In this case ringing will hardly occur, if for the 10µF caps electrolytics are used. In case of aluminium electrolytics the ESR is way beyond 1Ohm and a ringing cannot occur. If a tantal is used, a maximum will exist, but the impedance at this maximum stays well under 0.5Ohm.

By the way, the fact, that an impedance maximum occurs, doesn't necessarily mean, that this maximum is harmful! Some couples show relevant maxima, others don't. More, even the second or third best decoupling scheme can show a gratifying performance.

So, a combination of 10µF tantal and 100nF ceramics is a good supply decoupling scheme. Today I would use one 10µF/10V/X5S ceramic high cap in 0805 package instead, though.



Pi-filters are very well suited even in analog circuits. Due to the caps at input and output they are highly immune against complex wiring impedances. But even if a ringing would occur, which is rather improbable, then the ringing will take place at a rather low frequency, where the power supply rejection ratio of analog chips is very high. Take care, though, that the two caps of pi-filter to be used as power supply decoupling network is always in the 1...10µF range (or beyond) to provide this!






In the example above ringing could occur indeed. The decoupling isn't optimal, but seems to be a proven compromise, which at least seems to preserve the precicion of ADC!
Why compromise? Because here an additional requirement comes into play, which almost forbids the use of an additional series resistor or a larger cap to enhance the dampening: According to datasheet at power up or power down the supply voltage at AVCC must never go beyond or fall under the supply voltage VCC of digital core of microcontroller! So, if fast power ups and downs are to be expected, the time constant formed by shown components must be as small as possible!

More, if the VCC pin is carefully decoupled (10µF) and no spiky currents are drawn by the AVCC pin, then ringing need not necessarily occur, even if the decoupling scheme isn't properly damped. So, this scheme could work very well.

On the other hand, if you can guarantee by design (big storage cap) that only rather slow power ups and downs will occur, then you can increase the decoupling capacitance and even insert a series resistance. In any case, don't use an usual 10µH choke but a modern soft ferrite bead, even if the effective inductance is smaller than 10µH. This automatically adds the desired dampening resistance (ohmic losses) at frequencies above 1...10MHz and tremendously alleviates this ringing issue.

Kai