1

I'm designing a PCB with an AFE (Analog Front-End) for measuring/recording bipotentials and I'm trying to determine how much of the circuit I should shield with a Faraday cage. At the moment, I'm thinking of shielding the ADC and the digital isolators (used for SPI isolation). The PCB uses a MCU for Bluetooth communication so I see no reason to shield that.

Note: The AFE I'm using (ADS1294) has RLD (right leg drive) but trying to avoid a 3-pin configuration for measurements. Also open to avoiding RLD and Faraday and just using a low-pass or Butterworth filter in processing.

Any help is appreciated.

winny
  • 13,064
  • 6
  • 46
  • 63
Dionysus
  • 51
  • 5
  • Doesn't it require knowing how much a problem there is now, how much you need to make the problem smaller, and from where the problem comes from? Have you got any numbers? The shielding won't help if noise affects the system via data or supply wires, or Bluetooth radiation causes problems throughout the circuit. – Justme Dec 01 '22 at 21:31
  • My primary concern is protecting the analog signal. A faraday shield will definitely reduce the degree of interference from the power supply and improve CMRR. I'm trying to proactively address this, before looking into alternative solutions (like RLD). – Dionysus Dec 01 '22 at 21:37
  • 1
    Please only comment if you're genuinely interested in providing value. Coming on forums to disparage those with less knowledge is not productive for anyone. – Dionysus Dec 01 '22 at 21:38
  • 2
    This is a Q&A site, not a traditional forum. I just wanted to point out that your question about how much you need to shield needs more details in order to give some kind of answer. There is no schematic or PCB layout given for example. You might have a super great 6 layer PCB design with ground planes shielding everything already, or a 1 layer mess of airwires. – Justme Dec 01 '22 at 21:48
  • 1
    That's fair. Let me see if I can a schematic finished and add it to the question. – Dionysus Dec 01 '22 at 22:05
  • Doesn't really sound like a Faraday cage will help much. Nor will (probably) a low-pass filter at the input. Although as Justme pointed out, more info would be needed to provide a meaningful answer, the kinds of problems that you seem to be guarding against do not travel by air (which is what the F. cage would address), but they leak through poorly setup ground and power connections. Properly designed layout is likely to help _much more_ than a F. cage. Disclaimer: depending on your conditions, it may turn out that a F. cage could also help, possibly even being necessary. – Cal-linux Dec 02 '22 at 00:57
  • Wonder why you ask. Probably you already got into some problem. Supply like you said a situational photo/schematic. :) – RemyHx Dec 02 '22 at 05:33
  • 1
    In defense of @Justme - OP's problem is ill defined, and therefore hard and fast solutions cannot be provided. So OP's criticism of Justme was, IMO, unwarranted. – SteveSh Dec 23 '22 at 12:56
  • We sometimes make provisions for an EMI shield (or hut) to go over what, based on prior experience, may be a problem area. This could be a component or a group of components. This provision is something like a copper or solder ring around the suspect area that can be used to attach an EMI hut should testing show that is needed. – SteveSh Dec 23 '22 at 12:59

1 Answers1

0

Shielding is most helpful if it shields the high-impedance section, i.e. in this case patient, from the in-band EMI source, e.g. mains lines. Do you see how this isn't in the scope of your device?

Out-of-band EMI, e.g. from your PSU should be adressed with appropriate EMI filters instead. For example, place RLC lowpasses in your signal chain. These should be anyway present as anti-aliasing filters for the ADC.

tobalt
  • 18,646
  • 16
  • 73