Noise Filter for DC SMPS

[Kev]

This is such a common issue, yet I have not found a great DYI solution or an affordable commercial product for it, so I am asking here.

Pretty much all DC power supplies these days are "switching" power supplies (SMPS), which are known to generate a lot of noise (voltage deviations from average) in their outputs. Cheaper, ungrounded supplies are worse than grounded ones. When used to power most devices, the noise is not a problem because either (a) the electronics (say LED lamps) are unaffected by the noise or the disruption caused by the noise is imperceptible to the user, or (b) the device is expecting noisy power and has built-in filters to suppress it.

Unfortunately, battery-powered audio devices (including audio outputs of cheap laptop computers), fit neither category (a) or (b), and the power supply noise, especially 50/60 Hz mains frequency and its harmonics, is unpleasantly audible when these devices are powered by DC SMPS.

I expected that, as a result, there would be some readily available, cheap, plug-in/pass-through noise filters, where you could plug in the noisy DC output from the power supply and get clean DC power from the other side of the filter. By "clean", I mean that it can be used to power audio circuits without audibly affecting the sound. So far, all I have found is this: iFi DC iPurifier2 Active Audio Noise Filter/Conditioner for DC Power Supplies for US$129, which to me seems about $125 more than an adequate filter should cost. It also only operates at or above 5 V, while, for my purposes, I would like one that operates at 2.5 V.

Question 1: COTS product recommendations? Can anyone recommend a nicely packaged commercial product that suits my needs, preferably for under US$20?

Question 2: Build my own? If "no" to Q 1, then is there a reasonably easy way to build my own? Of course, by "build" I mainly mean "assemble", as I am not going to make any parts myself, or even get a printed circuit board printed. I'm talking about putting some components on a proto board and soldering leads together. I want it to work in the range of 2.5 V to 24 V automatically, so I do not think I can use something like a linear voltage regulator.

Question 3: Why is this so hard? Power line noise from DC SMPS seems to be a pretty common problem, so why is this so hard. I can understand why (expense) its not built in to the power supplies, but a plug-in filter seems like it should be easy to make and relatively easy to sell. Why isn't the market flooded with options?


Preemptive response filter: for the purposes of this question, I am not interested in any other sources of noise, including ground loops, EMI picked up by audio inputs, or thermal noise in the pre-amps.

[ejeffrey]

Unfortunately, battery-powered audio devices (including audio outputs of cheap laptop computers), fit neither category (a) or (b), and the power supply noise, especially 50/60 Hz mains frequency and its harmonics, is unpleasantly audible when these devices are powered by DC SMPS.

Switching power supplies usually operate between 50 kHz and 1 MHz.  If you are hearing 50/60 Hz noise from your audio electronics, the one thing it definitely isn't is "switching noise" from the power supply.

The power supply still is -- or at least can be the cause.  Line harmonic buzzing is generally caused by y the issue is common mode coupling of the AC mains onto the signal ground via the EMI filter caps.  Those filter capacitors are indeed needed particularly on switching power supplies to pass emissions rules.  The important thing is that the 60 Hz is not actually coming from the switcher -- its just that there is a capacitor from the AC line input to your signal.

The other potential source of line frequency noise is ground loop pickup, although this is in no way specific to switching power supplies.  In this case, the interference is caused by 50/60 Hz magnetic fields being picked up by your cables.  Ironically the biggest source of these are frequently large line frequency transformers from nearby *linear* power supplies.  I have 100% seen situations (not in audio, in scientific measurements) where someone got all upset about "stupid noisy switch mode power supplies", didn't do any measurement or analysis, and replaced them with linear power supplies only to have the magnetic field from the linear power supplies cause problems they never had before, then start looking on the other side of the room for "more evil switching power supplies"

This why simple in-line filters are not readily available to solve this.  In both cases the power supply itself can be quite low noise measured from V+ to local ground so there isn't really anything to filter.  Rather the problem is that module 1 and module 2 have a different local "common" voltage, so when ground referenced signals are connected between them the two sides of the interconnect will interpret the signal voltage differently.

The general solution to this, and what almost all pro audio gear uses is to use balanced signals.  That is, each signal has a + and - output.  They are usually twisted together to almost completely eliminate magnetic pickup.  In this case, the ground potential difference ideally doesn't matter as it is not part of the signal path.

It's possible to make single ended signals work with careful control of equipment grounding, choice of power supplies with low ground leakage, and so on, but it's really a lot more work than switching to balanced circuits.

[Kim Christensen]

You could fool around with some chokes and capacitors to make a lowpass filter. Or Google for a "DC line noise filter" or "DC EMI filter".

[luudee]

Hello,

perhaps this video (#1116) that Dave made a while back may help:



Best Regards,
rudi