# Is the TPS54202 buck converter a good solution to pass the FCC SDoC?

I want to design a PCB with the ESP32-WROOM module which has bluetooth. I want to pass the FCC verification. I would use a linear regulator, but it makes too much heat. My input is 14V car battery, output 3.3V, 0.5A peak. I've decided to give a try to TPS54202 as it's EMI friendly. The ESP32 says it can take up to 80mV ripple, so with a proper design, ripple should be fine. I'm just not sure about the EMI of TPS54202.

• Please consider safety circuits as well for voltage spikes from the car battery apart form the certification – User323693 Jan 24 at 2:48
• You have to deal with load-dumps when in a fossil fuel vehicle with a starter motor and alternator or generator that may be used or where it may be in motion. (Not needed if you will only be using the vehicle as a battery holder.) It's can be a nasty mess to deal with. These were my thoughts for a linear regulator, but even with a buck regulator you will need to be aware of at least some of the details you need to manage, somehow. Sadly, I really don't know much about passing FCC verification. – jonk Jan 24 at 7:36

As long as you follow the layout guidelines, there should be no concern over not passing EMC.

For this power class, it would still be work to use a linear regulator. Indeed, it can get a bit warm, but as long as you have some proper grounding for thermal dissipation, it should be alright. Furthermore, you would not have to worry too much about EMC.

A quick search at TI website gives the following LDO LM317, which delivers at least 500mA if the following condition is fulfilled: $$V_{in}-V_{out} \le 14V$$ which is your case.

According to the chip's datasheet the maximum power consumption should be around 1.8W:

If you want to increase the noise robustness, you could also add some additional filtering between the battery and the LDO. In order to reduce the common mode noise (mostly high frequency,) a choke-based filter would do the trick.

Concerning the certification itself it is hard to say whether you would pass or not in the first trial without running any pre-tests, since there is an iterative process between measuring the noise, produced either via conduction or emission, and its attenuation via filtering.

Despite all advice for filtering, my recommendation is to try to avoid creating noise in the first place. That means e.g.:

1. Avoid creating large loops in your layout through which a lot of current flows. (This might act as a loop antenna in some cases.)
2. Keep your high speed lines as short as possible.
3. If possible shield the housing.
• LM317 is a linear series regulator but it is not an LDO. – The Photon Jan 23 at 20:38
• Also, none of the packages listed on your linked datasheet are likely handle 5 W well. If OP's going to do this, they probably want a TO-220 package (and a big heat sink) as presented here. – The Photon Jan 23 at 20:44
• Im burning 5w at peak, 2w while esp32 is idle. Ive tried LM317 its hot as hell, I would need heatsink. But I dont want to have much heat in my sealed plastic enclosure as other parts will degrade faster. Thanks for your suggestions tho – TheAfrizzz Jan 24 at 0:36
• @TheAfrizzz I updated my answer. If efficiency is important, you should then follow your original proposal, using a Buck Regulator. – vtolentino Jan 24 at 11:51