7
\$\begingroup\$

I have the option of either using a vertical or horizontal bobbin for a flyback SMPS transformer, and I can't conclusively decide which would be best for minimal radiated interference to be picked up by nearby signal circuitry - of all the EMC/layout related material I've looked at, I have found nothing that mentions specifically this issue directly.

To be totally clear, I'm using 'vertical' to mean perpendicular to the plane of the PCB, and 'horizontal' to mean parallel. Here's a quick diagram - this is not meant to be to scale, just to illustrate the concept:

Diagram 1

The diagram shows the two transformer options mentioned above, and some signal circuitry which could be anywhere on the same PCB, not directly connected to the transformer.

So far, here are my assumptions - of course please correct anything that may not be quite right:

  1. The return currents from the signal circuitry should ideally flow in a plane directly underneath, therefore the signal circuitry should not contain a loop along the PCB plane with a vertical magnetic axis. However, instead, as there is a distance between the signal and ground plane, the signal traces and their return paths would create small loops which circulate down through the PCB and back up, producing a horizontal magnetic axis - see drawing below for an example. Hence, the transformer should have a vertical magnetic axis (Option A) for minimum pickup by PCB traces arranged in such a way.

Diagram 2

  1. If for some reason a signal cannot be routed so that the return flows directly below, this would create a loop which has a vertical magnetic axis - see diagram below for an example when this happens. Although this will be avoided, for minimum interference in such an arrangement the transformer should have a horizontal magnetic axis (Option B) Diagram 3

  2. In general, out of the cases in the 2 points above, the case in point (1) will be the most common and the one in point (2) will be avoided. Based on this, so far I think the vertical transformer (Option A) is best here.

I'm not too confident that I have considered everything that's important, so I'm asking for any suggestons from anyone with more experience for any other relevant factors, anything I've overlooked or otherwise verification whether or not I'm on the right line.

Thanks in advance!

Edit: some further information about the specific project following responses below - I will be using EE cores with a centre gap, and will most likely use a flux band aroud the windings to attenuate radiation, though I don't want to make the question too specialised to this project so it would be nice to establish some kind of general approach for dealing with this situation, as I think it's quite broadly applicable.

\$\endgroup\$
4
  • \$\begingroup\$ In case of things like emc there are many influencing factors. Frequencies involved, layout and function of your signal circuit and possibly many other factors. It is hard to give anything conclusive, and if possible imho the best option is a test. \$\endgroup\$ – PlasmaHH Jan 17 '15 at 15:23
  • \$\begingroup\$ Of course, I'm aware that the subject is complex and I wasn't expecting an answer to cover all cases - perhaps, instead, any 'rules of thumb' I should keep in mind, or any feedback on my thought process so far. Ideally I would have performed a test, however unfortunately my time scale to get this specific project finished is too short. \$\endgroup\$ – ew218 Jan 17 '15 at 15:33
  • \$\begingroup\$ Area in go-return loop on PCB surfaces as in 1. is usually seen as a necessary evil - but keep distances short. Loops as in 2. are usually unnecessary and to be avoided. Slot across path (eg in ground plane in some cases) is usually a very bad idea. \$\endgroup\$ – Russell McMahon Jan 17 '15 at 17:31
  • \$\begingroup\$ Of course, some finite area will exist - it's impossible to shrink the distance between two current paths to zero! I agree that the exact scenario in point (2) should be avoided, and I won't be using any slots in ground planes however it's possible to think of some scenarios where loops will exist in a design which have the illustrated magnetic axis (that specific drawing was meant as one example rather than the only case). The question is not how to route the tracks, but rather, given adequate attention to routing, how to orient the transformer to minimise coupling due to radiation. \$\endgroup\$ – ew218 Jan 17 '15 at 17:49
1
\$\begingroup\$

what's best for EMC?

Copper tape around the transformer - I've done this to shield the windings on one job that was causing stray radiated emissions that could be picked up - the "straying" magnetic field was measurably attenuated by the tape. Not elegant as a bodge in an EMC lab but it went into production with a more elegant looking shield.

\$\endgroup\$
3
  • \$\begingroup\$ Thanks, however I had already intended to use a flux band in order to attenuate the radiation - I didn't mention this in the question to try to make it general rather than too specialised to this project, though perhaps I'll edit this in as I imagine a flux band affects the radiation pattern and therefore might affect which orientation is best. I haven't had a chance to actually build a test prototype with the flux band to see how much of a difference it makes, however I'm assuming there will still be something to gain from orienting the transformer in an optimal manner... \$\endgroup\$ – ew218 Jan 17 '15 at 20:01
  • \$\begingroup\$ "I'm assuming there will still be something to gain from orienting the transformer in an optimal manner" - Not if it's an EMC test in a lab because the "test piece" has to be rotated and tested in 3 planes. \$\endgroup\$ – Andy aka Jan 17 '15 at 21:57
  • \$\begingroup\$ I'm not focusing on interference with the external world, but rather interference internally within the device - it's really just the scenario I described where some sensitive circuitry is on the same board as the SMPS - hence the question regarding the orientation of the transformer, which of course is fixed in relation to the PCB circuitry. As it happens this design will never see an EMC test lab (it's just a one-off for a custom piece of lab hardware), but regardless I was wondering if there have been any guidelines established around this issue of chosing the transformer-PCB orientation. \$\endgroup\$ – ew218 Jan 18 '15 at 0:34
0
\$\begingroup\$

There can be a bunch of EMI coming out of the gap of a flyback, depending on how it is constructed. If the center leg is gapped it's minimized. If all three legs are spaced, you'll get field lines coming out of each of the gaps.

\$\endgroup\$
1
  • \$\begingroup\$ I am using EE cores with only one centre gap in my design - the other two legs of the cores make contact with each other. \$\endgroup\$ – ew218 Jan 17 '15 at 16:47

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.