I am wondering if AC noise (EMI) gets coupled to the DC side of the AC/DC adapter, what are the design methods used to isolate AC noise getting coupled to DC output in an AC/DC adapter. What are the relevant AC/DC adapter (Wall Wart) specifications that can help identify the quality of AC/DC adapters?

Thanks in advance for your help!


AC/DC adaptors using line frequency transformers have the lowest EMI around. There aren't that many of them however, although some still exist and can produce very 'clean' DC.

Switching supplies obviously generate a lot of EMI by their operation. Some of this is indeed coupled to the DC output. The path from AC line to DC is often quite intentionally formed by a small capacitor that bridges the isolation barrier. This capacitor is there to improve high frequency EMI but it's regrettably a source of low frequency EMI, notably line frequency related at twice line frequency (rectified), aka 'buzz'.

There are various techniques to limit the impact of this. The switching transformer can be a significant source of EMI and careful attention to design will mitigate this. Various constructions minimise EMI.

Circuit board layout can influence EMI. In fact there are many subtle areas that can improve performance. Good snubbbing of the transformer helps for example.

Unfortunately EMI standards concentrate on conducted EMI to the AC supply (for good reasons) and tend to neglect the DC output.

The unfortunate consequence is that it is very hard to select an adaptor that works especially well in respect of minimising EMI coupled to the DC output from information that's readily available.

I'd suggest that you should be vary of any adaptor that's unusually inexpensive. Sadly, simply paying more is no guarantee of quality but for sure a 'good' adaptor in this respect will never be cheap either !

  • $\begingroup$ Are their any conducted EMI measurements and limits that can be placed to minimized conducted EMI on DC output. Also what is best method the identify low frequency that get pass the isolation barrier. $\endgroup$ Oct 9 '20 at 18:00
  • $\begingroup$ For a stand alone supply, the DC output will have to pass conducted emissions tests but those tests only measure RF noise. The real problem unaddressed by standards is the low frequency noise generated. If this enters an audio system, as it often does, it's likely to introduce an annoying highly audible buzz at twice line frequency + harmonics.. The source of this waveform is the bridge rectified AC line. $\endgroup$ Oct 9 '20 at 18:11
  • $\begingroup$ So if 20Hz - 20KHz is filter out then, that would be an improvement. $\endgroup$ Oct 9 '20 at 18:22
  • $\begingroup$ Not entirely clear what you mean. The only compliance measurements involve measuring RF frequencies only. The source of the 100/120Hz + harmonics is 'by design' effectively. $\endgroup$ Oct 9 '20 at 18:26
  • $\begingroup$ So when you say low frequency, AC frequencies and its respective harmonics. $\endgroup$ Oct 9 '20 at 19:07

The youtube video USB Wall-Wart Power Supply Teardown will answer most of your questions.

The capacitor marked high light in the image below is acts as suppression capacitor. A class Y capacitor is preferred

enter image description here

Also image has the DC and the live side which is there to separate AC from DC. The below layout is good example of how this is done in on the PCB layout.

enter image description here

Also look for AC/DC adapters with lower ripple voltage.

One thing note is there are linear and switch mode AC/DC adapters in the market. Switch mode AC/DC adapter are more efficient.


  • $\begingroup$ The “suppression” capacitor shunts the common mode component of the switching waveform to the AC line. Its performance is limited by common mode impedance of the load: it better be high or your load will gladly take over some of the capacitor’s job and thus expose itself to common mode interference (I don’t call it noise since it’s a terrible term: little of it is random, it’s just the totally predictable periodic output of the switcher). $\endgroup$ Oct 11 '20 at 3:32
  • $\begingroup$ @UnslanderMonica, Thank you for the comment. Can you please elaborate on Its performance is limited by common mode impedance of the load: it better be high or your load will gladly take over some of the capacitor’s job and thus expose itself to common mode interference . Are you staying that that if the AC/DC Adapter is design for 800mA and the current drawing from the device is 20uA then the load will take over some of the capacitors CM interference. If so what are the ways to overcome such common mode interferences. $\endgroup$ Oct 11 '20 at 11:51
  • $\begingroup$ No, I'm saying that the common mode shunt capacitor connects the COM of the DC output voltage to the AC line. Said AC line is connected to ground through the ground-to-neutral bond on the transformer supplying the installation location (e.g. the building). Then if your load has low impedance and offers a parallel path to ground (e.g. because it's grounded directly, or indirectly through connections such as ethernet, etc.), then your load is just another impedance in parallel with the shunt capacitance inside the power supply - and you don't want that, as it exposes you to switcher currents. $\endgroup$ Oct 13 '20 at 21:11
  • $\begingroup$ So, you must actually control the common mode impedance of your load and make it high compared to the impedance of the shunt capacitor, so that it lets the shunt capacitor in the power supply itself do the job of shunting the switcher noise. To do otherwise not only exposes your load to unwanted currents, but also turns the cable connecting the load to the DC output of the supply into a big antenna for the switcher. This not only adds uncontrolled emissions to you EMC profile, but also introduces susceptibility due to reciprocity: if you can emit well, you can also receive. $\endgroup$ Oct 13 '20 at 21:12

What are the relevant AC/DC adapter (Wall Wart) specifications that can help identify the quality of AC/DC adapters?

Sometimes there are output ripple specs but they are so arbitrary and limited as to be useless.

So, in practice? None. You have to qualify the supply you will be using and potentially repeat some of the tests on each lot you receive, because most manufacturers will try and cheat you once they got you as a customer. That is: they will try and cheapen the design. And many manufacturers don't quite understand the various ways wideband AC can be measured, so you will often find that you have to feed their hand and provide them the exact measuring setup they should use, otherwise you'll get gibberish results (been there done that, and the story repeats itself over and over).

Ideally you would design your own supply if the application is that critical. There is this idea that you can let "someone else" do the power supplies, but there comes a point where the amount of time you spend coaxing your PS vendor to do what you really need starts to dwarf adding the supply as a yet another module of your own design. Presumably your design is much more complicated and expensive than the power supply.

Yet another approach is to have sufficient regulation at the input from the external supply that you can expect to cope with most anything that has a chance of passing emissions tests. I have taken that approach often: the device has a fast and wideband "RF" LDO preregulator, with sufficient common and differential mode impedance in front of it. But take care to design the differential filters so that their node impedances are balanced relative to both the "+" and "-/COM" input lines, so they themselves don't act as common-mode-to-differential converters (and vice versa - again, due to reciprocity, if they convert one way then they will convert the other way too).

  • $\begingroup$ If collaborating with a AC/DC adapter vendor, can you suggest some test specification beside ripple. Any thoughts on filter specifications. $\endgroup$ Oct 12 '20 at 14:30
  • 1
    $\begingroup$ The specification is driven by your application. Certainly once you understand the application, you'll know what sort of parameters the power supply's output must have, and thus how to specify them. Ripple specification is quite powerful, since you're giving the bandwidth of interest and either the RMS of peak-to-peak limit. Testing is commonly done by amplifying the signal, acquiring it using a wideband digitizer (such as a digital oscilloscope at full bandwidth/maximum sampling rate), then digitally filtering the output and determining go/no-go. $\endgroup$ Oct 13 '20 at 21:07
  • 1
    $\begingroup$ You can also give a spectral envelope, like you would if you were doing conducted radiation tests. Of course such envelope must be accompanied by the means used to connect the power supply's output to the spectrum analyzer: the component values do matter for repeatability there. $\endgroup$ Oct 13 '20 at 21:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.