Hi. Long time no see.

First of all, let me tell you I do not intend to build this. So this is a theoretical discussion. And I'm not an expert on this, so don't try to build it.

The thing is all transformerless supplies are considered dangerous and I really don't understand why there's so much fuss about this. So I'm trying to design a schematic (attached to this post) that, along with proper building, would provide enough safety.

The green box labeled "Permanent insulation" contains all the components that should be encased in something like a permanent plastic case that cannot be opened. If something blows up inside (e.g. the fuse), you're supposed to throw it away and buy another one.

The schematic goes like this... C1 and C2 block any DC component that might be present on the power lines, just like a transformer. D1-D3 are avalanche diodes (somewhat like higher voltage Zeners): D1 and D2 step the voltage down, while D3 blows the fuse in case:

1. There's a voltage spike on the power lines.
2. Any other components fail and the voltage would rise to dangerous levels.

Outside the box, we filter the voltage in a way similar to other supplies. The input waveforms to the filter are different than in a regular supply (due to the way we lower the voltage with D1 and D2), but it can be worked around it.

As far as I can tell, the user would not be in any danger, except if D3 isn't properly connected or F1 does not blow and another failure occurs, which can't be handled anymore. Multiple D3 diodes could provide enough redundancy to reduce the probability of dangerous situations.

What do you think about this? Am I right?

 

It's not safe. There is no real isolation from the mains. Ground yourself and touch the return path in this circuit for a surprise.
Might be better to drop the voltage by R1 and a resistor in the return path.
At least include R1 in the enclosed area.
Large caps may be required.
IMO, use a transformer.

 

It's not safe. There is no isolation from the mains.

What if there's no isolation? Isolation itself does not make it safer unless you use a step-down transformer.

At least include R1 in the enclosed area.
Large caps may be required.

Yeah, R1 would better be inside, thanks.

 

It's not safe. There is no real isolation from the mains. Ground yourself and touch the return path in this circuit for a surprise.

The caps provide that kind of isolation, +/- can't really float at high DC component.

 

You get an "AC" shock. Quite unpleasant.

 

all transformerless supplies are considered dangerous and I really don't understand why there's so much fuss about this

If you really do not understand why it is dangerous to use a non-isolated power supply, why then do you imagine you can design one to be safe? The fuse provides no protection to the user in case of either capacitor failing. Further, those capacitors do not adequately limit line voltage from a user in contact with ground under any operation.

The best you can expect with this design is an unpleasant shock from the terminals. The level of that shock may be lethal, depending on the size of the capacitors. Any failure in the limiting capacitors makes the terminals carry an absolutely lethal shock hazard.

With no more regulation than you get from the zeners, a wall transformer will do as well, and with safety.

As far as I can tell, the user would not be in any danger, except if D3 isn't properly connected or F1 does not blow and another failure occurs, which can't be handled anymore. Multiple D3 diodes could provide enough redundancy to reduce the probability of dangerous situations

A transformer can never place a user in danger of a lethal shock. Your fuse provides no protection whatever. That "permanent insulation" is no isolation from the line voltage.

If your life is worth less than the 3 euro for a wall transformer, then that is your decision to make. Your design is inherently a lethal shock hazard, and cannot be made safe to use under any circumstances.

 

The hazard exists when you have your left hand touching the frame of some equipment that is grounded to the same place as CONN1 pin 3, and then you touch pin 1 or 2 of CONN2 with your right hand. 9 out of 10 people would not expect to get a shock by doing this.

In this case they would get a shock because there is no transformer to isolate the ground return path so the line voltage on CONN1 pin 1 will pass through your circuit via C1 to either pin 1 or 2 of CONN2.

If the impedance of C1 at 60Hz is high enough then the shock will be mild, but the circuit would work right either.

Note: CONN1 pin 2 (neutral) is also grounded at the service panel.

 

When considering safety in design there are a few considerations to bear in mind.

1) People do the wrong thing. Some deliberately, some inadvertantly, some carelessly.
They connect things back to front, miss out connections, form bad connections etc etc.
Sometimes they even disable designed in protection.

2) Whilst you may be the perfect user, you cannot control what someone else may do when you are not there or have passed the equipment on.

3) Even though the equipment itself may be safe it may be connected in such a manner as to produce a hazard elsewhere - eg undersized supply cables leading to fire danger.

4) If you connect directly to the mains supply you are connecting to an energy source of massive output capacity in the event of a fault current being drawn.

5) If you connect via a true isolation device such as a transformer the fault current output is limited by the transformer. This is why you can have shaver sockets in bathrooms.

This is why we design electrical equipment with guards, shields, covers, recessed off switches and outstanding off switches etc etc.

 

Thanks for your replies so far.

After a longish discussion on Freenode's ##electronics, I'm afraid there's no simple solution, like beenthere suggested. The biggest hurdle is removing potential difference from hot/neutral to ground, a thing which caps won't do (or they will, but only if they're way too small).

The only solution that remains is switching (using analog switches for example): connect rectified mains to caps, disconnect, connect caps to load, disconnect, repeat. But the issue is that it's kinda complex (which makes safety harder to attain).

Otherwise, you need something that blocks DC without storing a charge or bleeding it through until it's charged. I'm unsure if something else than a transformer can do it. Obviously, you can't/shouldn't pull neutral to ground just before the equipment.

The green box that's supposed to be sealed is safe enough as long as outside connections can't electrocute users. Transformers aren't safe either if you break the insulation and touch hot wires. So it matters what you put inside the box.

Any other ideas besides the one involving switching?

 

For a 'small' power supply, you can fit an AC powered light source surrounded by solar cells inside your "green box"
Miguel

 

Return to a classic of the past; a motor-generator set.

 

Thanks. Nice ideas, even if not really better than a transformer in most applications.

The Wikipedia article on galvanic isolation makes me think "isolation" is not the proper word to describe the transformer, since, as they say, capacitors also provide galvanic isolation. Yet resulting transients are dangerous.

I didn't make any calculations, but I suppose reducing those transients entails high impedance at 50Hz. On the other hand, transformers also pass some DC, so I'd appreciate a comment on this.

I wish there was something like a gyrator transformer . Well, we've still got that switching capacitor idea.

 

On the other hand, transformers also pass some DC, so I'd appreciate a comment on this.

Uhh, no. There is no circumstance where a transformer can pass DC voltages or currents.

 

Uhh, no. There is no circumstance where a transformer can pass DC voltages or currents.

AFAIK, an infinite inductance transformer could work on DC. Even on real transformers, small transients may pass, just like a capacitor while it "fills up" (though the waveform does not look like DC in either case). Can't find a reputable source for this, but here's something:

http://wiki.answers.com/Q/Why_doesn't_a_transformer_work_on_a_DC_supply

The difference is transformers that pass very little DC do not impede the AC like capacitors, I think.

 

By definition, if a signal passes through it is AC. We are talking real world here, but even an "infinite inductance" will need an infinate DC power supply, lightning will follow shortly.

Futhermore, you won't even get a haverwave, a DC pulse. If it goes through there will be a rebound. A transformer is not an inductor or capacitor, where you can get a DC signal through. If you charge the primary up, you can have a pulse, but sooner or later you will discharge the primary, and there will be a equal but opposite pulse. You can hide the secondary pulse by stretching it out, but it is there none the less.

This is core to the nature of transformers.

 

AFAIK, an infinite inductance transformer could work on DC

Utter rubbish.

Transformers work by inducing a current in the secondary.
This can only be done by a varying magnetic field. A steady one, however large, induces zero current.
To obtain a varying magnetic field you require a varying primary drive current, or AC by another name.

 

The only way a battery (DC) voltage source will work a secondary current in a transformer is if the DC voltage source is interupted, and never stays in steady state.

But a steady state (DC) voltage source cannot produce a transformer action, because it takes a change in current velocity to induce a voltage, wether its a acceleration or decceleration, in current flow.

 

This seems to have played out. Any direct connection to a power line must be considered as a lethal shock hazard.