Mains AC to DC PSU without neutral / earth question

[Wilksey]

Hi,
Firstly, apologies if this has already been asked.

I understand the dangers of electricity, etc, I am not 17th edition, but have a G39 certificate allowing me to wire things up on the side of the road without supervision.

Now the safety disclaimer is out of the way

I have seen several designs for home automation, one that springs to mind is the Itead smart switch (1,2 or 3 channel), which goes in place of an existing light switch and has a soft switch on the front and a 433MHz RF link inside.
I am guessing that this is either logic or micro controlled, thus needing 3.3 or 5V DC power, the RF receivers are typically 5 or 12V and I guess they are using a relay to switch the contacts.

My question is, bearing in mind how small these switches are, how on earth (no pun intended) do they get DC from a single AC hot (live) wire without using some monstrosity of electronics?  I guess rectification must come into play, and maybe some kind of transformer? Albeit a small one or perhaps not isolated at all, but how they they do it without a neutral wire?

Light switches generally work on the principal that you have a constant live and a switched live (which goes to the lamp live) and Neutral is already connected at light source, I could understand if the light switch was earthed as that can act as a return, or if the light was switched on, there is a path to Neutral through the lamp, but when the lamp is switched off, or the relay is open, I can't seem to figure out how they (and other companies) did it.

Any shedding of light greatly appreciated!

[cyr]

AFAIK the in-line / 2-wire dimmers/switches are never fully off, they draw enough current through the load to keep running.

[Wilksey]

Hi,

Thats understandable as the RF link needs to be kept alive at all times, but I am wondering HOW they do it.

[Richard Crowley]

Of course, the circuits are "hot" and not safe to touch, so they are well insulated inside various plastic pieces.
They often use capacitors to "drop" the current down to the low value needed by the circuits, and voltage regulation is typically a simile Zener diode shunt. As Cyr said, they use such low power that they can "steal" current through the load when they are ostensibly "off" or "open circuit".

[Richard Crowley]

LowPower Labs (makers of my favorite wireless Arduino clone, "Moteino") make a similar product, a replacement for a mechanical wall switch called "Switch Mote" http://lowpowerlab.com/

Because of the regulatory requirements for isolated and certified components, he uses a tiny commercial power module from Recom:  http://www.recom-power.com/pdf/Powerline-AC-DC/RAC01_02-SC.pdf

[Phaedrus]

Presumably they use a capacitive divider, maybe followed by a schottky and some kind of regulation (zener, TL431, LDO, I dunno). I haven't taken one apart, but that seems reasonable as long as it's a low-power application.

[gxti]

SwitchMote has a neutral connection, so that doesn't apply to the original question. It's certainly the easiest way to make things work.

I don't know exactly how no-neutral power schemes are implemented, but it's easy enough to guess. When the switch is open you let a little bit of current through, and when the switch is closed you add a little bit of resistance (or use the resistance of the switching element) to get a voltage drop you can boost, or maybe use a current transformer. The concept is called "power stealing", you can find a few patents if you search on that term. It's also common in smart thermostats, which do the same thing on 24VAC thermostat control lines. Thermostats do have the benefit of always cycling periodically so they can get away with only stealing power while on.

[German_EE]

The following link may provide some insight into methods used:

http://ww1.microchip.com/downloads/en/AppNotes/00954A.pdf

As for running a supply such as this without a neutral connection, it might be possible to wire the supply in series on the live wire using some sort of fixed voltage drop but I would not like to try this. Wiring the supply between live and earth will (or should) trip the ELCB

[Wilksey]

Hi,

The Microchip AN shows neutral, so not what I was after, thanks anyways though.

I can't find any good circuit diagrams on how one might go about using just the hot line to get power, how does it reference ground?

I searched for power stealing, and I got some odd results

[gxti]

It's referenced to neutral by way of the terminal that the load is connected to. The circuit is: hot -> switch -> lamp -> neutral. When the switch is open, its two terminals see the full mains voltage. You can draw a milliamp from it and it will flow through the lamp but won't be enough to light it. If you draw 100mA however it would be enough to turn on the lamp.

[jlmoon]

I could understand if the light switch was earthed as that can act as a return,

I would hope someone wouldn't use "earthed" as a return.. A BIG NO NO!
Ground rule # 1 .. never use "earth ground or referenced earthed" as a return line for a live circuit!
... rule # 2 .. if confused go back to rule #1.. period

[jlmoon]

It's referenced to neutral by way of the terminal that the load is connected to. The circuit is: hot -> switch -> lamp -> neutral. When the switch is open, its two terminals see the full mains voltage. You can draw a milliamp from it and it will flow through the lamp but won't be enough to light it. If you draw 100mA however it would be enough to turn on the lamp.

and needless to say would stop functioning if you unscrewed the load (lamp) from it's receptacle.

[eneuro]

It's referenced to neutral by way of the terminal that the load is connected to.

Isn't classic 230VAC light dimmer works like this? 

In schematics below potentiometer was replaced by LDR resistor and triac phase offset is controlled via external circuit low voltage driving light LED 

[Wilksey]

It is a bit naughty to use the earth as a return, but people do operate this way! I am not saying I would do this though.

I mean, realistically, there appears to be no earth or neutral in the Itead smart switch, I assume they are using a switch (SCR or something) or a relay to physically do the switching of the constant live to switched live, grabbing a few mA would not be enough to turn this on would it?

Maybe there is a hidden battery? lol

[jlmoon]

It is a bit naughty to use the earth as a return, but people do operate this way! I am not saying I would do this though.

I mean, realistically, there appears to be no earth or neutral in the Itead smart switch, I assume they are using a switch (SCR or something) or a relay to physically do the switching of the constant live to switched live, grabbing a few mA would not be enough to turn this on would it?

Maybe there is a hidden battery? lol

I think the concept is similar to throwing a long wire parallel to the >300KV transmission lines .. like a fence wire and tuning for max V.. linear transformer if you will.

[Wilksey]

I notice a lightwaveRF switch also just has the 2 wires going to it.

I am half tempted to buy one and see what they do inside, but worried I might be horrified by the way they do it, maybe not so the lightware, but the Itead.

The installation guide for the lightwave doesn't really show much, but I know it works without neutral, and the installation video I saw the guy removed the earth and the switch was powered down until the breaker was flipped back on.

Does anybody know of any good websites that might explain the concept of grabbing power from the single live wire (+ return to load wire), I still can't think how they can get enough juice to turn on the relay or switcher that ultimately turns on the load (lamp in this case).

[eneuro]

I still can't think how they can get enough juice to turn on the relay or switcher that ultimately turns on the load (lamp in this case).

There are endless possibilities and it only depends how often one needs switch ON/OFF load and how powerfull it is 
For example just made simple prototype using random unknown ferrite filter core in this very tinny SMPS driven by limited to 0.1A 5V input power in H bridge controlled by 2 output pins of ATTiny85 connected to inverted totem pole BC327/BC337 at output stage.

There is only about 5 turns on its primary additionaly current limited by 2x27R 1W resistors while I do not have specs of this ferrite core, however one of its 6 galvanic isolated outputs charges 220uF capacitor by fast 1N4148 diode and 100R resistor in series to 12.2V. When additional 100R load is connected to this capacitor I have 8mA. White power led with 2x1k resistors has 1mA current.

One may think that 1mA is very small current, but it is a lot of charge 
Now lets take two 600V mosfetsand make AC switch like this to turn on/off our load "magically".

We have no neutral, only "live wire" in out primary SMPS and very small AC current and quite small controll current from its output, but... those mosfets used might have about 60nC total gate charge each, so to switch two of them 1000 times per second we need in average... 0.000000120×1000= 0.00012 A = 0.12 mA 
So, my magic SMPS can do strange things if one do not know that I simply charged quite big capacitor and one could use supercapacitor and do other things even when no curent flows in this one primary wire 

BTW: Buy this magic thing and send to Dave for teardown or do it yourself and let us know what is inside this black box.
It is  always fun to see how something is made 

[Wilksey]

Hi,

That looks dangerous 

For example, a few power relays I have looked at in the past operate off of 5-12V (depending on range) and can take anything from 15mA to 75mA, so i would have thought 100mA would be required for "safe" operation?