I'm trying to find out how "good" 5V power adapters can be with respect to power consumption if:

  • there's no attached power consumer, and/or
  • the consumer takes very little power, such as an ESP32 in one of its low-power sleep modes

Are there particular circuits that are known to be particularly good? And how would I hunt for products that implement those?

Clarification: I'm not meaning to ask about the state of the art in what's sold in the mass market market. Instead, I'm asking "what are the best circuits known", even if they are for very specialized niche markets, or aren't available in commercial products at all. E.g. -- just a wild guess -- could it be possible to build something that has essentially two separate circuits for different load situations and dynamically switches between them? Where the low-power one is made to produce, say, 10 micro-amps at 20% efficiency for the sleep states of the processor, and the high-power one is a "normal" one that is kept disconnected during the sleep state?

  • \$\begingroup\$ I got some hits with 'usb quiescent power supply' \$\endgroup\$ – Digiproc Feb 26 '19 at 21:18
  • \$\begingroup\$ Most tend to actually be pretty good in that respect. Maybe not state of the art in relative terms but they tend to be low power to begin with so even a higher than ideal idle consumption percentage is a small fraction of a watt. \$\endgroup\$ – Chris Stratton Feb 27 '19 at 1:13
  • \$\begingroup\$ What are you trying to achieve? Save the planet from global warming? \$\endgroup\$ – Ale..chenski Mar 1 '19 at 7:15
  • \$\begingroup\$ @Ale..chenski no just not making things worse :-) \$\endgroup\$ – Johannes Ernst Mar 1 '19 at 17:55

The efficiency level of AC-DC power adapters is regulated in all developed countries. Here is a good article on the subject from CUI. The adapters are assigned "efficiency level", from "I" to "VI". Products sold in the US should conform to "level VI":

enter image description here

Since 2011 all low-power adapters (<49W) were regulated by IEC AS/NZS 4665 Part 1 and Part 2. They should conform to level "V", with the standby power (aka vampire power, vampire draw, phantom load, ghost load, or leakage) under 300 mW.

With newer "level VI" the DOE requires the low-power adapters (with nameplate power under 49W) to have the standby power under 100 mW.

  • \$\begingroup\$ This is a nice article. Not what I meant to ask for (I'll update the question), but nevertheless I learned something useful I didn't know! \$\endgroup\$ – Johannes Ernst Feb 28 '19 at 17:10
  • \$\begingroup\$ @JohannesErnst, you probably wanted to ask what causes and limits the idle consumption of AC-DC adapters then. Keep in mind that the primary side is 140V at least after rectification, and switching the 140 V up and down carries switching losses, which will be difficult to reduce in economical way. \$\endgroup\$ – Ale..chenski Mar 1 '19 at 7:21

Assuming you mean the highest efficiency at low loads in Buck DC-DC regulator to provide 5V, I found a Rohm design with 50% efficiency at 0.1% of Imax which indicates the overhead current was 1/2 of the load but then around 90% at 1~3A.

So your load duty cycle determines the overall energy efficiency.

In order to be a stable control system feedback needs to be in the same units as the regulating variable. Such as regulating forward current using current sense and forward voltage with voltage sense and not the integral of voltage for current from reactive loads to avoid loss of phase margin and efficiency losses during switching so it must also use zero current switching near zero voltage phase which gets very complex in design terms with traces being 10nH/cm and stray capacitance. So low load efficiency is a tradeoff with many other variables.

The best design must sacrifice or tradeoff many variables for power efficiency, dynamic load range, input/output ratio limits, step load margin, max ripple etc.

Once you have a spec, then all parameters in a design can start or you can examine suppliers to these specs. But you cannot examine 1 parameter like **low load current efficiency "in isolation."


But if you mean, lowest quiescent LDO for 5V with some Imax... this is a different topology that is possible with CMOS LDO's. For example 1uA idle with 1A out max is possible.

Or you simply use a PMOS switch with >1M off resistance and say 10 mOhm on resistance for load regulation error RL/(RL+Ron)*100%.


Considering an AC-DC converter with a high-efficiency requirement that draws 1W is challenging (due to cost/W sensitivity) and ensure the safety to disturbances to humans without compromise.

The best solutions recentlyEU NA found are there.

3W continuous @5V 0.6A
< 0.075 W no load power consumption
Insulation 4242 Vdc 1 minute @ 10mA ( from Ic=C dV/dt )
safety approvals GS (IEC/EN 60950-1, IEC/EN 62368-1)
EMI/EMC CE (EN 55032, CISPR 32, EN 61000-3-2, EN 61000-3-3, EN 55024, CISPR 24)
Climate 20~40'C max ambient 85% RH max
OCP = 1A auto-recover

  • \$\begingroup\$ It is most likely that the question is about a mains supply, not DC/DC conversion. The former is everyday and implicit user/consumer experience, the latter tends to be a specialized need which people state explicitly. \$\endgroup\$ – Chris Stratton Feb 28 '19 at 17:57
  • \$\begingroup\$ Even the AC LPF line Balun filter would draw >100uA minimum without consideration to the rest. So again we have a question without full specs \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Feb 28 '19 at 17:58
  • \$\begingroup\$ To clarify further, here is an example of the scenario I'm thinking of: imagine an ESP32-based, self-contained temperature WiFi sensor, that plugs into mains. Most of the time (say 290 out of every 300sec) it has nothing to do, so it makes sense to activate one of the ESP32's low power modes (eg <10uA). When it wakes up, it talks to the WiFi, so we are talking 100's of mA. If the power adapter always takes 100's of mW, there is no point in even activating that low-power mode. So: what are the options for a higher-efficiency one for when it is in that state? \$\endgroup\$ – Johannes Ernst Feb 28 '19 at 21:26
  • \$\begingroup\$ @JohannesErnst "So: what are the options for a higher-efficiency.." One option that comes to mind is to run the device from rechargeable battery, and connect the mains electro-mechanically only when the battery needs recharging. Not sure however how much this green initiative will cost to produce, and what its carbon footprint will be. Or you can entertain the idea of capacitive transformerless power supply, with same idea of controlling the regulator. "Phantom power" on domestic level was always a big concern for green enthusiasts. – \$\endgroup\$ – Ale..chenski Mar 1 '19 at 7:39

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.