The Myth of Three Capacitor Values

[exxos]

Interrsting article. While it can get complicated, in general when using SMT caps, using 3 values increases the track lengths which just causes problems with higher impeedences. I've just used 1 cap for years and when in as doing digital audio circuits, it's actually greatly improved the noise figures.just 5mm track length can break a design. These days keeping track lengths down is more of a factor than the caps, assuming SMT ceramics are used of course.

https://www.signalintegrityjournal.com/ ... tor-values

Edit:
I've also seen 3 caps values causes ossilations. All the inductances and cap values build a very complicate mix of oscillators, found that out the hard way years ago. :roll:

[cmorley]

Analysing it is very complicated - you're in to modelling rather than beard stroking. I just use one value 100nF everywhere... one per power pin... keeps the BOM down to one type of cap and I've never had a problem with any PCB I've done that way. I'll use other values where datasheets ask (mainly power regulator stability caps) but exclusively 100nF for decoupling.

Since what I do works for me I pretty much ignore the topic..

edit: typo changed uF to nF

[exxos]

I don't really worry about. In terms of decoupling, I think the old school rule was 100nf every 6 cm or something. I just plaster 1uF -10uF close to main chips now. In terms of what we are generally doing, it really does not matter that much anyway.

[troed]

Same hour as I read through that I made an order to my local electronics company.

...
Capacitor MLCC 100nF 50V X7R 2.54mm × 15
...

I happily noted that they said that if you use MLCC then just go with a "high" value everywhere.

/Troed

[exxos]

Is probably worth noting as well that when you start getting down to 0805 packages, The actual capacity start dropping dramatically when voltages are present across it.. It's why I mostly trying keep to 0805 & 1206 packages and up.

Dave Jones did a video on all this *somewhere* as well not long back.