Is this transformer-less power supply a good idea? [duplicate]

Question

This is based on a design I found somewhere , I just wanted to know whether this is a good idea ,I intend to use this to power a microcontroller and a 70 ohm 5 V relay . The cap c1 limits the current to about 200 mA , well below the 500 mA rating of my zener diode . Without the zener diode , it looks like there would be about 7 volts across r1 which the zener should regulate . Ripples might be a problem though but I don't think it should affect switching that much , I'll also add a flyback diode to this .

Also , I get it that no insulation is a bad idea in general but I need this to be cheap and small . If anyone has got any better idea , I'll be happy to hear .

edit: C1 should be 3 uF

Answer 1

I've used this type of design before but it should only be used with care and packaged in such a way so that nobody can easily access any connections - all connections have to be regarded as lethal.

I would add a current limiting resistor in series with the input capacitor because you cannot rule out fast transients occurring on the incoming AC supply and this could cause very large current peaks and potentially destroy the zener and then you have a knock-on problem if it goes open circuit.

Clearly also your diodes need to be rated at the correct reverse potential for AC mains in your neck of the woods.

C1 does seem excessive at 3uF - it's impedance at 50 Hz will be about 1000 ohms and, from a 230 V AC supply will draw a standing current of about 220 mA RMS - that's quite a lot of bad power factor load but if it's a one-off then probably OK.

Answer 2

Subject to certain precautions, it's not too bad. As long as you appreciate the lack of isolation means it must be double isolated from any possibility of being touched.

C1 must be a mains rated type, like an X2. They have specially designed electrodes to survive the 1500v transient over-voltages that happen all the time on mains.

C1 should have a small resistor in series with it to limit the inrush current that could flow when you first turn on. It protects the diode bridge from overcurrent. If your bridge consists of something like 1N400x series diodes, they can stand 30A for one half mains cycle, so an 8ohm resistor would be enough, and not get too hot under normal operating conditions.

As Dave Tweed points out below, a large resistor across C1 is also a good idea, to bleed off charge, to avoid surprising you a few minutes after disconnecting it.

The diode bridge only needs to be rated for the zener voltage. If that fails, you will lose your load and C2.

Answer 3

I think you'd be better off using a wall-mount power supply. They're widely available (my local grocery store carries them) and inexpensive (< $10 US). If "small" is important to you, you might be surprised at how big an across-the-line rated capacitor is. You've probably seen them in home appliances without noticing their values. They're about the same size as the smallest power transformers.

Adding a few thoughts:

All the current estimates - in the question and in the answers - are too high. It's easy to calculate the maximum current that would be drawn by a worst-case load (a short-circuit). In each half cycle the voltage across C1 changes by 325 volts (230 Vrms is 325 Vp), which pushes 975 microcoulombs through the capacitor. This happens every 10 msec., so the average DC current is 97.5 ma. The actual current will be less, but not a lot less, since the peak voltage seen by C1 is reduced by two diode drops and 5.5 volts.

Troubleshooting the circuit could be exciting. For example, if you probe either side of Dz (or its parallel cohorts) with a scope, you'll see a (more or less) square wave, with an average value depending on which side of Dz you chose. Don't even think of grounding the probe.

There's a similar circuit that uses fewer components: connect the load side of C1 to the cathode of a zener whose anode is grounded (actually neutraled - neutered?). Then, getting back to the zener's cathode, connect it to the anode of a rectifier whose cathode supplies the load. The load return would be neutral. The ripple would be higher, for a given C2, but it saves three rectifiers. Not a big parts cost savings, but it uses less room on the schematic, which could lead to using less paper. Spotted owls will thank you.

Electric shock is not a big consideration. All modern power tools, for example, have plastic cases, and don't use the ground connections provided by electrical outlets.

Damage from power line transients is, really, a significant consideration. Neil_UK mentioned 1500 volts. I read 6000 volts somewhere.

The biggest worry is some component failing and starting a fire. That hazard, and the other two just mentioned, are best dealt with by making them someone else's problem. That was the reason for my first round answer: wall-mount PS.

Your concerns with size and cost suggest you're considering producing more than a handful. You should start your design by reviewing requirements for certification by agencies like UL. It used to be said, in antiquity (1970s) that the only requirement for starting a power supply business was having a two car garage and only one car. That didn't mean that there were a lot of PS makers, but that the scope of the effort was usually underestimated.

Answer 4

This kind of supply is acceptable IF everything that is being powered by it is always well insulated from the user. This is the case, for example, in a LED lightbulb where the whole thing is sealed.

A shunt regulator is also rather wasteful with power. You said the current is limited to about 200 mA, but is actually much less. I get 50-60 mA. If it really was 200 mA, then this would be constantly dissipating (200 mA)(5.5 V) = 1.1 W. That may not be a big deal in terms of overall energy consumption, but is enough that you have to think about the thermal issues. That's more, for example, than a TO-220 package in free air, especially inside a small closed box, can safely dissipate.

Again, you need to look at C1 more carefully. The 70 Ω load will draw 79 mA at 5.5 V, but you're only going to get about 54 mA with the 750 nF you show.

Another issue is that this kind of capacitive coupling transfers current largely as a function of the derivative of the voltage. That's fine when the voltage is a sine, and most of the time it will be. However, spikes happen. These have large dV/dt, so can cause large current surges thru the capacitor. A resistor in series with the cap can help with this, but it's a tradeoff with causing even more dissipation during normal operation.