Could I change decoupling capacitors values on ATmega power pin?

Question

I have a question about some capacitor values. I'm building, as part of my project, an Arduino with an ATmega328. For some reasons, in my design I have capacitors of 47 k that allow me to construct 100 μF and others of 100 nF. That's fine, in order to keep using fewer different capacitors, I want to know if this change that I made is correct.

In the original design this is the construction in the Arduino's original schematic:

And I made this change:

I think it's a filter control of the power signal to try to mitigate the problems with spikes, but I don't know how critical it is, that is, if it's not too critical to support these changes (with critical I mean, how strict must be the values).

I see this other question: Changing decoupling capacitor values

I know the duty of them, my question is about the "grade" of difference, because I'm changing from 1 μF and 4.7 μF to 100 nF and 47 μF, and I don't know if that scale of change is too much.

Answer 1

Experience shows that nothing good ever comes from deviating from manufacturer recommendations... is your original schematic actually a manufacturer recommendation though, and not just some random Arduino board?

Anyway, these are large enough to act as "bulk" caps at some extent, to counter small surges. It's custom for MCUs to have a cap between 4u7F to 10uF somewhere on the main supply. I have no idea why you want to go up to 47uF - in that case I'd probably place the larger bulk cap directly after the voltage regulator (since the whole 5V net would benefit from it) and keep a smaller one on the MCU Vcc pin.

The exact value of the MCU Vcc cap is however unlikely to be critical, and you certainly don't have to use an expensive cap with polarity there. Sprinkling 3 caps with very similar values also seems needlessly expensive and I can't think of any good reason for that, unless there are multiple pins far from each other. In that case one cap per pin would be recommended, but since the pins are called 4 and 6, this doesn't seem to be the case.

I'd pick a single 6u8F X5R 10V ceramic cap instead. Tolerance etc is not critical at all. Should be available from 0603 or larger.

This doesn't rule out a decoupling cap of 100nF in parallel on the same pin though, if that's what the manufacturer actually recommends. Filtering fast spikes and filtering surges are somewhat different purposes.

Answer 2

There is no bypass cap value that is theoretically required.

For example, AVR hardware design considerations suggests a value of 100nF.

So of course you can change the value within reason.

It is unknown why your example has 1uF but it can be just as good value as 100nF.

Answer 3

A very common combination is a 10uF tantalum or ceramic in parallel with a 100nF ceramic. Each type offers different filtering capabilities in terms of the frequencies they are able to absorb. Tantalum caps, for instance, possess low ESR, which is important at high-frequencies filtering as it dissipates less heat. As a general rule, the most important parameter is the type of dielectric (as someone already mentioned). XR5 or XR7 dielectrics have great properties for EMI rejection and ripple filtering.