Whats the deal with using two capacitors ?

It seems like every project and schematic I look at, they always want to use two capacitors.

For example, connecting a 1 uF and a 0.1 uF across the power input of the chip to attentuate voltage ripple.

And they are usually different types of capacitors. I know there are a dozen types of capacitors, but this
is a capacitor manufacturing issue and not an electronic functionality issue, no ? The circuit analysis software
doesn't care what type of capacitor it is, only the capacitance.

I studied and used electronics extensively about 25 years ago but not recently. I never read about this business of
using two capacitors until I returned to model-making recently.

It can be a pretty complicated issue, and depends upon the parasitic properties of the capacitors, inductance and resistance. Every capacitor has a resonance point in frequency, where it is most effective in decoupling. Different capacitors with different capacitance and parasitic inductance will have a different resonant point. Putting them in parallel broadens the range of effectiveness.

--
The Gadget Shield: accelerometer, RGB LED, IR transmit/receive, speaker, microphone, light sensor, potentiometer, pushbuttons

Different types of capacitors (mylar, ceramic, tantalum) have/exhibit different ESR - equivalent series resistance.

and not an electronic functionality issue, no

It is an electronic functionality issue.

The circuit analysis software doesn't care what type of capacitor it is

That's one reason why prototypes are built.

It can be a pretty complicated issue,

Could that be the reason that it takes so many years to get a technology degree ?

Don

OK, so what you are saying is that having a 1 uF and a 0.1 uF both there, isn't just to have 1.1 uF of capacitance.

You are actually making a little network of 2 capacitors plus their internal resistances and stray capacitance, which
has a better effect on regulating the power supply or eliminating unwanted frequencies, than a single capacitor.

I guess if you could measure those "second order" effects, then you could model them. Or build a prototype.

Thanks a lot for the explanations, I have been trying to find an answer to this question for a couple of weeks
and now I know.

All components suffer from parasitics. Basicly that means there is no such thing as a perfect resistor, capacitor, or inductor. A good simulator will model these parasitics, this is really important for RF and microwave circuit design.
To answer your question: An electrolytic, even a tantalum, operates very poorly above a certain frequency. This is because the device starts to become a poor inductor, rather than capacitor. This can happen at well less than 1MHz with an electrolytic. Ceramic caps do much better. That being said, even a radial lead 0.1uF ceramic starts acting poorly above 10MHz, today 0.01uF is usually a better choice for decoupling. Chip caps are even better.
Try this. Connect an RF generator to a capacitor through a 51 omh resistor. Use a good O-Scope and measure across the cap as you increase the frequency of the RF generator. It's amazing how poor some caps really are.

The circuit analysis software doesn't care what type of capacitor it is, only the capacitance.

Which is why some some capacitor companies offer simulation models that are far more detailed for specific capacitor types than just the simple RLC model.

For a discussion of why we need decoupling see:-
http://www.thebox.myzen.co.uk/Tutorial/De-coupling.html