Yup, still cares if it's a kit. Citation:
https://www.ecfr.gov/current/title-47/chapter-I/subchapter-A/part-15/subpart-A/section-15.25As for testing, or when law is actually employed... as is almost always the case in the US, it's employed at the behest of any authorized agent that takes notice. You could probably get by without anything, and just never move enough product that someone notices it's malfunctioning, and that those consequences come back to you. The risk of failure is, whatever it says in this part of course, and if doing a recall and testing and modifications and return would destroy your business... maybe that's a risk you're willing to take, maybe not. The business case could even be made to proceed on a best-efforts basis (perhaps invest in some precompliance testing, or contract/hire someone for same), turn some profit, then go back and do it, and hope that you didn't screw it up the first time (and if you did, you only need to recall those first sales).
Basic EMI principles, if you have a device with one connector or less, it's at least very easy to improve it to such point that it passes, and the risk is most likely radiated if anything -- there's no conducted emissions at all without a connector obviously, and with just one, the "reaction mass" can only be the enclosure itself. Two or more connectors, you have risks for EMI going up and down both -- equal and opposite emissions are the most likely case, and so conducted is a strong concern. This is obviously a big problem for power supplies where those two connections are supposed to be isolated so you can't just tie input/output grounds together and be done. If you don't need isolation, conversely, tying grounds -- and ensuring there's no induced voltage in the ground path between them -- gives at the very least a strong starting point for
Ground is king; if you can extend PCB ground plane up and over the connectors themselves (shielded or metallic connectors, braid cable or coax), you keep whatever nastiness is in the cables, neatly tucked away. You might avoid filtering digital comms with the use of shielded cables (pedestrian example: USB is
mostly differential, but it also delivers a common mode 1.6V pulse (SE0, both lines low) as part of its protocol, which would be unreadable, and impossible to filter, without the use of shielded cables and connectors). Even with, filtering to the maximum amount permissible given the nature of the signals is always helpful, and without shielding, filtering allows you to use bare naked signals at low bit rates (some kbps, maybe even up to 100s, depending on length) while avoiding emissions and susceptibility problems (similar example: USB HIDs with unshielded cables!).
FCC doesn't require susceptibility testing BTW, but does recommend it. CE requires it.
I take it, this is just an inline signal processing box, perhaps with an audio port on one end, and serial cable on the other. That's quite good, it means required bandwidth is small. The serial port might even be shielded (DE-9), for which you can recommend the use of a shielded cable in the manual. Even without, I might place a little filtering on RS-232 (more on "TTL level" serial, it needs more help), and a TVS for ESD protection. The audio side, I would not assume shielded cabling, and simply filtering it will do well enough (and a TVS / clamp diode).
Tim
Home
Help
Search
Login
Register
