1
\$\begingroup\$

Designing a 4 layer board I've found that if I place some power lines running on top layer I can easly keep them wide and route them without headaches. But this lines are very noisy becouse they powers a group of PWM leds(at 5V/1.5A) and one brushed motor(at 5V/1A). Since this is not a professional board, maybe I can ignore this, but this made me think about how much this will affect signals and components on the other side of PCB. I've studied that metal covers acts like a Faraday's cage creating an "immunity" on circuits that are inside and having the ground connected to the cover.

My generic question: Since one layer of the PCB is a solid ground plane and it's like a wall between signal/components and power lines, this assumption it's still true? If yes, this is used in professional boards?

My specific question: In my case, can EMI from this lines disrupt my 3.3v logic circuits or will they be so little to even affect an ADC read? (it's just a small general board with with an STM32 L0, lines are impedence matched, no longer than 2cm signals)

\$\endgroup\$
1
  • 1
    \$\begingroup\$ A ground plane will not stop magnetic coupling from one track to another unless the frequency gets into the MHz ranges. \$\endgroup\$
    – Andy aka
    Commented Nov 9, 2016 at 11:50

1 Answer 1

1
\$\begingroup\$

Yes, different conductors on a PCB have some capacitance between them. This can cause coupling of signals between traces. The usual ways to avoid this are:

  1. Distance. Capacitance between traces falls off with distance.

  2. Shielding. Put another low-impedance conductors (like ground) between the two traces. They will each couple to this center conductor, but not to each other. This applies to parallel traces in the same layer, and something like a ground or power plane on a layer between the two conductors.

  3. Good design. Try to limit voltage dV/dt, since capacitive coupling is proportional to it. Try to reduce the impedance of traces you don't want picking up signals from elsewhere. Feedback around a amplifier section, for example, also helps to attenuate noise picked up within that section.

You also have to consider inductive coupling. Passing current thru a wire causes a circular magnetic field around the wire. This also works in reverse in that a changing circular magnetic field around a wire induces a voltage along that wire.

when two conductors are parallel and near each other, each is partially within the magnetic field of the other. Running a alternating current thru one will induce some alternating voltage in the other. If you think this is far-fetched, consider that transformers work by deliberately exploiting this principle.

Avoiding magnetic coupling can get tricky, because something like a ground plane doesn't block it. In general, avoid long runs of parallel conductors where one carries high current and the other is sensitive to induced noise. Just like with capacitance, distance helps. For magnetic coupling, orientation also matters. Two conductors crossing at right angles in theory have no magnetic coupling between them.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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