Will be low pass filter be good enough to remove switching supply noise

Status
Not open for further replies.

Thread Starter

iinself

Joined Jan 18, 2013
98
Hi,
is a simple pi filter good to remove ripples/noise coming from a switching regulator? My intuition tells me it is lot complex but was wondering since switching is happening at high frequency will Pi filter with 2 caps of 4700uf and an inductor 47oouH do any good?

Thanks
 

crutschow

Joined Mar 14, 2008
34,283
It will help, but how much is determined by the amount of ground noise running around due to the switching transients.
For that you may need a common-mode choke in series with both the output and the ground line of the output of the regulator.
 

MrChips

Joined Oct 2, 2009
30,712
Yes and no. It depends. What is the switching frequency?

Why 4700μF?

Going to a higher capacitance does not necessarily mean better attenuation, not a high frequencies.

You want three capacitors in parallel, 0.1μF, 10μF and 100μF.
 

WBahn

Joined Mar 31, 2012
29,978
The question isn't whether or not it is good, but rather or not it is good enough. That depends on how much switching noise you can tolerate. For many applications you don't need to filter at all because the load can tolerate lots of switching noise. For other applications you can't use a switcher at all because you can't, in any practical sense, filter enough of the noise out to keep it from being a problem.
 

madsi

Joined Feb 13, 2015
107
The question isn't whether or not it is good, but rather or not it is good enough. That depends on how much switching noise you can tolerate. For many applications you don't need to filter at all because the load can tolerate lots of switching noise. For other applications you can't use a switcher at all because you can't, in any practical sense, filter enough of the noise out to keep it from being a problem.
You could probably don't need such large cap and inductor values. You could do probably do well just adding a small inductance (22uH for example) and a low-esr capacitor <=22uF at then maybe have this also paralleled by a .1uF low-esr ceramic cap. If you can, isolate the ground return as well with a 22uH. Your best bet would be to shield the switcher from the sensitive circuit with some metal if noise persists.

I just built a battery powered low-current switching convertor power supply and was able to eliminate almost all noise just using a low-esr SMD .33uF capacitor at the output followed by a LDO regulator to banish all remaining noise.
Your scope will tell you the truth about this matter.
 

Thread Starter

iinself

Joined Jan 18, 2013
98
Sorry for not being clear about what I am trying - since I am running my audio amplifier using walwarts that is boosted using a switching regulartor (boost converter), thought there could be some value in adding a filter at the end? I do not have a idea about the switching frequency of the walwart or the boost converter (purchased both of them from ebay).
 

Kermit2

Joined Feb 5, 2010
4,162
i dont know what kind of amp you have, so this may not apply.
if you have at least two stages of amplification, the last or final stage will utilize a majority of the power. If fed a 'clean' signal, it can use power from a noisy supply and still produce a clean sounding signal.
to get the clean signal the pre-amp stage will need filtered power. since its power usage is much smaller than the final stage you power it with a "shunt" regulator. google the term and study the concept. it is a Very effective filter for noisy voltage sources.
 

Thread Starter

iinself

Joined Jan 18, 2013
98
i dont know what kind of amp you have, so this may not apply.
if you have at least two stages of amplification, the last or final stage will utilize a majority of the power. If fed a 'clean' signal, it can use power from a noisy supply and still produce a clean sounding signal.
to get the clean signal the pre-amp stage will need filtered power. since its power usage is much smaller than the final stage you power it with a "shunt" regulator. google the term and study the concept. it is a Very effective filter for noisy voltage sources.
Thanks. It is a chip amp LM3886.
 

ronv

Joined Nov 12, 2008
3,770
Your getting 7 amps from a wall wort and a boost regulator? Please tell us more about what you have and what you are doing.
 

JMac3108

Joined Aug 16, 2010
348
Hi,
is a simple pi filter good to remove ripples/noise coming from a switching regulator? My intuition tells me it is lot complex but was wondering since switching is happening at high frequency will Pi filter with 2 caps of 4700uf and an inductor 47oouH do any good?

Thanks
Madsi and MrChips are right. You don't need a large cap. Try a low-ESR tantalum and maybe a couple ceramics in parallel.
 

Thread Starter

iinself

Joined Jan 18, 2013
98
Your getting 7 amps from a wall wort and a boost regulator? Please tell us more about what you have and what you are doing.
I have 2 12V/5A walwarts. I feed this into 2 boost to boost converter ( $3 on ebay each) and I adjust its output to 27V. Next I connect them in series to get (+27V)-0-(-27V). It works pretty well. Reading around people seem to think the switching noise is going to affect the quality of the audio, although I don't hear anything possible because the switching is happening well about audio freq range.

But just to see if filtering made any difference I built a LP filter with the above values, did not hear any difference, so was wondering if the inductor was doing its job?
 

Thread Starter

iinself

Joined Jan 18, 2013
98
You could probably don't need such large cap and inductor values. You could do probably do well just adding a small inductance (22uH for example) and a low-esr capacitor <=22uF at then maybe have this also paralleled by a .1uF low-esr ceramic cap. If you can, isolate the ground return as well with a 22uH. Your best bet would be to shield the switcher from the sensitive circuit with some metal if noise persists.

I just built a battery powered low-current switching convertor power supply and was able to eliminate almost all noise just using a low-esr SMD .33uF capacitor at the output followed by a LDO regulator to banish all remaining noise.
Your scope will tell you the truth about this matter.
Thanks. Working without a scope, I have to rely on good theory and hope my components are doing their job :)

A question - you mention using a LDO regulator, but some places they say that LDO introduces noise by themselves, especially at higher freq their PSRR drops, is that right?
 

ronv

Joined Nov 12, 2008
3,770
A 4700 uh inductor that is good for 7 amps would be huge, so my guess is yours is not doing much. It sounds like you are running a pretty good size power amplifier so your problem (if your having any) are probably due to the power supply dropping when the volume is loud maybe costing you some bass.
 

Papabravo

Joined Feb 24, 2006
21,159
Hi,
is a simple pi filter good to remove ripples/noise coming from a switching regulator? My intuition tells me it is lot complex but was wondering since switching is happening at high frequency will Pi filter with 2 caps of 4700uf and an inductor 47oouH do any good?

Thanks
Only if you can be certain that the noise power spectrum is above your cutoff frequency and provides sufficient attenuation in the transition and stopbands. You have zero chance of succeeding with theory and reliance on components. Buy or rent the tools you need.
 

Lundwall_Paul

Joined Oct 18, 2011
236
Yes and no. It depends. What is the switching frequency?

Why 4700μF?

Going to a higher capacitance does not necessarily mean better attenuation, not a high frequencies.

You want three capacitors in parallel, 0.1μF, 10μF and 100μF.
Question here. I know this these values are common in designs. My question is why these three particular values chosen after all the values just add up when placed in series?
 

MrChips

Joined Oct 2, 2009
30,712
You may think that the equivalent of capacitances in parallel is simply the sum of the individual capacitance.

Each capacitor has different Equivalent Series Resistance (ESR) and therefore has different frequency response. You want to stagger the decoupling capacitors to cover a wide band of frequencies.

Choosing and Using Bypass Capacitors
 
Status
Not open for further replies.
Top