Post History

Most reliable galvanic isolation technology for extreme EMI environments?

I'm working on another project with extreme EMC requirements as per various notorious military EMC standards. Civilian product but for military use, and as such subject to the toughest levels of conducted/radiated susceptibility, 200V/m fields across an 2MHz-18GHz range, peak pulses of far higher energy yet, extremely low radiated emissions down to 0 dBµV/m average @ 30-450MHz etc etc. The details aren't important since my question is conceptual, just assume that the EMC requirements are on the ridiculous side of things. We've done similar projects several times before, so we know what we are up against in general terms.

As one requirement, we need to use a CAN bus and it needs to have galvanic isolation. Usually not a big deal, I'm a fan of using inductive digital isolators such as [these](https://www.analog.com/en/analog-dialogue/articles/digital-isolators-protect-rs232-bus.html) with a part like  [example datasheet ADµM1201](https://www.analog.com/media/en/technical-documentation/data-sheets/ADuM1200_1201.pdf). Since these a digital signals there's no worries about current transfer ratios and the like. These are cheaper, widely available and more rugged/reliable than optocouplers. 

(One optocoupler part or another in our product range tended to go end of life/LTB once every full moon, until I banned them from use in new designs. Some ten years later we've successfully phased most of them out, good riddance. EOL announcements are now blissfully quiet.)

I plan to clock the CAN bus as slow as possible without messing up the real-time requirements, but realistically I can probably not get it below 50kbps. Toying with slope control will probably happen though, either directly on the CAN transceiver or indirectly though external filters.

**My question is if I should be concerned when using these kind of digital isolators in a project with extreme EMI requirements?**

They specify a bit of transient immunity etc in the datasheet, which won't be a concern since there will be external TVS and filters. I'm not concerned about transients or conducted EMI - that's one reason to use galvanic isolation in the first place. But other than that, I don't have much of a clue how it performs when it comes to radiated emissions/susceptibility, other than that they've passed industrial/automotive EMC testing flawlessly in the past (RS in levels of 50-100V/m, typically). But military EMC is another matter.

Whereas an old-school optocoupler is... optical. So maybe I should get back to using them for special projects like this?