Noise doesn't 'hurt' the traces or other passive components, only the ICs, so bypass caps should always be mounted as close to the ICs they are 'protecting' and use as short leads as possible.
One insight that might help appreciate why digital circuitry especially needs thorough decoupling:
3kW mains heater at 240V 50Hz, 12.5A rms, max rate of change of current is 5.5kA/s
1000W audio amplifier into 4 ohm load at 10kHz (16A rms signal ) - max rate of change of current is 1.4 MA/s
20mA LED driven from logic circuit (7ns switching time) - 2.8 MA/s
Lightning is something like 1000MA/s for comparison.
So a little digital circuit driving one LED could easily be causing twice as fast a current change as a large PA amplifier, and far larger than anything on your mains wiring - and this interacts with stray inductance in your circuit to develop unwanted voltage spikes without decoupling real close to the chip.
Incidentally the only output switching time for the Arduino processor I could find on the datasheet was a "typical" value for the SPI SCK output of 3.6ns, so I'm probably being conservative with these switching times.
http://forum.arduino.cc/index.php?topic=92017.0
One insight that might help appreciate why digital circuitry especially needs thorough decoupling:
3kW mains heater at 240V 50Hz, 12.5A rms, max rate of change of current is 5.5kA/s
1000W audio amplifier into 4 ohm load at 10kHz (16A rms signal ) - max rate of change of current is 1.4 MA/s
20mA LED driven from logic circuit (7ns switching time) - 2.8 MA/s
Lightning is something like 1000MA/s for comparison.
So a little digital circuit driving one LED could easily be causing twice as fast a current change as a large PA amplifier, and far larger than anything on your mains wiring - and this interacts with stray inductance in your circuit to develop unwanted voltage spikes without decoupling real close to the chip.
Incidentally the only output switching time for the Arduino processor I could find on the datasheet was a "typical" value for the SPI SCK output of 3.6ns, so I'm probably being conservative with these switching times.
http://forum.arduino.cc/index.php?topic=92017.0
Comments
Post a Comment