Selecting a ferrite bead for AVCC

[Starlord]

I'm trying to select a ferrite bead for the SAMD21:

http://ww1.microchip.com/downloads/en/DeviceDoc/40001882A.pdf

On page 856 it says:

The bead should provide enough impedance (e.g. 50? at 20MHz and 220? at 100MHz)
for separating the digital power from the analog power domain. Make sure to select a ferrite bead
designed for filtering applications with a low DC resistance to avoid a large voltage drop across the
ferrite bead.

Finding a bead that fits this description is easy enough if a select an 0603 part, but I was hoping I could go one smaller and use an 0402 here because I need the room. However at that size I'm having trouble finding an exact match.

So I'm wondering how crucial these parameters are and how far I can push them?

I assume that a higher DCR will result in a voltage drop, but since the max current the AVCC pin can accept is 60mA even if I double the DCR of the bead, the voltage drop will still be only .006V.  I would imagine the voltage drop could be 10x that and still not cause an issue. And that's a worst case scenario. I don't imagine the current draw on AVCC would ever approach 60mA in most realistic usage scenarios. This is a general purpose board, so I don't have a specific usage scenario in mind here to design around. If I were to assume 20mA then that 50mOhms could potentially be 30x higher... 1.5ohms, and not cause an issue. And even if AVCC was drawing 60mA I still don't think 1.5ohms would be a problem.

But I don't know for certain.

On the flip side, I don't know how much more is okay on the impedance.  220ohms impedance at 100MHz is what they mention, but is 400 okay? And what about 1K? I assume if I dropped it to 100 it would still be okay but give slightly noisier readings. But it would still be far better than no ferrite at all.

I selected all the 0402 beads with less than 500mOhm DCR and 150-1K impedance:
https://www.digikey.com/short/qqhc72

Here for example is the least expensive one which has a lot in stock. It's 500mOhms DCR and 600 ohm impedance @ 100MHz:
https://www.digikey.com/product-detail/en/taiyo-yuden/BK1005HM601-T/587-1839-1-ND/1465309

Is that okay? Or would it perform really poorly?

If I choose to sort by power line, then three good contenders with 150-220 mOhm DCR and 220-330 ohm impedance rise to the top:
https://www.digikey.com/short/qqhc3d

But I don't know if this really counts as a "power line" when it's 60mA max.

So, thoughts? What range of values should I be considering here?

[Rerouter]

the main thing is the series resistance needs to be there to damp out LC oscillations on power on, or load spikes on other parts of your system, if its too low, it can "twang" your LC filter and make it oscillate outside of the specs of the chip.

Moving on the L value, vs the C size you can calculate in excel. essentially you form an impedance divider for given AC frequencies. for a noise spike of X amplitude, it gets cut down by the ratio of the inductors impedance to the capacitors impedance,

How far you walk that path comes down to how much resolution you need out of your ADC's and what your error budget allows. most of your noise likley will be 100KHz and up on a digital system built around an ARM core, you are likely using a 0.1uF AVCC cap, which means your capacitors impedance is tiny, so you just need some inductance to keep that ratio tiny.

And you want some series R, or else you can make an LC tank circuit by accident.

[tszaboo]

Ferrite beads should be over-rated for current, because they stop acting as a ferrite as current increases.
http://www.analog.com/en/analog-dialogue/articles/ferrite-beads-demystified.html

With 50% of the rated currents, the inductance decreases by up to 90%.

So, go with the bigger one, and over specify on the rated current, not on the Ohms@100MHz.