Which is the best enclosure for Sub-GHz and 2.4 GHz applications?

[sangar]

Hi All,

It is just my curiosity to know which kind of enclosures gives the best performance when they work with RF...I have heard that Plastic is always the first choice and then Aluminium for RF...But, a stainless steel is not good to go...Could someone briefly explain about RF performance with the different enclosures...

Thanks,
Muthu

[David Hess]

What matters are the seams which form slot antennas.  Finger stock or a conductive gasket or lots of closely spaced screws are used to close seams and prevent leakage.  Aluminum is a problem because of its insulating oxide.

[joeqsmith]

Aluminum can be plated with something like yellow chromate.

https://www.finishing.com/387/55.shtml

[coppercone2]

that is the most no nonsense explanation of aluminum conductivity I read. I feel like I got hit with a brick wall of bad information in a horror movie. They just need the track from 'the thing' playing in the background of that page (like when blair is looking at the microscope). Yea, when it reaches 150 nm, we are all fucking doomed. I want to chromate EVERYTHING. 

[cdev]

Chop off a few cm from a copper pipe and polish it up and clean it. Tin the parts where it connects to your PCB and make the caps for its ends out of sheet copper. Use edge mount coax connectors like SMAs but use the kind that also have a longer section that holds two nuts so you can get both edge PCB grounding and grounding to the shield. Have your PCB fit snugly to the inside of the copper pipe and dab it with a bit of solder at the edges too so its well grounded. Don't solder the point where it goes through the copper foil (or opening it again would be too difficult) just use the nuts to get good contact there. It looks a little homemade but works really well and likely (I have not needed to do this) could be made to be fairly waterproof if it needs to be..

Hi All,

It is just my curiosity to know which kind of enclosures gives the best performance when they work with RF...I have heard that Plastic is always the first choice and then Aluminium for RF...But, a stainless steel is not good to go...Could someone briefly explain about RF performance with the different enclosures...

Thanks,
Muthu

Plastic might be the best choice for people who did not know what they were doing as far as using metals for shielding effectively. Look for a PDF document about audio shielding and the "pin 1 problem"

It depends on the frequencies involved.

[David Hess]

For aluminum seams they make special finger stock which has deliberately sharp edges to puncture the aluminum oxide coating.

[coppercone2]

In regards to the pin 1 problem, what I did for my amplifiers that work at low frequencies is I connected the PCB to the chassis by star washers and metal standoffs and only wired the center conductor of the BNC to the PCB.

But, when I connected a power connection which is bipolar, the ground from the power supply does have a internal ground loop, I used heavy gauge wire from the ground connection to the PCB and also connected the stud to the chassis. Is this wrong?

[cdev]

coppercone2, I've had the same experience.

But you said that you figured out how to fix it at low frequencies which often is all you need. I had exactly the same problem with my SDR and I eventually figured out that using a really thick extra ground wire strapping the chassis of the SDR to the chassis of my computer with very tightly screwed on soldered lugs and thickish copper wire in addition to the shield on the audio cables helped a lot. (just the audio cables shield was not enough)

Lock washers is definitely the way to go there. I need to buy a bunch of them.

Doing something similar in the past is how I discovered that anodized aluminum is an insulator!

Also, I often put a capacitor large enough to be a fairly low RF impedance between the negative and ground connection on my power supplies when I am using them for an SDR (what works best though is running them off a battery).

The capacitor seems to help more often than hurt.

[coppercone2]

This is actually for the jim williams 100KHz preamplifier for noise measurements, I doubt it has any business connected to a SDR.

Should I lift the 0V (its powered by +-4.5V from a lab PSU or a internal switched in batteries) from the chassis, have it enter through a feed through, and go directly to the PCB?

 It just seems odd to connect the ground through a feed through (the +-4.5V are isolated bananas, but the 0V is a grounded solid metal banana). I don't really see the problem the ground loop between the interior of the chassis and the heavy gauge ground wire causing a problem with a system consisting of a few op-amps.

I also need to know this because I have the parts to build a active RF circulator with high speed op-amps which will have the same problem.. I was planning on using RF-Feedthrough caps on the +-rails and bonding the 0V ground at both the PCB and the chassis.

It defiantly seems weird to use the chassis as a ground return (is this good practice at all?). To be honest I also questioned it with the bananas, and I was planning to actually make it twinax BNC instead of regular BNC, since the chassis is made of steel.

I thought small signals fine, it can go through the chassis through 4 star washers, but the power supply ground???

What should I do as best practice for this LNA that uses LT1028s? I mean there is a loop, but its inside. Does it pick up noise from the power wires? Also there are some capacitors between the 4.5's and ground (tantalum <100uF).

To be fair I only did it because I ran out of isolated banana plugs, but does it matter? I actually thought it was making it better because the loop is internal and im shorting out a peice of cable with a big sheet of steel so the impedance should be lower.

[cdev]

I totally agree with you, if you can avoid using the chassis for a ground return don't do it. Just connect all the floating chassis's together to maintain shield integrity unless that causes problems.

Is it a differential signal you're amplifying??

I am also using two conductor plus shielded ground for the cable from my power supplies for low current situations a lot of the time. It just seems smarter.

Maybe you could try doing similar, grounding your devices case to your power supply ground, and running the two power leads, positive and negative to the appropriate power rails.

Whatever works.

No capacitor.

[coppercone2]

no its single ended. My idea for input twinax was because then if I have ANOTHER shielded enclosure to connect it to, I can get a single point ground on one of the PCB's acting as the entire shield ground, and then basically float the other PCB and connect the shield through the BNC, but I was not sure about capacitance to ground plane from the shield if its a big system. Like lift the PCB of the thing that connects to this amplifier off the ground and use the amplifiers PCB ground plane as a star ground for the shield, but I don't know. It might depend on the length of the cables too, with the impedance of the shield compared to the impedance of the wiring.

One day I will need to try both ways but its too much of a hassle to measure it without a dynamic signal analyzer so i just leave it alone for now. I get a headache thinking about it, it must be much simpler to replace a few connectors once I have the proper equipment. And you also need to think about cable type and the rammifications of twinaxial vs triaxail in this situation.

[cdev]

Those kinds of situations are where a really sensitive high bandwidth, differential probe would be nice to have.

I have an idea. Did you see the graphics from that German PDF which was posted a few days ago, it was about quieting a PC power supply? In there there are a bunch of photos - some detail a clamp on probe which goes around cables to get an idea of what signals are riding on them. Its simple, it uses a ferrite split core and several turns of wire around it. One could likely also make it into a directional coupler to measure the CM signals flowing in both directions separately. If you have even a cheap spectrum analyzer or even an RTLSDR you could probably learn a lot more about what improved the noise that way than anoything else, because it is just going to measure the common mode noise.

I have a huge split core that I think I am going to make into a CM probe, its the ideal thing to do with it.

If you can make a clip kind of like a clamp on ammeter that would be the most flexible.