How come on many digital circuits there is both a 4.7uF capacitor and a 100nF capacitor in parallel next to each other? What difference is between this and just having a 4.8uF capacitor?

 

The larger cap is used to filter the lower frequency noise components while the smaller valued capacitor is used to filter the higher frequency noise components.

The lower valued capacitors have better high frequency response while the higher valued capacitors perform better at the lower frequencies.

hgmjr

 

You shold make a graph of the capacitative reactance as a function of frequency for these two capacitors. The difference will be apparent in great graphic detail...so to speak.

 

So ceramic filter high frequency better while electrolytics filter low frequency better?

 

You shold make a graph of the capacitative reactance as a function of frequency for these two capacitors. The difference will be apparent in great graphic detail...so to speak.

It will??

The reason you use both types is that electrolytic capacitors are rubbish at passing high frequencies. A graph of theoretical reactance vs frequency will not show this.

 

So ceramic filter high frequency better while electrolytics filter low frequency better?

Yes, that is correct.

Like papabravo has indicated, you have only to look at the manufacturer's graphs for each of the capacitor types and you will quickly see how the impedances of the two capacitor types vary with frequency.

hgmjr

 

It will??

The reason you use both types is that electrolytic capacitors are rubbish at passing high frequencies. A graph of theoretical reactance vs frequency will not show this.

I thought the graphs would show precisely that. Is it your position that they show something else?

The bulk electrolytic capacitor is also there to provide current during a load transient when logic gates, especially CMOS gates, are switching. Ceramic capaciotors are incapable of providing this function.

 

I think the key word was 'theoretical', which would be true. In reality, parasitics come into play and you can witness them on a 'real' reactance vs frequency graph.

Steve

 

Except that they are second order effects. The original question was about why two separate capacitors in parallel are different than a single capacitor with the same capacitance. The answer obviously depends on more than the total capacitance. That there are two separate paths to ground is essential to understanding the difference.

 

It is precisely due to these second order effects that an electrolytic capacitor is no good at high frequencies and the only way that a graph of reatance will show this is if you take into account ESR and ESL when calculating the reactance or actually measure the actual reatance at a various frequencies, neither course of action would be straight foreword or easy for a beginner.
If you just graph reactance using the standard formula X\(_{c}\)=1/\(\omega\)C then about all you will see is that the ceramic capacitor is too small to pass low frequencies but the electrolytic on its own is fine.

 

perhaps to add to the OP's question, why are other types of caps better? next to the Vcc pin of my counter why should i use a .01uF tantulum vs a .01uF electrolytic in an effort to scrub out high freq noise? my educated guess is it is in the physics of the materials being used.... i forget where, but i saw a chart online that listed cap types and what they are best used for, etc.

 

should i use a .01uF tantulum vs a .01uF electrolytic in an effort to scrub out high freq noise

Neither, at that value. You should use X7R ceramic 0.01uF.

However, for +5V or less digital circuits, I would suggest using X7R ceramic 0.22uF @ 10V (these come as small as 0603 size) for each and every IC bypass. If you need to filter higher freq, then parallel a 330pF - 1000pF or so with the 0.22uF.

Also, include some larger 2.2uF - 10uF caps (or possibly larger to 100uF) at varied locations around the board (these can be tantalum or electrolytic and act similar to large water towers stationed about the board, the smaller caps are like a toilet tank).

Although there is a lot of math that can be done to find the exact resonance of each IC + cap(s) + trace inductances, and thus choose the appropiate bypass values for each circuit, the above rule of thumb will tend to serve for DIY.