I am designing a dual supply powered system, where certain loads, depending on some conditions will be powered by one or other power rail, in order to achieve some safety levels.

The concept

The problem arises when trying to design an EMI/RFI filter for the system. First and foremost, adding the filter after the switch is not possible. This is because there are certain loads powered by the redundant power, whereas others not. Apart from the filter not being nearest to the device input, it would imply replicating filters where loads exist, which may bring more issues than benefits.

The first approach is this

First, naive approach

But common mode filtering will never work due to the current splitting.

Common mode filtering not working

I have been struggling with this, and the only solution I have come up with is to use FETs as switches to open the return power path only if voltage is present in the respective positive rail.

Power path enablers

This way, the current will flow through where is supposed to.

Expected power path

However, I find the solution too appealing. I have the feeling that I am missing something.

  • Is my solution to the problem valid? Any details/quirks to be considered, apart from the thermal ones?
  • Do you have any other solution to my problem?

ADDED: One of the answers suggests a 3P filter. This solution is to be applied to small scale electronics, so that would be my last resort.


1 Answer 1


It seems to me that a three-phase filter might do the job for your example case. These usually have all the windings on the same core.

enter image description here

Figure 1. A random three-phase filter. Image source: RS.

Connect Va via L1, Vb via L2 and connect G via L3 or N. I'm assuming that the N winding is the same as the other three. This leaves you with one spare feed path.

Just to be clear, you would then have the two supply commons connected at NIN in and have all the load commons connected at NOUT.

  • \$\begingroup\$ That's a good solution. However, for small scale yet high current embedded electronics might be a bit bulky. I will add that insight to the original question. Do you have any comment on the FET solution, which I consider less bulky? \$\endgroup\$
    – Manex
    Oct 26, 2020 at 14:26
  • \$\begingroup\$ Your question doesn't specify current magnitude. Smaller filters will be available or you may be able to wind your own for a limited run. Regarding the FET solution, I don't like switches in the negative lines. \$\endgroup\$
    – Transistor
    Oct 26, 2020 at 14:56
  • \$\begingroup\$ why don't you see that switch in the negative line? Thank you \$\endgroup\$
    – Manex
    Oct 26, 2020 at 15:09

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.