Idea for powering low power circuits directly from mains

[rs20]

Hey,

I've had this idea knocking around in my head (I'm sure I'm not the first by a long way) about how to power 5V logic type stuff off mains voltage without burning watts in an adapter. My particular application is completely enclosed (radio controlled or something, haven't got that far yet) so isolation is not a requirement. I'm budgeting ~1mA typical draw, with bursts of 20mA.

One classic approach that I heard once was coupling AC in via a capacitor, and using two diodes to turn that into a charge pump. I think this suffers from a pretty bad power factor, but apart from that seems pretty nice. The only problem is that you basically pick the output current by choosing the capacitor, if your load varies, you have to burn off excess current in a zener or similar. So my idea was, use an optotriac to turn the mains input on and off, and have the logic monitor the input and control the optotriac accordingly. Basic schematic attached.

One obvious issue is that although it works fine once it's up and running, it'll be stuck off at initial turn-on. I could perhaps use some kind of battery backup, and design things so that that battery never runs out. Something involving depletion-mode fets perhaps. The huge capacitance of the supercap isn't really necessary, a standard eletrolytic would probably work fine with a higher frequency on the CTRL line.

But that issue aside, I'm curious to hear your thoughts on this idea. Obviously it's a fairly complex replacement for batteries, but are there simpler solutions that last for ten years (a claim I fabricate) and consume a similarly small amount of (real) power?

EDIT: After writing all this, I just realized this circuit pulls the full 5V load current (let's say 10mA) from mains, which works out at 2.4VA (similar to a wall adapter from memory), but just almost completely 90 degrees out of sync with mains voltage. How would my electricity meter report this?

EDIT 2: OK, further processing going on in my mind. Why not remove the charge pump sillyness, and just use a little transformer instead? In other words, hook up a transformer with input voltage chopped at a 5% / 10% duty cycle, f << 1 Hz. This actually sounds like a good idea to me now...

[SArepairman]

keep in mind that you probably need a mains rated foil capacitor from a reputable manufacturer for this circuit. Plus that power rail might be too dirty for a MCU, I think it will be much dirtier then a transformer. I think these types of circuits are considered to be on the "less reliable" end of the spectrum, but maybe thats because of poor component choices.

https://www.eevblog.com/forum/suggestions/capacitive-dropper-psu%27s/
be sure to read post #5, you need input protection too. Not that cheap when you make it safe.


there are lots of traps for young players with this topology. I would stay away from it unless your selling a billion things. If you wanna actually use it in your house just use a transformer + switchmode/linear regulator, your requirements will cost like peanuts

i think that a triac will still hurt your capacitor unless you wanna do zero crossing detection.

[rs20]

keep in mind that you probably need a mains rated foil capacitor from a reputable manufacturer for this circuit. Plus that power rail might be too dirty for a MCU, I think it will be much dirtier then a transformer. I think these types of circuits are considered to be on the "less reliable" end of the spectrum, but maybe thats because of poor component choices.

https://www.eevblog.com/forum/suggestions/capacitive-dropper-psu%27s/
be sure to read post #5, you need input protection too. Not that cheap when you make it safe.


there are lots of traps for young players with this topology. I would stay away from it unless your selling a billion things. If you wanna actually use it in your house just use a transformer + switchmode/linear regulator, your requirements will cost like peanuts

i think that a triac will still hurt your capacitor unless you wanna do zero crossing detection.

Thanks for that link, I hadn't considered that. It seems that the capacitive divider formed by the foil cap and the super cap (with suitable help to maintain good high frequency performance) would render that particular point a non-issue, though. Especially if you add a bit of extra ESR to the foil cap.

But yeah, I'm definitely leaning (see "EDIT 2" in my original long-winded post) towards a transformer* that's cut on and off by a triac as my current plan. A traditional always-on transformer has just too much phantom power draw; the purpose of this project is an intelligent timer/thermostat type thing that doesn't draw as much phantom power as the thing that it's turning on and off because that would be the height of pointlessness! I'd be much prouder of a solution that cost $100 but consumed tens of milliwatts rather than having to replace batteries every year. Almost certainly not a marketable product, but that's why I have to build it myself .

* Either a standard 50/60Hz steel transformer, or an AC/DC converter like this $12.50-from-digikey one, or maybe that FSAR001 thing.

[Post_Apocalyptic_Inventor]

You could use a capacitive voltage divider, followed by a rectifier and +5V linear regulator. I guess that's maybe the standard approach you already mentioned, but think about it: The overall efficiency will not be brilliant, but since you will only draw 1mA of load current the overall power will be so small that bad efficiency  will not matter much. This technique is used in mains powered infrared motion detectors (PIR). I guess a Power Factor Correction will not be neccesary due to the low power level. (Using a triac will in any case not be a good idea if you are concerned abot EMI, or power factor). The circuit however is of course a bit dangerous, since it doesn't provide real electrical isolation.Therefore I wouldn't use it in a commercial product. I'm not even sure how it is legally possible that these circuits can be used in commercial devices like  PIRs. 

You can see an example-circuit like it is used in PIRs in one of my videos:



If you don't like that circuit at all, I would recommend to build a flyback converter based on an integrated SMPS-Controller with integrated power mosfet like the fairchild "FSDM 311"

Datasheet:

http://www.fairchildsemi.com/ds/FS/FSDM311A.pdf

To do that you will however need a special transformer.

The device which you linked (named "PBK-1) could be used too. As it seems, it does provide electrical isolation. Haven't seen one of those before.

[NiHaoMike]

What about add a smaller cap across the optocoupler? You could size it to just barely meet the steady state power requirements.

You'll also want to add a series resistor for inrush limiting, a zener for backup overvoltage protection, and use a X2 rated cap for the series impedance. And, of course, the whole circuit must be insulated to mains spec.

Another trick is to use a 240V input transformer on 120V mains or a 480V input transformer on 240V mains. There are also very low standby switchers available nowadays.

[ejeffrey]

I remember seeing, I don't know where, a transformerless supply that worked by using a fast switch to only turn on when the line voltage was below 5V.  I don't remember who makes it, I think it was intended for things like power monitors that are connected to the line voltage anyway.

[tom66]

You could use a capacitive voltage divider, followed by a rectifier and +5V linear regulator. I guess that's maybe the standard approach you already mentioned, but think about it: The overall efficiency will not be brilliant, but since you will only draw 1mA of load current the overall power will be so small that bad efficiency  will not matter much. This technique is used in mains powered infrared motion detectors (PIR). I guess a Power Factor Correction will not be neccesary due to the low power level. (Using a triac will in any case not be a good idea if you are concerned abot EMI, or power factor). The circuit however is of course a bit dangerous, since it doesn't provide real electrical isolation.Therefore I wouldn't use it in a commercial product. I'm not even sure how it is legally possible that these circuits can be used in commercial devices like  PIRs. 

It's legal if it's a doubly insulated device or it has an earth and it's singly insulated, I think. And all outputs will be through relays or optos or the like.

By your logic nothing on the primary but a transformer primary would be possible - but frequently a very complex circuit is found on the primary side of a power supply.

It's definitely dangerous for testing so you should probably debug the circuit on 5V directly before powering it from the mains - and use a variac  or adjustable  current-limited ac power source to test it slowly!

[NSA]

Microchip did a data sheet for this type of supply: ww1.microchip.com/downloads/en/AppNotes/00954A.pdf

[Post_Apocalyptic_Inventor]

@tom66: No that's not what I meant. I know my way around SMPS and it is true that there can be complex circuitry directly connected to the mains, but  good "off-line SMPS" are based on those DC to DC converter toplogies, which do offer electrical isolation, like the flyback, forward and push pull converter. opto couplers and driver transfomers are often used in addition  and they allow that signals can be send from the primary to the secondary side (or vice versa), without compromising the electrical isolation from earth potential. You  can say that many measures are taken to make sure that non of the output voltages of an SMPS  are earth referenced / elctrically directly connected to the mains. For that same reason you should not  use Buck and Boost Converters to directly convert the rectified mains voltage.

If you use voltage divider circuits of some kind( be it  resistive, capacitive or a combination of the two), or also phase fired controllers, the output volatge will in any case be earth referenced. And that will be a problem for a commercial product if your device will have any voltages "leading out" of its insulating enclosure. Even if these voltage are "only" 5V logic level signals.   

(I said that I wonder about the typical PIR-circuits because they are the only common devices that I know which use capacitive deviders to directly convert the mains voltage down. They also have neither metal cases (with connection to the protective earth), nor do they have a double insulated enclosure. My guess why they are legal anyway is, that they have no other ouput voltages "leading out" of their insulating plastic enclosure than the mains itself (switched over a relais). And that is of course dangerous no matter if it comes out of the PIR or directly from the socket. Other than that there would be no danger unless you open up the device. And that is of course not what a normal  consumer is supposed to do.)