4
\$\begingroup\$

I'm not sure if it's appropriate to ask here, but I would assume many people in electronics would have experience with this.

I'm enclosing an RF sensitive PCB into an sealed aluminium enclosure. Aluminum is chosen because it is lightweight for shipping, and also a good thermal conductor which is required as the enclosure acts as the heatsink.

The big concern however, is that the circuit needs to be grounded to the aluminium enclosure for proper shielding. If the enclosure is not conductive and low impedance GNDed to the PCB then there are problems with internal noise generation from switching circuits being transferred to RF sensitive parts of the circuit.

The PCB slides into a railing and the edges of the PCB have exposed GND rails. Aluminum is not conductive on the surface due to its oxidization effects, and I have no way of applying some sort of pressure or screw to the enclsoure to force conductivity through the oxide layer.

What can be changed however it the coating on the aluminium. From doing research it looks like the coating I require is nickel plating, or tin plating. I've also looked into RoHs safe class 3 chromate conversion, but even this writes that some contact pressure is needed for conduction. Tin plated steel may be an alternative, but is a poor thermal conductor and is much heavier.

Does anyone have experience with conductive aluminum for RF shielded enclosures? What is commonly used? Does tin or nickel plating aluminum cost a lot more?

EDIT: Added an example image of the enclosure and PCB.

Front View of the Enclosure

\$\endgroup\$
  • \$\begingroup\$ You cannot install some type of wiper or spring finger on the rail? Are you constrained to simply accept the rail exactly as is with the exception of applying a coating? It seems like modifying the railing is the best bet unless it is impossible. \$\endgroup\$ – mkeith Apr 16 '16 at 5:27
  • \$\begingroup\$ I added a picture to show what the situation looks like. I'm not sure how i'd add a wiper or spring finger. Any examples? The enclosure is made from extruded aluminum and I could change the mould, but it would cost a lot. \$\endgroup\$ – Projectile Fish Apr 16 '16 at 7:27
  • \$\begingroup\$ Are there metal end caps screwed to the extrusion that can be used? \$\endgroup\$ – Transistor Apr 16 '16 at 8:24
  • \$\begingroup\$ Yes, but actually i've found that if the PCB is grounded only at the end caps, then EMI is very bad. Grounding the PCB at many points, e.g. along the rail eliminates EMI. \$\endgroup\$ – Projectile Fish Apr 16 '16 at 8:27
  • 1
    \$\begingroup\$ Potentially sounds like a PCB layout problem if you need so many grounding points. \$\endgroup\$ – Andy aka Apr 16 '16 at 8:40
3
\$\begingroup\$

If I need to ensure good contact to a housing, regardless of the metal it is made of, I use EMI-fingers or collars where possible.

You can get them made to almost any spec, from cheap tin plated to 3μm paladium-gold finish.

Here are a few examples:
Weird little spring-finger that might be slotted in
A type of EMI-Gasket/collar
And another EMI-Gasket
A wrap around EMI-Finger, that gets mounted on the lower flange and then presses down onto the board after wrapping around it

Those beat out any, and I mean any solid metal object with any kind of finish, because rails and other such do not apply any force. Try it at home: Get a nice clean metal surface (copper, steel, whatever), apply a voltage and then measure that voltage, preferably with a light load, like a 1k resistor, while softly laying a contact on that plate. Then apply a tiny bit of pressure. Note the difference in the measurement accuracy.

Contact resistance is directly proportional to the amount of pressure applied.


That said, I agree with Andy's comment that if you need good contact to the housing to eliminate noise that comes from your own PCB! there's something very wrong in the design.

It sounds more like you require the casing to be a ground plane, than an actual shield.

Good use of ground planes and stitching vias and though through layout of noisy and noise sensitive areas should take care of all of that, even if the board is dangling in free space. The reason to need reliable contact between housing and board ground is noise from the outside, not noise from the inside. The latter being one that requires much fewer connection points. (usually one global connect and one accompanying any sensitive signal, through coaxial wire for example)

\$\endgroup\$
  • \$\begingroup\$ Thanks for the info on spring fingers. I should be able to come up with a way to use these. You and Andy are probably right about the PCB problems too though. Is it probably the problem that both ends of the PCB are not connected with a low impedance enough ground connection? \$\endgroup\$ – Projectile Fish Apr 17 '16 at 3:59
  • 1
    \$\begingroup\$ @ProjectileFish Something along the lines of too little ground connection, or bad ground design is what I would bet, yes. But it's nearly impossible to say without source files and the appropriate time to review them. Just that adding more connection externally makes it better is a good hint it's at the least in the ground plane design. An example would be not having a star ground and no ground plane, which allows the switching electronics to inject spikes into the ground of the signal system. \$\endgroup\$ – Asmyldof Apr 17 '16 at 16:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.