# Creating oscillations on the output of a DC power supply

## Main Question or Discussion Point

Is there some common parameter for DC power supplies that provides the maximum oscillation amplitude allowable at the output?

For some context - I would like to generate an oscillating signal powered by a DC power supply, but to prevent most of the oscillations reaching the DC supply I am going to implement a radio frequency choke (RFC). I need to know how large the RFC must be to satisfy the power supplies maximum allowable oscillations at it's output.

Related Electrical Engineering News on Phys.org
Usually it's better to view this the other way: your 'filter' (usually a well sized capacitor bank close to the power parts) is not there to protect your PSU but to provide local (low impedance) 'backup' for the circuit at high frequency.

It is already bad enough if you have RF running on the power cords.

anorlunda
Mentor
The answer depends on the details. Can you tell us more? What are you trying to accomplish? What voltages, powers, frequencies are you working with? Maximum instantaneous power or max average power?

I'm unsure on the specifics but I was more just hoping someone knew the name of the parameter that details the maximum allowable oscillations at the output. I guess the answer to the following question will help me understand:

why is it common to block AC frequencies from reaching the output of a DC power supply?

Last edited:
why is it common to block AC frequencies from reaching the output of a DC power supply?
It is common to 'block' that AC as close to its source as possible, so it won't affect the DC power rail across the whole circuit (the power rail is not only about connecting something to the PSU: it is also about connecting everything together). When there is RF on the PSU what requires external/additional choking you are most likely already failed with the design process: since most circuits are far more sensitive than the PSU, when you have to start protecting the PSU the main stuff is usually a complete mess.

Ripple
Decoupling

Last edited:
berkeman
Mentor
Glad you found that reference. I was just about to post something similar. Testing for the stability of switching power supplies when the load is varying is an important part of characterizing their regions of stability (and regions of potential instabilities).

BTW, can you say more about the application? It's kind of unnatural to impose a signal on a low-impedance power supply rail. Is this some sort of "Link Power" communication network, like "Power Over Ethernet" (POE)? There are some tricks to such applications that can help to keep the costs down, compared to brute-force things like large blocking filters...

sophiecentaur
Gold Member
why is it common to block AC frequencies from reaching the output of a DC power supply?
It's good practice to reduce RF Interference as much as possible. There may be other equipment nearby and also there are regulations about RFI levels.
It's standard to have decoupling capacitors between the supply rails and ground in order to make sure that the circuit works 'as planned'. A signal path along the internal supply rail in a circuit can be a severe embarrassment, sometimes causing an Amplifier to Oscillate or an Oscillator just to Amplify.

Tom.G
A signal path along the internal supply rail in a circuit can be a severe embarrassment, sometimes causing an Amplifier to Oscillate or an Oscillator just to Amplify.
Or a digital logic circuit to operate illogically!

berkeman
Mentor
LOL

sophiecentaur
Gold Member
Or a digital logic circuit to operate illogically!
Yes.
When you get down to it, every 'digital' circuit is analogue. It is only the information that's digital. It's a waste of capacity to have the waveforms (internal as well as external) looking like the box-cars that we are presented with when digits first come our way. Timing is what it's all about in high speed digital circuitry and a sniff of something on a supply rail can cause deadly intersymbol interference.

.Scott
Homework Helper

I am just going to assume my RFC has to be large enough to reduce the ripple current and voltage bellow the rated ripple current and voltage.