Why are ripples bad?

[rainman1]

Lets say a PSU supplies VDD for MCU, 3.3V for example, and this VDD has ripples of +-100mV-peak.
What is the problem with that?
Thanks.

 

[steveB]

Lets say a PSU supplies VDD for MCU, 3.3V for example, and this VDD has ripples of +-100mV-peak.
What is the problem with that?
Thanks.

It depends on the details of the device. Assuming the voltage (with ripple) is always in an acceptable range for device operation, then you are dealing with an issue of power supply noise rejection ratio (PSRR), which depends on the frequency of the ripple (or noise). Some devices have good PSRR at low frequency, but not at high frequency. So a 60 Hz ripple may not be an issue, but noise from a switching power supply (50 kHz - 1 Mhz) could be.

You really need to understand the device you are working with to figure out if +/-100 mV is a problem.

 

[audioguru]

That is too much ripple. The MCU will think, "WTF?"

 

[crutschow]

For circuits that are completely digital, 100mV of ripple is usually not a problem since digital circuits have a high tolerance for noise.

But for analog circuits (or a MCU with analog circuits such as an A/D) that amount of ripple can get into the signal path and corrupt the signal. Thus analog circuits usually require a much lower ripple and more power supply filtering than digital circuits.

 

[Willbe]

And ripple currents cause heating in filter capacitors.

 

[confounded]

can you tell me what does MCU stand for?

 

[rainman1]

From all of the above i understood that ripples are only bad for analog circuits.
(You cant say that ripples heat capacitor which you use for ripples).

So how come my crew found it very important to spend much board space (2 capacitors for each VDD input pin of digital IC) for reducing the ripples?

 

[steveB]

So how come my crew found it very important to spend much board space (2 capacitors for each VDD input pin of digital IC) for reducing the ripples?

That is good standard practice, and you will run into problems if you don't do that. Even if the small ripples are not an issue, glitches due to power transients become likely without capacitors on every chip.

However, you may be able to get away with one capacitor per chip. If you do try that, keep the larger value. In the old days two capacitor were used because the larger capacitor did not perform well at high frequency, so the small capacitor was there to pick up the fast transients. If you choose the capacitors carefully, keeping the larger value capacitor is sufficient. However, you don't save much money or board space removing the smaller capacitors. And, a recall of products could be much more expensive !!!

 

[crutschow]

From all of the above i understood that ripples are only bad for analog circuits.
(You cant say that ripples heat capacitor which you use for ripples).

So how come my crew found it very important to spend much board space (2 capacitors for each VDD input pin of digital IC) for reducing the ripples?

High ripple currents from high frequency switching regulators can heat a capacitor.

The capacitors on each chip are not for reducing power supply ripple, they are to reduce the high frequency transients that are generated on the power bus when high speed digital circuits switch. If you don't decouple them, they can be large enough to cause errors in the digital circuit operation.