4

During reparations and teardowns I have frequently come across PCB-mounted RF shields and covers. While I don't typically work with GHz processors or design RF modules I wonder if such covers could be a way to efficiently avoid EMI issues in projects where unit cost is not a significant concern.

What type of designs scream "We should look into getting a RF shield cover!"?

What type of circuitry is not well suited for RF shield covers? (Heat dissipation? EMI peaks in sub-100 MHz range?)

Is it common to add such a cover from the start or is this typically a measure to combat EMI in later stages of a design project?

NoobPointerException
  • 730
  • 4
  • 18
  • Typically all manner of discreet RF may need it, no matter frequency. Single chip "RF IC" solutions are far less sensitive. – Lundin Apr 19 '22 at 12:54

2 Answers2

2

Usually these designs are tested with a spectrum analyzer or at a lab, if the processor is radiating beyond the FCC limits it gets a shield. Some designers might opt to add a shield from the beginning just to be safe or from previous experience. Shielding can also happen at the enclosure level, but if the enclosure isn't conductive or passes RF then it may be necessary to add the shield on the PCB.

Is it common to add such a cover from the start or is this typically a measure to combat EMI in later stages of a design project?

Since shields add cost to a project, most designers would not want to add them to a project unless necessary. With many shields (especially SMT shields) you can design them into the PCB and then test for radiation. If the PCB and/or chip is radiating more than the requirements (FCC or if you have radio interference), shielding will be needed. If you don't need it then you can leave it off of the bill of materials.

In many cases a near field wand and spectrum analyzer can be used for pretesting

Voltage Spike
  • 75,799
  • 36
  • 80
  • 208
  • It struck me later, that just as you say, there is an argument to be done for adding space for a shield but not mounting it if not needed. Could you elaborate on the question regarding when to add such option/possibility? – NoobPointerException Apr 14 '22 at 08:20
2

One place RF shields find use is to provide isolation between the input and output of amplifier chains, and between adjacent channels, or RF paths, within a module.

We all know that unwanted feedback between output and input can cause problems. In a limiting case, this feedback can cause an amplifier chain to oscillate if 1) the gain through the amplifier chain is greater than the feedback/leakage loss and 2) the phase shift of the leakage path is such that the fedback signal adds to the input.

But, this is not the only case. Many systems have very strict IPR (Impulse Response), or time sidelobe requirements. These can be on the order of -30 dB or greater. If the amplifier chain has a total gain of say 60 dB, and the IPR requirement is better than the -30 dB mentioned earlier, then the attenuation. or loss of the output to input leakage path needs to be greater than 90 dB. This is a very sporty requirement and so to meet this requirement RF shields (or huts) may be placed over the amplifier chain.

A similar situation may apply to multiple RF channels in a module, or on a board. Isolation between these channels may not be achievable unless selective shielding is employed, such as RF huts over key components.

All of this has nothing to do with meeting a government or user mandated EMC requirement. It's all about self-compatibility.

SteveSh
  • 9,672
  • 2
  • 14
  • 31
  • Very interesting information, I have yet to encounter the issues and requirements you mention so this was completely new to me. Which are the typical applications for such amplifier chians with strict IPR requirements? – NoobPointerException Apr 14 '22 at 08:15
  • 1
    Radar systems for one. And others, especially when you're trying to accurately characterize the received signal. – SteveSh Apr 14 '22 at 08:42