Why your cheap mains inverter could kill you
A mains inverter is a device that converts the DC power provided
by a battery (often as 12V car battery) to an AC mains supply.
These devices can be very useful, and are generally safe if used
carefully and as designed. However, all inverters -- particularly
inexpensive ones -- create electric shock hazards that are not
obvious.
These hazards are most troubling when the inverter is used in
a way for which it was not really intended -- for example,
in a permanent
or semi-permanent installation, or in environmentally hostile conditions.
I see a lot of discussion of this topic on web forums, and it is usually ill-informed. In this article I will illustrate in detail why the shock hazards exist, and why the steps that are often recommended to mitigate them will not work reliably.
Note:
In this article, I'm only discussing electric shock hazards. Careless use of inverters also creates an overheating and possibly a fire hazard. It's important to operate these devices within their design limits.
The problem with cheap inverters
Mains inverters with nominal power ratings of several kilowatts are available for less than £100. These inverters are invariably "unisolated" (or "non-isolated"). This means that there is a direct electrical connection between the battery side of the inverter, and the mains side. If you attach an AC voltmeter between either of the battery pins, and either of the live/neutral pins, you'll measure a voltage.
Depending on where you measure, and the design of the inverter, this voltage will probably be in the range 60-200V -- at least, that's what I've measured when I've tested these devices. You might find that one of the input terminals is directly connected to one of the mains terminals -- the negative battery terminal to the neutral mains terminal, for example. In that case, you'll read no voltage between negative and neutral, and probably 200V between positive and live.
It's also important to understand that, while a domestic mains socket has live, neutral, and earth (or ground) pins, these designations are meaningless for an inverter. The earth pin on the mains outlet of a cheap inverter is probably not connected to anything. The notion of "live" makes sense in a permanent electrical installation, where the neutral pin is at roughly the same potential as the Earth (the lump of rock we're standing on). The "live" pin is live because its voltage is different from that of the Earth. In Europe we call that difference "230 volts", but all that matters in this context is that it's large enough to hurt. With an inverter, though, you can't read anything at all into which mains pin is "live" and which is "neutral" -- they might as well be called "1" and "2".
In summary, "unisolated" means that current can, and will, flow between the mains terminals and the battery terminals of a cheap inverter. Why that's a big deal is something I will return to shortly.
An "isolated" (or "galvanically isolated") inverter has a transformer (magnetic) coupling between the battery side and mains side. This isolation solves some, but not all, the electrical safety problems I will describe in this article.
How do you know whether the isolator you bought from eBay is is isolated or not? The glib answer is that if you can lift it with one hand, it isn't isolated. Also, if you didn't have to think very hard about the expenditure, it's probably not isolated. An isolated inverter will cost 10-100 times the price of an unisolated one.
The better answer is that you'll have to ask the manufacturer -- advertisements often do not make it clear. I think that's because professional-grade inverters, designed for permanent installation by a specialist, are usually isolated, so this isn't stated. Cheap inverters are always unisolated, and this isn't stated either, because it's not a selling point.
If there's any doubt, you should assume that your inverter is unisolated, and use it accordingly.
How to use a cheap, unisolated inverter safely
The figure below shows the only way in which it is reasonably safe to use an unisolated mains inverter. Here the battery and the inverter are completely isolated from any kind of ground connection (by the rubber mats). The user can connect exactly one appliance, directly to the socket on the inverter. Of course, nobody's going to use real rubber mats; but careful operation (e.g., in completely dry conditions away from other metallic objects) can approximate to such a configuration. This set-up is often referred to as a "floating" supply, meaning that the mains pins are not constrained to take up any particular voltage with reference to earth -- there is no earth.
Now, I've shown the user -- the blue fellow on the right -- sticking his fingers directly into the socket, and nobody would be daft enough to do that. The point I'm trying to make, however, is that this is a safe configuration. It's unlikely that any other fault condition will expose Bluey to a worse hazard than direct connection to the mains outlet; if that's safe, everything else is probably safe.
Why is this safe? To get an electrical shock, an electric current has to flow through the body. The body has to be part of an electric circuit. In the configuration shown above, the physical separation between the electrical equipment, and the ground, ensure that no such circuit can be formed. If Bluey sticks his finger in the socket, what will actually happen is that his body will pull the mains socket down to earth potential. The rest of the equipment, including the battery, will take on some voltage relative to that. Still, that won't affect Bluey (so long as he is the only person involved in the installation).
How to use an unisolated inverter unsafely
So that's the safest way to use a cheap inverter: no connection of any kind to anything except a single electrical appliance -- a fully-floating supply. So now let's look at the most dangerous way to use such a device.
In this configuration, some part of the electrical installation is deliberately earthed, perhaps using an earthing stake. The fact that the earthing connection short-circuits the rubber mats should give us a clue that the arrangement is likely to be problematic.
Again, I've shown Bluey sticking his fingers into the mains socket, which nobody would do. However, an electrical short-circuit inside a metal device (an electric drill, for example), will have exactly the same effect -- the user will be left holding a "live" part.
Remember that "live" and "neutral" are dubious terms when using an inverter. Both the "live" and "neutral" pins can potentially carry a dangerous voltage in the "earthed" configuration, so it doesn't matter whether the putative fault is on the live connection or the neutral.
By earthing the battery (or any other part of the system) we have created exactly the electrical circuit that the rubber mats were intended to prevent. Current can now flow from the mains socket, through the body of Bluey, through the ground, and back to the battery via the earth stake.
People do use their inverters like this, because they have some notion that "earthing" is a good thing. You can even buy portable earth stakes for exactly this purpose. The kind of danger that is created by careless earthing depends on what, exactly, is earthed; but it will never be a good practice with a cheap inverter.
How bad the danger is depends on the conditions. If Bluey is wearing rubber boots, and standing on dry ground, he'll probably get away with a tingle. If it's wet, and he's barefoot, he could be in trouble.
In practice, though, you don't need to drive an earth stake to create an inadvertent path for fault current. Suppose the battery, or the inverter, are standing on wet ground. In such a case, you don't even need the earth stake -- you've directly earthed the equipment via the water.
Or consider the case where the battery is actually in a car -- this is probably a very common application for a mains inverter. The car doesn't have rubber mats, but it does have rubber tyres. These tyres isolate the inverter from the ground and are, in effect, safety features.
But suppose somebody is standing outside the car, and leaning on it. That person is creating an additional path for fault current, bypassing the insulation of the rubber tyres. This scenario is not as dangerous as deliberate earthing, but it's not entirely safe, either. When cars are provided with built-in mains inverters, it's really only safe to use them inside the vehicle.
In short, real-world applications of mains inverters typically fall between the "best" and "worst" scenarios I've described above. Nobody is going to stand their equipment on a rubber mat, but the nearer we get to that configuration, the safer we will be.
The dilemma of earth-neutral bonding
If earthing is so dangerous, why is it so ubiquitous in permanent electrical installations? There are a number of reasons for this, but we only need consider one of them here. If the electrical equipment has an earthed enclosure, then if a fault occurs in the equipment, such that an electrical part short-circuits to the enclosure, the resulting high fault current will blow a fuse (or trip a breaker).
You'll notice, perhaps, that none of the electrical faults I showed above will generate enough current to trip a fuse. A few tens of milliamps flowing through the human body could be quite dangerous, but we need amps to blow the fuse.
It is for this reason, I think, that some people recommending bonding the earth and neutral connections of an inverter. You'll need to make a piece of equipment to do this, which should be a red flag from the start; but it's common advice. Let's see why this might be considered a safety feature.
In the diagram above, the inverter's neutral and earth are connected together, and the mains socket is powering a piece of equipment with a metal enclosure. The enclosure will be connected to the earth of the mains plug (because this is necessary for safe usage in a permanent electrical installation). The red line shows the faulty short-circuit between the live input and the case.
The plug of the equipment has a fuse (and, if it doesn't, there will be a fuse or a breaker in the inverter itself). It should be clear from the diagram that there is a low-resistance path from the inverter's live to the inverter's neutral, sensible via the sensible case of the equipment. This will blow the fuse (or trip the breaker) and render everything safe -- sort of. At least it will stop us using this faulty equipment until the defect is repaired.
This earth-neutral bond looks like a good idea at first sight; to see why it might not be, we have to consider what happens in the no fault scenario.
In the diagram below, there is no electrical fault in the equipment, but somebody (who is standing on the ground) has hold of it. The equipment is, perhaps, a power tool, which the user is likely to grip tightly.
Bluey on the left can blithely carry on using his electric drill, until somebody else -- who is also standing on the ground -- touches any other part of the electrical system. Because the inverter is unisolated, there is a path for current through both the human bodies, via the Earth.
To be fair, this is probably a less likely fault scenario than the one that arises if the equipment is operated in damp conditions. But that fault scenario can be present as well.
In short, the earth-neutral bond replaces one danger -- not detecting that there is a fault in an appliance -- with a different danger -- unexpected voltages raised elsewhere in the system.
Earth-neutral bonding with an earth stake
Suppose we bond the earth and neutral (as above) and then connect the combined earth and neutral to true earth, using an earth stake. In this configuration the fellow on the right will be fine -- his equipment enclosure will now be at earth potential, so he cannot be shocked. But the person on the left will receive double the voltage with respect to earth. So that's hardly an improvement in safety.
In fact, the use of an earth stake means that the car battery is live, with respect to earth, even if the person on the right is not even present. The earth stake is creating the risk he previously created -- a path to Earth for fault currents -- but much more effectively. Not only is it not an improvement in safety, it's potentially lethal.
The use of an earth stake might make RCD protection (see below) more reliable, but it's still wrong.
A note on RCDs and similar safety devices
People often ask whether it's a good idea to use an earth-fault detector like an RCD on the output of an inverter. If the inverter system is fully floating, as in the first picture in this article, then an RCD won't work, and will be unnecessary. In fact, it won't work because it is unnecessary. Since there's no path for fault currents, there's nothing to trip the RCD.
What about a situation where the supply is supposed to be floating, but is inadvertently earthed by, for example, being run in the wet? In such a situation an RCD might trip before somebody is hurt -- it depends somewhat on the design of the device. In general, RCDs sense an imbalance in current between the mains supply terminals; if these currents are unequal, then some current is going somewhere it should not.
The problem is that RCDs are not really designed to work on floating supplies where, as I've said, they are unnecessary. A particular device might work, but I'm not sure it should be relied on. In any case, the "test" button on the RCD won't work, because there is no path for the simulated fault current, and that alone makes the set-up unsafe.
What about the situation where the inverter's earth and neutral are bonded? The RCD probably will detect the fault condition I showed in the picture above, where the path for fault current is through two people, one of whom is (for example) leaning on the body of a car that is powering the inverter.
The RCD might trip in this situation although, again, this isn't an application for which it will have been designed. But that's not really the point -- the bonded earth-neutral creates a shock risk even in a no-fault situation. It does not need any kind of equipment failure for there to be a hazard. Maybe, if we're lucky, the RCD will prevent serious harm; but nobody ought to be exposed to an electric shock from car bodywork, particularly in a no-fault situation.
The problem of distribution
At the start of this article, I said that part of running an inverter safely was to supply exactly one appliance. If the supply is "floating", why does it matter how many appliances are connected?
Remember that the floating supply is safe so long as nothing is earthed. Earthing any part of the system potentially creates a problem. The more appliances are connected, and the more people who use them, the more likely it is that the "no earth" rule will be compromised.
Suppose a person is using an electrical drill with a fault, such that there is a short-circuit from the motor to the drill body. Although he is "safe" using this faulty appliance with a floating supply, the fault will not be detected; so long as he uses the appliance, the system is no longer a floating supply.
A serious hazard is unlikely; but the more people and appliances are involved, the greater is the risk. For my part, I think that powering a bunch of cellphone chargers (for example) from an inverter is an acceptable risk. Why? Because each charger contains its own isolating transformer (if it complies with European safety standards). The earth pin on the plug is not connected to anything. Still, it's difficult to tell what kind of equipment can (relatively) safely be used this way.
Why an inverter is different to a generator (and why this matters)
In a sense, an inverter is similar to a portable generator: it provides a mains supply and can be operated in a fully floating configuration. There's a lot more information and discussion about the electrical safety of generators than there is of inverters, simply because generators have been around longer.
There is a way, however, in which an unisolated inverter is quite different to a generator, meaning that any advice you might receive that applies to generators has to be treated with caution. A generator is truly self-contained -- it is electrically connected to nothing but the equipment it supplies. The same is not true with an inverter. As a minimum the inverter needs a battery supply. In some cases this may be augmented by equipment like solar panels. Whatever you do to the inverter affects everything connected to it.
The safety implications are therefore quite different. An isolated inverter is more like a generator, and can be set up in a similar way.
Isolated inverters -- are they an improvement?
An isolated inverter has a built-in isolating transformer. This means that there is no electrical connection between the input (battery) side and the output (mains) side. Power is coupled from one side to the other magnetically. It's not perfect isolation, because some electrical current can leak across the transformer by capacitive coupling. Unless you're powering medical equipment, though, this small amount of leakage is unlikely to be problematic.
When used in a fully floating configuration, the isolated inverter offers no advantage over an unisolated one. The "isolation" is provided by the fact that there is no earth connection. However, when a totally earth-free operating environment cannot be guaranteed, the safety benefit of the isolated inverter quickly becomes apparent. For example, even if the battery side of the inverter is earthed (intentionally or unintentionally), there is still no path for fault currents across the inverter. The mains side of the inverter is still a floating supply. If the neutral and earth terminals of the mains output become bonded (accidentally or intentionally), then the battery side will still be floating, and cannot cause a shock.
An alternative to using an isolated inverter is to couple an unisolated inverter to a separate isolating transformer. These devices are widely available: they are the yellow and blue buckets often seen on construction sites. An isolated inverter will usually be smaller and lighter than an isolating transformer with the same power rating, because the transformer will be supplied with high-frequency AC power, rather than mains frequency. Why this is the case is a matter of electromagnetic theory, and beyond the scope of this article.
In practice, isolated inverters are more commonly used in permanent installations, where the greater range of risks justifies the additional expense.
Summary
Electrical safety is a complicated subject. In this article I've touched on shock hazards, and tried to show why, where they exist, they are not easy to mitigate.
Of course, shock hazards are not the only risks associated with mains inverters. It's easy to overheat an inverter or, more likely, its associated cabling. With many vendors, it's difficult to know how far their specifications can be trusted. Even if you can avoid the shock hazards, there are plenty of other things to take care over.
