BNC connector is nickel plated, or uses a nickel diffusion layer, or possibly might even be made with clad steel, or nickel silver or some other semimagnetic alloy.


ALso, same is true for the coax itself. Alot of coax is clad steel middle, because it gets weak if its all non ferrous.. notice that in a coaxial cable there is nothing to grip?


Not sure how it effects the analysis, but literarly anything can happen to the magnetic properties of the can depending on what hardware its put together with, and what it plugs into.


I noticed in the lab, GOOD LUCK trying to get someone to find the 'special' cable to plug something in. If it fits it sits. Especially if they have expensive stuff and its geometrically correct.


So I can see a editor of the document possibly go into 'quarantine' mode after considering what to do with the magnetic shield of the container.  Asked the lab maybe what they think the most reproducible result from using random material BNC and cable?

[EC8010]

All that the LT diagram means is that you should insulate the BNC shell from chassis on the output side to avoid earth loops. You can either buy BNC chassis connectors made like that, or use standard types plus insulating kit. Insulated BNCs are easier to use on biscuit tins as the insulating kits presume a thicker "chassis" than a biscuit tin. That note about sanding the tin to get electrical contact with the lid is nonsense; solder a wire between the two for a reliable low resistance bond.

A mumetal box big enough to contain two "D" cells would be very expensive and probably not very effective. For mumetal to shield against magnetic fields it has to divert flux, so the ratio between its thickness and internal dimensions becomes important. A thick wall and small internal dimensions is far more effective than a big box with the same wall thickness. Thus, for effective magnetic shielding, you want small electronics inside a box only just big enough. There is no need to magnetically shield the batteries. I have 12V lead-acid batteries inside Kit Kat tins for electrostatic screening and power goes to the electronics over BNCs, leaving everything electrostatically screened.

[coppercone2]

its not nonsense.. its a option.


They could have specified double RF finger gasket if they felt like it.

I had anamoulous errors from (KS) test equipment at work that seemed like were caused by co-workers new phone. That's when I started to start ignoring people that always downplay shielding practices

Granted, if its poorly fitting, the wire might be better then sanding. But as I noticed on the welding table, simple mechanical fitup, even with steel, can sometimes make a fine ground, so long its cleanish and there is some pressure/tightness. For me, a wire is to help it with high current after its damaged (ground safety) much more so then for anything related to shields. This can be a requirement for panels.

Tins are often made with very high end dies to tight tolerances, so sanding them can have good results, so especially if you are gentle. They often even nickel plate them before painting too (usually a thinner nickel layer then you would want for engineering purposes)


The biggest problem with tins is that they are not designed for connectors, so unless you add a brace, they often deform substantially, especially with cables hanging off them, if you get extreme, if you see that, then you might wanna add a soldered wire, but its better to reinforce and maintain the dimensions of the precise box. Its unlikely you will see a problem from soft test leads, but something like a thick coaxial cable can cause the side of a small tin to warp when it bends, or if it hangs from equipment (like a DIY splitter, amplifier, filter, etc that you attach directly to a test equipment... )


You can simply cut a piece of thicker metal so it fits inside of the box, with holes for the connector. then when you bolt on your connectors, the entire wall is stiffened by a plate. I use 3/32 aluminum flatbar to good results here, brass would be better.




But I totally prefer pomona shield boxes, they are great quality, fit together tight, look professional, very stiff... if you have the money its worth it over a tin in many cases, particularly for passive stuff that does not need batteries. Oh but they aint steel ~!

[AnalogTodd]

Therefore, I'm confused on how to actually wire the enclosure. . Would it be done as pictured in diagram 1, diagram 2, diagram 3, or some other way?


(Attachment Link)

We found diagram 2 provided the best immunity to ground loops and the issues they can cause in such high gain circuits. We weren't spending huge amounts on specialized connectors or gaskets, but instead trying to figure out what worked. Most RF was beyond the frequency range we were looking at so we didn't find a need to go in depth on RF shielding. The lower frequency signals that were in the lab showed themselves to be a MUCH bigger issue, such as the 20kHz signal from the nearby oscilloscope. That was why we didn't go for aluminum or copper enclosures--they work great for high frequency but are poor when it comes to keeping low frequency signals at bay.

[Overspeed]

hmmm this is going to sound crazy
but i saw a cast iron pot on amazon US$24
6.8kg

https://www.amazon.com/AmazonBasics-Pre-Seasoned-Dutch-Handles-7-Quart/dp/B074DF92HW?th=1

from the weight estimation, it looks like 4mm to 5mm thick
what do you all think ?
there is also larger sizes

(ok i laughed at myself when i saw it 
because i remember this discussion here)

and then under recommendations
there is a 1 for bread which has a very nice shape
looks like it could fit a eurocard inside !

4mm cast iron
anybody want to hazard a guess the shielding capability ? 20 ? 30dB ?
** output estimate from grok = 22 to 35dB (ur = 200 to 1000) 12cm dia sphere with 5mm thickness
i guess then a smaller pot may have a higher shielding effect

*** and a smaller confit pot that fits inside the larger
grok estimate = 28dB upwards (ur = 200)
US$24 + 20. its crazy. is this a good price for near 50dB result ?
the lids are probably wonky as hell

Hello

As low carbon ( SS304) quite cheap that pressure cooker , you can even two as russian dolls ( remove the handle on the smaller ) and keep a gap contant  by using stainless steel sponge

available in Aluminium or SS304 , can be purchase cheap on flea market

Lid have a hole for pressure valve so easy to pass through

Regards
OS

[LooseJunkHater]

A mumetal box big enough to contain two "D" cells would be very expensive and probably not very effective. Thus, for effective magnetic shielding, you want small electronics inside a box only just big enough. There is no need to magnetically shield the batteries. I have 12V lead-acid batteries inside Kit Kat tins for electrostatic screening and power goes to the electronics over BNCs, leaving everything electrostatically screened.

This is something I had actually been thinking about, so I'm glad you clarified this, thanks!

[LooseJunkHater]

Therefore, I'm confused on how to actually wire the enclosure. . Would it be done as pictured in diagram 1, diagram 2, diagram 3, or some other way?


(Attachment Link)

We found diagram 2 provided the best immunity to ground loops and the issues they can cause in such high gain circuits. We weren't spending huge amounts on specialized connectors or gaskets, but instead trying to figure out what worked. Most RF was beyond the frequency range we were looking at so we didn't find a need to go in depth on RF shielding. The lower frequency signals that were in the lab showed themselves to be a MUCH bigger issue, such as the 20kHz signal from the nearby oscilloscope. That was why we didn't go for aluminum or copper enclosures--they work great for high frequency but are poor when it comes to keeping low frequency signals at bay.

Perfect, thanks! I guess I'll be building the enclosure illustrated in image two.

[EC8010]

Tins are often made with very high end dies to tight tolerances, so sanding them can have good results, so especially if you are gentle. They often even nickel plate them before painting too (usually a thinner nickel layer then you would want for engineering purposes)

The biggest problem with tins is that they are not designed for connectors, so unless you add a brace, they often deform substantially, especially with cables hanging off them, if you get extreme, if you see that, then you might wanna add a soldered wire, but it's better to reinforce and maintain the dimensions of the precise box.

Yes, to make a tin that is sufficiently air-tight that the food inside doesn't go stale requires tight tolerances. Unfortunately, the resulting tin isn't terribly rigid (it's only 0.1mm thick), especially if it's a square. That means that although you can get reasonable contact at the corners of a square box, there is likely to be negligible contact along the straight edges. And connectors need penny washers or large plates to spread the load. Nevertheless, even with those issues, I've made a lot of successful electronics in biscuit tins. Particularly, I've made low current measurements down to fA without hum being a problem. Beware that there's often an epoxy coating on the inside of tins, hence the need to abrade the surface to make electrical contact between lid and box, and to abrade the surface (glass fibre brush works well) to allow soldered joint at each end of the wire bonding lid to box.

[LooseJunkHater]

Tins are often made with very high end dies to tight tolerances, so sanding them can have good results, so especially if you are gentle. They often even nickel plate them before painting too (usually a thinner nickel layer then you would want for engineering purposes)

The biggest problem with tins is that they are not designed for connectors, so unless you add a brace, they often deform substantially, especially with cables hanging off them, if you get extreme, if you see that, then you might wanna add a soldered wire, but it's better to reinforce and maintain the dimensions of the precise box.

Yes, to make a tin that is sufficiently air-tight that the food inside doesn't go stale requires tight tolerances. Unfortunately, the resulting tin isn't terribly rigid (it's only 0.1mm thick), especially if it's a square. That means that although you can get reasonable contact at the corners of a square box, there is likely to be negligible contact along the straight edges. And connectors need penny washers or large plates to spread the load. Nevertheless, even with those issues, I've made a lot of successful electronics in biscuit tins. Particularly, I've made low current measurements down to fA without hum being a problem. Beware that there's often an epoxy coating on the inside of tins, hence the need to abrade the surface to make electrical contact between lid and box, and to abrade the surface (glass fibre brush works well) to allow soldered joint at each end of the wire bonding lid to box.

For your various MI + EI enclosures, do you create standoffs (ex. to hold PCB's, holding a smaller metal enclosure within the larger one) by piercing holes through the enclosure for the standoffs, spot-weld standoffs, or simply leave the items "floating"/unsecured within the enclosure? I'm thinking that by piercing the outermost (MI) enclosure, you then create a path for MI to enter the enclosure, but I'm not sure realistically how much of a problem this will create.

[EC8010]

Biscuit tins are only 0.1mm steel, so penetration depth means they only start screening against magnetic interference >200kHz, and none of my experiments have been that high in frequency. I'm not expecting to achieve magnetic screening, only electrostatic.

But I entirely take your point that drilling holes for screws creates apertures and I rarely use magnetic fasteners, so any holes are not magnetically plugged. One particularly fussy experiment needed self-adhesive copper tape over the gap between tin and lid all the way around to stop electrostatic hum crawling in through the gap.

For tests inside biscuit tins, I have breadboards with a piece of FR4 screwed to their back and with neodymium magnets glued to the copper foil. This stops the boards moving around and (slightly) reduces capacitive coupling between strips. If I do dead bug circuits on FR4, I usually earth bond the FR4 foil to the tin using a few bits of copper foil soldered both to foil and tin, thereby anchoring the circuit; some experiments have been very fussy about their bonds to the tin.

[tszaboo]

Biscuit tins are only 0.1mm steel, so penetration depth means they only start screening against magnetic interference >200kHz, and none of my experiments have been that high in frequency. I'm not expecting to achieve magnetic screening, only electrostatic.

But I entirely take your point that drilling holes for screws creates apertures and I rarely use magnetic fasteners, so any holes are not magnetically plugged. One particularly fussy experiment needed self-adhesive copper tape over the gap between tin and lid all the way around to stop electrostatic hum crawling in through the gap.

For tests inside biscuit tins, I have breadboards with a piece of FR4 screwed to their back and with neodymium magnets glued to the copper foil. This stops the boards moving around and (slightly) reduces capacitive coupling between strips. If I do dead bug circuits on FR4, I usually earth bond the FR4 foil to the tin using a few bits of copper foil soldered both to foil and tin, thereby anchoring the circuit; some experiments have been very fussy about their bonds to the tin.

You can buy one of those near field probe kits, and place it inside the box. See if the noise floor decreased compared to the outside.
Also, fun fact, coax cables have 3 conductors. You have the inner conductor, the outer shield, and the other side of the shield. Because RF, penetration depth, the two sides of the coax cable shield can carry different currents, which can upset antennas, but in this case, measurements.

[EC8010]

Also, fun fact, coax cables have 3 conductors. You have the inner conductor, the outer shield, and the other side of the shield. Because RF, penetration depth, the two sides of the coax cable shield can carry different currents, which can upset antennas, but in this case, measurements.

I imagine you have to be pretty high in frequency for that to become noticeable; do you have any figures?

[tszaboo]

Also, fun fact, coax cables have 3 conductors. You have the inner conductor, the outer shield, and the other side of the shield. Because RF, penetration depth, the two sides of the coax cable shield can carry different currents, which can upset antennas, but in this case, measurements.

I imagine you have to be pretty high in frequency for that to become noticeable; do you have any figures?

Good question, I've never thought about it, but it should be interesting to calculate.
A RG316 cable has 2.4mm diameter outer shielding, and 2.9mm jacket. Let's say this 0.5mm is mostly the shielding.
A skin depth calculator for copper gives me the distance of 65um for 1MHz.
206um for 0.1MHz.
So I guess at pretty low frequencies this can become an issue.
Made a quick check with Grok, it gave me 0.5MHz, because it was calculating with 0.27mm thickness shielding.

[temperance]

You have the inner conductor, the outer shield, and the other side of the shield. Because RF, penetration depth, the two sides of the coax cable shield can carry different currents, which can upset antennas, but in this case, measurements.

Under which circumstances will a HF current flow on the outside of the shield and not return on the inside of the shield?

[coppercone2]

but a coaxial cable is braid that has some litz like effects for skin depth. I think the basic calculation is for a hard line.


Interestingly, at lower frequencies... it might be better not to use a hard line for this reason. I think semi rigid (solder soaked) is also like this. Alot of measurements show that braid by itself is a bit more complicated then most rudimentary analysis.

And its further complicated by foil shield options

[tszaboo]

You have the inner conductor, the outer shield, and the other side of the shield. Because RF, penetration depth, the two sides of the coax cable shield can carry different currents, which can upset antennas, but in this case, measurements.

Under which circumstances will a HF current flow on the outside of the shield and not return on the inside of the shield?

Radio operators typically bring it up because they have dipoles, and it comes up with the discussions about baluns.
For measurements, I think... common mode noise?

[iMo]

..yep, you have to use a balun ([balanced-unbalanced) at that dipole's feedpoint, sure..
But the output of your LNA is unbalanced..