I am planning to use use NCP1117 for another project. So decided to check it up in Arduino UNO Rev3 schematic. In the NCP1117 datasheet there is Cin and Cout corresponding to PC1 and PC2 in arduino schematic. But in Arduino schematic there is also 100nF capacitor named as C2. In some other schematics using other voltage regulators there is also a similar 100nF capacitor. So I guess it is something necessary.
Can anyone please tell me What is that C2 for?
Thank you in advance.
The datasheet recommends 10uF on both sides.
The output capacitor also depends on the load. For higher currents (e.g. > 300mA) required by the RF devices (WiFi etc), I would use a 470uF capacitor.
A regulator needs decoupling caps for stable operation.
Sometimes a combination of electrolytic and ceramic is used.
Refer to the regulator's datasheet.
Look at the board's eagle files if the caps drawn are actually positioned near the regulator.
They might be drawn near the regulator, but used to decouple another chip.
Leo..
florinc:
The datasheet recommends 10uF on both sides.
The output capacitor also depends on the load. For higher currents (e.g. > 300mA) required by the RF devices (WiFi etc), I would use a 470uF capacitor.
There is the output capacitor PC2 47uF. C2 is something else. Parallel to it in the schematic.
Wawa:
A regulator needs decoupling caps for stable operation.
Sometimes a combination of electrolytic and ceramic is used.
Refer to the regulator's datasheet.Look at the board's eagle files if the caps drawn are actually positioned near the regulator.
They might be drawn near the regulator, but used to decouple another chip.
Leo..
--- Sometimes a combination of electrolytic and ceramic is used.
--- Refer to the regulator's datasheet.
That makes sense now. But in the sample circuits there is no combination is mentioned actually..
--- Look at the board's eagle files if the caps drawn are actually positioned near the regulator.
--- They might be drawn near the regulator, but used to decouple another chip.
Hmmm. I wouldnt call this near but they are close.. C2 is somewhere else on the board file. I am trying to find out where it actually goes now. Not too good with board files, you know. 
I get lost easily.
Nope. I couldnt make it. Its just some random capacitor connecting 5V to GND when you look at the board. I dont think it is a decoupling capacitor. If it was a decoupling capacitor why would they draw it near the voltage regulator in the schematic?
I still think it is something about voltage regulation but cannot be sure.
C2 is very close to the power pin of the USB chip.
Leo..
A mix of at least three design principles is involved here.
A regulator requires an output capacitor connected close to its output for stability. 0.1 µF is usual.
Each digital logic chip - and that is what an ATmega328 is - requires a bypass capacitor connected close to each supply input for stability. 0.1 µF is usual and the ATmega328 has (at least) two supply terminals. This means there must be not just one, but two, three or many 0.1 µF bypasses on the supply line because there will almost always be significant distances between regulator and digital chips..
Electrolytic capacitors generally have significant inductance, so while they may be fitted as "reservoirs", they do not function as high-speed bypasses, so you require both electrolytics and (miniature) ceramics with negligible lead lengths.
Finally, ground paths should be "flood filled" as far as possible with multiple "thru"s between tracks on opposite board sides to reduce ground impedance and supply lines similarly, as wide as practical.
Paul__B:
A mix of at least three design principles is involved here.A regulator requires an output capacitor connected close to its output for stability. 0.1 µF is usual.
I think the output capacitor is 47uF PC2. NCP1117, LM1117, AMS1117 all similar ICs and their datahseets advice output capacitors between 4.7uF to 22uF. I think the designer picked it 47uF for more stability.
Paul__B:
Each digital logic chip - and that is what an ATmega328 is - requires a bypass capacitor connected close to each supply input for stability. 0.1 µF is usual and the ATmega328 has (at least) two supply terminals. This means there must be not just one, but two, three or many 0.1 µF bypasses on the supply line because there will almost always be significant distances between regulator and digital chips..
I could find only one bypass (bypass and decoupling, are they same?) capacitor in the schematic. That is C6. It is just in front of the VCC(PIN7) and goes to the VCC(PIN20) with a trace on the top layer going below the chip on the boardfile.
Paul__B:
Electrolytic capacitors generally have significant inductance, so while they may be fitted as "reservoirs", they do not function as high-speed bypasses, so you require both electrolytics and (miniature) ceramics with negligible lead lengths.
I think that is the actual reason. Those capacitors are big so they end up being electrolytic.
Paul__B:
Finally, ground paths should be "flood filled" as far as possible with multiple "thru"s between tracks on opposite board sides to reduce ground impedance and supply lines similarly, as wide as practical.
Eh.. I couldnt get this. By "flood filled" you mean that I should have a groud plane on both sides and connect them vias and make supply lines wide?
Damien82:
Eh.. I couldnt get this. By "flood filled" you mean that I should have a groud plane on both sides and connect them vias and make supply lines wide?
Yes. Look at any of the Eagle board files from SparkFun. They have a polygon (not a rectangle) on both sides of the board which is named "GND". This becomes the ground connection for all components. Any through-hole components join both ground planes so you don't usually need to add vias. The board design rules will make the flood fill flow around all your other traces automatically whenever you hit "ratsnest".
It is often annoying to design the board while those planes are filled but you also need to have the grounds connected. After ratsnesting, "ripup" the polygons. This will remove the flood fill so you can see your other traces.
The final trick to Eagle to find a specific component on the board is to use the "i" information button to highlight the component on either the schematic or the board and then alt-tab to the other Eagle window. The component will be highlighted there too. This works best on nets or traces. Using Eagle's button to switch to the other window will drop the highlighting.
Finally, ground paths should be "flood filled" as far as possible with multiple "thru"s between tracks on >opposite board sides to reduce ground impedance and supply lines similarly, as wide as practical.
Can you explain what you mean by this? Bit confusing.
"Flood fill" is where ground connections are not only made as tracks of a given width, but expand to fill up all available copper area between the other tracks. This reduces impedance - resistance and inductance - reducing voltage drops between ground connections.
Where there is such a ground track on both sides of the board, extra "via"s - plated-through holes - can be provided so that both sides become one conductor. This is in addition to the "thru"s which are required to continue the ground connection and as was mentioned by MorganS are frequently a hole where a component lead mounts already.
Damien82:
Nope. I couldnt make it. Its just some random capacitor connecting 5V to GND when you look at the board. I dont think it is a decoupling capacitor. If it was a decoupling capacitor why would they draw it near the voltage regulator in the schematic?
No, its perfectly standard practice is to plonk all the decouplers in one place in the schematic, since
its entirely boring and a waste of time placing in the 'right' place there - they need to be in the
right place on the board layout, that's all you care about. Decoupling is "infrastructure"