Dear Team,

When measuring the ripple of a Switching regulator using an oscilloscope,how to select the bandwidth of the oscilloscope.

Do I need to use full BW or BW limit option.

Regards
HARI

 

@5:26

 

I only use bandwidth limit when the fuzzy and spikey bits interfere with seeing what's going on. Depending upon the application, you might find yourself spending most of your time using the scope's full bandwidth because even the very short duration pulses can be a problem.

 

Many power supplies have have a noise spec like this; 2mV at 75% load, 20mhz bandwidth.

Switching power supplies have many sources of noise.
-120hz from the power line.
-100khz from the switching frequency + all the harmonics from a 100khz square wave.
-If the duty cycle is 25:75% then there is noise from 2.5uS and 7.5uS.
-Inductors, transformers and capacitors have a resonant frequency where they like to ring. There may be 10mhz noise traceable back to a component on the board.
-Transistors and diodes have a speed they switch at. A MOSFET might take 10nS to turn on.

When I measure power supply noise I am looking for 120hz and switching frequency noise and the first few harmonics. I don't look for 100mhz noise. That is measured under a different specification.

I would look up some commercial power supplies and see how they spec the ripple.

When measuring noise keep the ground loop small as possible.

Just connecting the scope to the power supply might cause noise that is not really there before. I add two of these to my scope probes coax, to help with high frequency noise. It does not effect the high frequency you can see on the scope. It breaks up high frequency ground loops.