Pros/cons of powering from 4.2V buck or 3.3V LDO

[okwestern]

EDIT: I have a li-ion charger ic (buck converter with built-in power path, auto-selecting buck/battery) giving 3.7-4.2V, which then powers a 3.3 LDO.

When the product is connected to 12V, the output will be buck regulated (with noise, etc.), but when 12V is absent, the built-in MOSFET will switch to output the battery instead. The 12V absence is very rare, but this is the critical time where efficiency is crucial, and I would avoid powering the nRFs from LDO if possible, since the nRFs can handle 4.2V anyway (nRF GPIO and other peripherals/sensors will be 3.3V). The LDO steals a little energy in reducing the voltage, plus the dropout lowers the voltage a at least few hundred mV (even on low consumption). Both contributes to a considerably shorter life on one battery cycle.

Is it recommended to power the nRF9160+nRF52840 directly from a buck? Or will the radio (LTE / BT /GPS), ADC, etc. be degraded by the switching noise?

If buck is OK/recommended, which configs to go with for 9160 and 52840?

And if not (LDO is needed), which configs for highest efficiency?

[ketiljo]

You can power both the nRF52840 and the nRF9160 from the buck converter. But you might need additional filtering depending on the power supply noise from the buck. There are internal regulators in both the nRF52840 and the nRF9160 so some of the ripple will be taken care of. Try to keep the ripple below 100 mVpp. 

REG0 the the nRF52840 will take care of the efficiency by itself by changing mode when needed. You need to the set the voltage though, but nothing more.