How does input voltage ripple affect output voltage of linear regulator?

Question

The problem starts with my 3D printer which is powered by a 12V SMPS. When it drives a heavy load such as heated base and hot end, it causes a ripple on the 12V rail. Also it affects the 5V rail with the same ripple as the image shows.

Note: My printer uses Ramp 1.4 with Arduino Mega2560

Note:
Chanel A is 5V from Arduino Mega, Chanel B is 12V from PSU.

For ramp1.4 , For Arduino Mega 2560

What I want to know is:

1. How does this happen? As I know, voltage regulator should regulate all above 5V+(dropout voltage), so it should be a flat 5V line.
2. How to fix or prevent this? (Do I need to add some component filter or change regulator ICs?)

Answer 1

1. In the datasheet of a linear regulator there should be a parameter called "power supply rejection ratio" (PSRR), which describes how much the ripples on the input voltage get suppressed. In fact this parameter is frequency dependent, wherefore there should be a graph about the course of PSRR over the frequency of the input noise.
   Example: Linear regulator with 5 V output voltage and 60 dB PSRR @ 1 kHz. Input voltage around 7V but there is an oscillation of 1 Vp-p and 1 kHz frequency on the input signal. The output voltage will be around 5 V with the oscillation suppressed by 60 dB, so Vp-p of the oscillation would be around 1 mV.

2. Without further information about your circuit and components I can't really say much about that. A simple point would be to get a linear regulator with a higher PSRR, but I don't know if that is possible for your application.

Answer 2

You need to be extremely careful when probing noise and ripple as the unwanted signal may be on the ground because of where you are probing, not on the supply rail itself.

It is preferable to use the probe in the x1 setting not x10. And select a ground point very close to the supply you are measuring. The spring ground attachment you probably got with your probe is one way of helping that.

Is it causing a problem? If not it may not be real.

The scope trace shows about 200mV p-p ripple. That shouldn't affect the digital circuitry.

Answer 3

Let us talk heating, of the silicon die/package/PCB.

Thermal time constant of cubic meter of silicon is 11,400 seconds.

Thermal time constant of 0.1 meter cube is 100X faster, at 114 seconds.

Thermal time constant of 1cm cube is 100X faster, at 1.14 seconds.

Thermal time constant of 1mm cube (about the size of a silicon die) is 100X faster, at 0.0114 seconds, or 11.4 milliseconds.

If you have a 2mm size die, you get 4X the time constant or 45.6 milliSeconds.

Or if you examine the mounting flag/paddle inside the plastic package, you'll have about that size.

Notably, copper has nearly the same ( 9,600 versus 11,400) seconds/meter^3 as silicon.