I read the Murata application notes on power supply decoupling and noise filtering. I guess it works great as advertisement because I now want to replace all the decoupling capacitors to my analog-related ICs with something like this :
CF1, CF2 : NFM21 three-terminal capacitors
As a relative beginner in electronics, I'm fine with wasting money on this kind of overkill circuit, if I can just drop it in and reap the advertised benefits (less Vin noise, equivalent voltage buffering compared to MLCC decoupling caps).
Are there hidden downsides ?
For the sake of argument, Vcc is 3.3v from a LDO regulator (which takes 5v from dirty USB power). The analog parts of the circuit are: a CCD (IC2 on the schematics, TCD1304DG), an op amp (TL971) and an ADC (AD7380). The CCD is clocked in at 2 MHz, and there are 12-20 MHz SPI lines running on the board. I can more or less add this without disturbing the routing of the actual signals.
Best Answer
The big- value feedthroughs are useful to knock down high-frequency noise, so you could possibly get rid of the PI filter.
Next to the device, the big benefit of FT caps is their low ESL at a given value. You can replace multiple 0.1uF with one 1uF for example. So they can save space under a large device that needs a lot of bypass.
For these reasons FT caps are starting to show up in mobile phones as their main application. With that volume driving them it’s not such a concern sourcing them.
There is close cousin called the XY cap that has 4 terminals. It’s more expensive, is patented, and has relatively few sources. It’s debatable whether it’s better than the FT.
That all said, with the relatively low frequency you’re using you probably won’t see a benefit using FT caps. You can get by with conventional caps and a ferrite bead, like 4.7uf + 600 ohm + several 0.1uF. If improved PSRR is needed consider a separate LDO for the ADC.