Yes, any switching regulator creates EMI. Good PCB layout is the best way to reduce this. Can you post your layout and EMI results?

Yes, you can add a snubber. The gate driver in inside the IC, so you cannot affect this part of it. A properly selected and placed input cap also helps. Typically, a 470pF 0402 is the right value to use.

Thanks for sending me the files. That input cap is too small and so has quite bad DC bias characteristics. You need to have more effective cap at the input.

On the EMI side, I recommend turning off the other IC when you do the measurement. Otherwise, it is not clear where the EMI comes from.

It's not clear if you are measuring radiated or conducted emissions? Can you describe more about your setup?

Is this EMI an issue for your application? Any SMPS will make EMI. This is nothing unusual.

Chris, thank you for your response. What value would you recommend for the input cap? The datasheet for the regulator says: "For most applications, 10 µF is sufficient and is recommended, though a larger value reduces input-current ripple further." For this reason, we selected 10uF.

I am measuring radiated emissions. The reason we started investigating the emissions in the first place is because our circuit failed emissions testing at the testing lab. So we borrowed a spectrum analyzer and a far field probe and I can capture rough emission plots, like the one I sent you.

We tried your recommendation, i.e. turning off other parts of the circuit, and there were some reductions in the emissions. But the largest reduction by far has been disabling the 3.3V reg and using DC power supply. Unfortunately, power consumption is an important requirement, so we cannot simply shift to a linear regulator.

For this reason, I am considering using a snubber. Do you know what values may be good to start with? I was considering a 390 ohm resistor (P/N: P390ADCT-ND) and a 22uF cap (P/N: 445-8045-1-ND). An alternative is to use a low noise buck. Do you know if TI offers any good options?

Thank you very much!

10uF nominal is fine. The issue is that the effective capacitance at ~7V is too low due to the DC bias effect. Here, we have only 3 uF effective C. This gets much worse as Vin increases. A good rule of thumb is to have about half of the nominal capacitance at your biased voltage.

But we likely should add another smaller Cin right next to the IC. The best value to add is whatever value has a self-resonant frequency at the frequency that you want to snub. Could you send me your actual EMI test results to see where it failed?

Here is an app note on how to make a snubber: www.ti.com/.../slva255.pdf 22uF sounds waaaaaay too big.

This is a 2A device. Since you have a lower load current, a device like TPS62120/25 is much more appropriate. These lower current devices switch slower and have slower rise/fall times so there is less EMI.

Based on a 280 MHz peak, a capacitor with a self-resonant frequency at 280 MHz can best snub it. Here is one I found: GRM1555C1E102JA01 Add this between the main input cap and the IC as close as you can.