[utensil4778]

As far as I understand, modules like an ESP32 for example, carry their own FCC certification. If you include them in your product, you do still need certification of the product overall, but you don't have to worry about the radio certification, just unintentional radiators.

For instance, if your widget includes an ESP32 and a switching power supply, you are (notionally) guaranteed to never fail certification due to bad behavior from the ESP, but if you botch your power supply design and are spewing out noise in the KHz to MHz range, you still fail certification.

Even if every individual component in your device carry their own certification, you still have to certify the product as a whole. Poor PCB design can produce bad EMI. Maybe you're running SPI over a long wire or your traces are routed in a way that accidentally creates an antenna at your SPI clock frequency. Hell, even something as simple as toggling a GPIO pin once a second can emit high frequency EMI under the right conditions.

There are a lot of ways to unintentionally produce harmful EMI, and that's exactly why FCC certification is required for everything. This stuff is hard to get right and there are endless gotchas and exceptions and edge cases and you have to know about and account for all of them.

[petsfed]

>This stuff is hard to get right and there are endless gotchas and exceptions and edge cases and you have to know about and account for all of them.

And this is also a major source of the cost of the testing. You're not just paying $5k+ for a piece of paper that says "FAIL" on it, and "better luck next time". The test engineers want you to pass, ultimately (if for no other reason than because you can't get repeat business from a customer who goes out of business), so they're going to point out the common sources of harmful EMI they've seen in other designs.

[jvanderbot]

This makes sense. Thanks for ELI5.

Presumably EMI certification is easier than the FCC RF certification.