dknguyen said:I dunno...high frequency is just a descriptive term. But after searching a bunch of documents on typical impedance curves for ceramic capacitors...the maximum frequency a X7R surface mount capacitor (leaded is a possibility for larger capacitances at less than 1MHz) is good for bypassing (theoretically according to the measured impedance curves) is about...
300MHz for 0.001uF surface mount
60MHz for 0.01uF surface mount
10~11MHz for 0.1uF surface mount
4~6~8 MHz for 1uF surface mount
4MHz for 2.2uF surface mount
2MHz for 4.7uF surface mount
1-2MHz for 10uF surface mount
1Mhz for 22uF surface mount
500kHz for 100uF
References:
**broken link removed**
**broken link removed**
**broken link removed**
http://www.ttiinc.com/object/tech_seminars_052505muratapacket
Remember digital square waves contain harmonics that are 100x or 1000x greater than their frequency. Also be aware that just because a capacitor is better for bypassing at a higher frequency doesn't mean it's impedance will be as good (as low) for bypassing a lower frequency as a larger capacitor.
Neat stuff...if you're obsessive compulsive.
dannix said:Don't worry, I'm not going to sue... :s
I'm a bit obsessed with capacitors because so much if it seems like a black art. I know you don't trust the rules, but after digging and digging and digging to see where they are coming from, I always end back at the beginning just following the rules knowing they aren't always true just because it's so much easier.Optikon said:Thats not the problem...
We just don't want you walking away thinking a 2.2uF capacitor is only good to 4MHz... maybe it is maybe it isnt.. the details will determine that.
I generally don't like seeing hard-fast rule tables like that for these reasons. It is much more satisfying to understand the fundamental issues with components.
dknguyen said:I'm a bit obsessed with capacitors because so much if it seems like a black art. I know you don't trust the rules, but after digging and digging and digging to see where they are coming from, I always end back at the beginning just following the rules knowing they aren't always true just because it's so much easier.
Lately I've been getting really paranoid about using a few different sized caps in parallel to bypass due to impedance spikes. I know what a impedance curve for two caps looks like, but I'm paranoid about using 3 because I can't predict how they will interact with each other.
The chart seems to suggest 0.01uF for a 40MHz PIC (which in itself isn't accurate since it's a square wave and the harmonics are much higher suggesting an even smaller value!), but it seems to work just fine with 0.1uF.
And that reminds me, the reason you tend to use a smaller capacitor for higher frequencies isn't because small capacitances bypass better there (on the contrary, in theory where capacitors have no parasitic resistances or inductances, a larger capacitor is always better). But in real life, larger capacitors have higher inductances which completely nullifies the effectiveness of the capacitance at higher frequencies. Hence, the packaging and capacitor type also affect this heavily. So all that stuff really says more about the trend of parasitic inductances with capacitances rather than how capacitance is related to effectivebypass frequencies.
Optikon said:I agree with what you're saying... but when you say 0.1uF works fine, what are you saying works fine? The PIC still runs, yes - I'm sure that is true but there is an effect. It is all really a matter of what the design requires to be acceptable. Do you know that your 0.1uF case isnt radiating more electromagnetically? How about power supply noise? You can be sure there is an electrical difference somewhere. The question is whether or not it remains acceptable.
Bypassing is really a power supply impedance problem. It is especially important for low noise analog work and very high speed digital designs.
I have found that below a 0603 surface mount part, there is little to be gained in terms of reducing parasitic inductance (i.e. 0402, 0201 or similar)
power and ground plane capacitance become dominant at >100 MHz and often that only constitutes < 1000 pF
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?