Decoupling caps with series inductor

[loudaslife]

I've been looking at the NXP BGA3018 datasheet, and I'm a bit confused by the application circuit on page 10.



My understanding of decoupling caps is that low value/small package caps are chosen to minimize parasitic inductance. If so, why place 10nF and 100pF caps on the other side of L2, which has 3-4 orders of magnitude more inductance than any ceramic cap? If you look at ESR instead of inductance, the story is the same. It seems to me like you may as well just use one large capacitor (maybe 1uF) instead.

[T3sl4co1l]

Yes, at high frequencies there is an inductance divider from the RFC to the caps, and between caps.  A large attenuation is desired, to keep RF out of the power supply.

A large capacitor will have the same ESL (or more, because it's physically larger), so will perform worse.

Tim

[tggzzz]

If so, why place 10nF and 100pF caps on the other side of L2, which has 3-4 orders of magnitude more inductance than any ceramic cap? If you look at ESR instead of inductance, the story is the same.

You would benefit from doing a quick simulation of the relevant parts of the circuit (an AC sweep in this case). LTSpice is a good starting point, but I expect there are online simulation programs.

As for the different sizes of capacitor, the answer will be found in the datasheets for different capacitors. The impedance vs frequency is the key point. For bonus points, find the equivalent parasitic inductance and capacitance, and include that in the simulation.

[Kleinstein]

The inductor L2 is part of the amplifier, not part of the decoupling. Decoupling is there to keep the RF out of the supply. I am not so sure the 100 pF really have an effect or are only there because of tradition from times when a 10 nF cap was large.

[loudaslife]

For some reason I hadn't even thought about keeping noise out of the power supply, I was only thinking about it in terms of providing a stable voltage to the IC. That makes much more sense, thanks.

[srb1954]

In this case the inductor is not for decoupling but supplies the DC supply current to the amplifier while also presenting a high RF impedance to minimise loading on the amplifier output. A resistor could be used in this position but the available amplifier gain would be lower and a higher supply voltage would be required.

[T3sl4co1l]

The reason for the inductor is not to supply stable voltage, but stable current.  It's not so much an IC as a prebiased transistor.

Tim