12VDC - 240VAC 50W Inverter.

50W continous or 60W for short periods at 240V AC is provided by this inverter from a 12V DC supply.

Continuing from the other articles on this site about vibrator power supplies, the inverter described here is a practical example of this method of DC voltage conversion.
Whilst most vibrator power supplies are used for DC to DC conversion; e.g.; for a  car radio, there is nothing to stop one using the AC at transformer secondary directly. With the right turns ratio of the power transformer, it then becomes possible to provide AC mains voltage from a battery.
Some background first:

DC to AC conversion.
As transformers rely on a constantly changing magnetic field to operate it is not possible to convert DC voltages without ancillary parts. When the first vibrator power supplies were developed, they were based on Victorian era induction coils operating in half wave. In fact, some constructional articles describe using Model T Ford ignition coils as B battery eliminators. Indeed, my first inverter experiments were along these lines. This mode of operation has disadvantages in that DC is present in the transformer core, reducing efficiency, and the type of vibrator used had to interrupt a high current when starting. Additionally, the loading was critical for proper operation. All these problems were overcome when full wave operation was implemented. See this article for more on half wave vibrators.
The basic method to create the changing magnetic field is shown below and is fundamental to all inverter designs:

Basic inverter circuit shows method of DC to AC conversion and voltage step up.

It can be seen that as the switch connects to the upper contact that the magnetic field created in the primary coil will be opposite to that when the switch moves to the lower contact.
Hence, an alternating magnetic field is produced, and normal transformer action allows this to be converted to the required voltage. This can be higher or lower than the DC supply voltage. The switching must be done rapidly and continuously, which obviously means some kind of automatic switch. It is obviously totally impractical for the user to flick back and forth a two way switch at the speed required! A motor driven switch can, and has been used, but the more popular alternative is the vibrator. This works on the same principle as a buzzer, but with much higher quality construction and predictable performance.
Since the 1960's, transistors have been used to perform the switching, and these days, MOSFETS are used. These don't produce the same mechanical noise, but are much more susceptible to failure due to overload or voltage spikes. They also require much more circuitry to drive them, and require the correct supply polarity.
The other alternative for DC to DC, or AC voltage conversion is a motor generator. These also go under the names of "Genemotor" and "Dynamotor". Here, a DC motor fed from the supply drives a generator on the same shaft, or more usually, on the same armature. Whilst reliable, they are noisy and less efficient. However, they don't suffer the limitations of contact carrying capacity that vibrators do, so they are used more for high power applications.

Power rating.
As can be seen from some of the other projects on this site, my house is equipped with a solar 12V DC supply, so inverters are a useful item when one wants to run mains appliances. I have also operated various 240V items in cars. I have built many inverters over the years, some vibrator, and some solid state. Here, I describe my most powerful home made vibrator inverter.

The obvious choice of vibrator was one of the dual interrupter types; Oak V6612.  The contacts can be wired in parallel, but for better current sharing are used singly with two identical primary windings. Oak non synchronous car radio type vibrators have a contact rating of 4A. To use one of these (e.g., V5123) limits the input power to 4 x 12V = 48W. In view of transformer efficiency, say 80%, this results in about 38W for the maximum output power. I have built many inverters of this kind for low power devices with excellent results, but for a 50W inverter something more powerful is required.

Data for the V6612.

Dual interrupter vibrators are manufactured identically to ordinary sychronous types in that there are two sets of contacts; one for primary switching, and the other for secondary rectification. One may assume these could be paralleled to increase the ratings. However, with the synchronous type, the secondary contacts are designed to close just after the primary contacts close, and open again just before they open. This means while the contact current rating can be doubled while they are actually closed, the actual switching current rating is still the same as one set of contacts.
Nevertheless, where a synchronous vibrator is used in a non synchronous application, it is worthwhile paralleling the unused contacts because they still make a useful contribution.

The dual interrupter vibrator has both sets of contacts adjusted to close and open with the same timing. While this may appear to double the rating, in actual fact it is somewhat less. This is because it is impossible, in practice, to have both sets of contacts open and close at exactly the same time, all the time.
Looking at the data above, the V6612 maximum input current is therefore only 4.5A. This incidentally, applies to each set of contacts switching its own primary winding. In other words, the transformer has two separate centre-tapped primaries. The reason for doing this is to provide more equal current distribution between the contacts. It should be clear that if one set of contacts has a slightly higher resistance than the other, then most of the current will flow through the lower resistance set. A further improvement in current sharing can be obtained by using two separate identical transformers with their secondaries paralleled.
Another limitation with regards to vibrator current is the input voltage. It will be noted that the current rating falls with higher voltages. This is because contact arcing occurs more easily as the supply voltage is increased.

With that information in hand for our 12V inverter, we are limited to 4.5 x 12V = 54W output, assuming 100% efficiency, and two primary windings. For intermittent duty, the current rating is 80% greater, or 8.1A. This would allow an output of 97W. The definition of "intermittent" is not given, but it would no doubt be something along the lines of two way radio transmitter use.
The transformer used with this inverter has only a single centre-tapped primary so this would derate the contact switching current. However, it is a toroidal type and this has much greater efficiency than the conventional E-I laminated type normally used.
On that basis, it would probably be safe to specify 50W for the continuous output power rating, with 60W drawn for short periods. Higher power loads which cause contact arcing are not acceptable in any circumstances.

The V6612 type has a 12V driving coil. It is quite in order to use the 6V type V6606 (along with the other series drive Oak/MSP 6V types) on 12V by means of a 27R 5W resistor in series with the driving coil.
There are of course other more powerful types of vibrator available such as the 50 cycle Van Ruyten types which can provide 100W. These are much larger and have four sets of contacts. They were used in some Australian made inverters from the 1950's up until the early 70's.

Vibrator with can removed to show operation. Correct circuit design ensures no sparking at contacts and long life. Driving coil is visible at top with vibrating reed and contacts beneath. Nut at side locks the adjustment for the driving contact.

It might be assumed that one just uses a 12-0-12V to 240V transformer in reverse. Not so, and sadly many inverter circuits appear on the internet doing just this. Their designers have obviously never tested their circuits properly, because they'll find the output voltage somewhat lower; about 180-200V instead of 240V.
First thing to consider is turns ratio. One would assume that a normal 240V to 12V transformer has a turns ratio of 20:1. In the ideal world it would, but given losses in the transformer, it has to be slightly less than this. The "12V" output is at the transformer's rated current. Run with no load, the voltage may be something like 14V, which means a turns ratio of 17:1.
Now, what happens when this transformer gets used in reverse? Assuming 12V is fed in, 204V comes out. And, that's not taking into account things like transistor saturation voltages and the supply voltage drop between battery and inverter.
For vibrator inverters, another factor comes into play, and that's the time between the contacts opening on one side and closing on the other. The transformer is fed with no power during this dead time. While the peak to peak voltage output may be correct; the RMS is less.
In practice, a 9-0-9V : 240V transformer is required. I have found over time that the conventional 240V laminated type of transformer is often a poor performer with vibrators. Some do work well, but many don't. This is evident when the idling current is higher than it should be, along with difficulty eliminating contact arcing. This is mainly a function of winding inductance and lamination design, causing a high magentising current. Transformers designed for vibrator power supplies do not have these problems.
In the modern world, I have found toroid transformers to be excellent substitutes, and possibly even more efficient than a genuine vibrator transformer. To illustrate this, I did initially use a conventional 240V to 18VCT E-I laminated transformer for this inverter, but magnetising current was about 1A, and it was impossible to eliminate the contact arcing completely. Changing to a toroid (Altronics M5109) dropped the magnetising current to an insignificant amount with no arcing.

Peak to Peak and RMS.
Vibrator and simple solid state inverters produce an output based on a square wave. This is because the switching devices are either on or off. This means that the 0 to peak output voltage is the same as the RMS. So, for a 240V square wave, the peak is also 240V. With a 240V rms sinusoidal supply, as from the mains or an alternator, the peak is 1.4142 times this, or 340V.
What is the result of this? For heating appliances light incandescent lamps or heating elements, there is no change. Resistive loads like these only care about the RMS, and in fact can work on 240V DC just as well.
For electronic loads, there can sometimes be a problem with the lower peak voltage. For example, assume a radio is powered off a square wave supply. The valve heaters will see 6.3V RMS, but the high tension DC will be about 1.4 times less voltage. This is not usually a problem as such circuits do tolerate such a loss in supply voltage fairly well. However, a television with an unregulated supply is likely to suffer a shrunken picture.
Generally, appliances with regulated supplies, especially switchmode ones, have enough regulation to compensate for this.
Induction motors and other reactive loads can be a problem with a square wave supply as the square wave is really a sine wave with infinite harmonics. And it's this power present at higher frequencies than the harmonic which causes extra losses and heating.
Note that ordinary voltmeters will give an innaccurate reading with a square wave. One that measures RMS should be used instead.

While vibrators intended specifically for 240V AC inverters do run at 50c/s, the more common radio type is 100c/s, and this is what is used here.  Having said that, there were a number of Australian commercially made inverters which were car radio vibrator based. Typical brands include Ferris, Smoothflo, Robinson, Champion, Jewel, and Bland Radio.  What were commonly sold as shaving inverters sometimes use synchronous vibrators and put out DC. This was acceptable for motor driven shavers which were quite common at the time.
Appliances not compatible with 100c/s are things like clocks that rely on mains frequency for their timekeeping, and synchronous motors. Heating appliances, or appliances where the incoming AC is rectified do not care about mains frequency.

Top view shows main components. Note the simplicity compared to solid state designs.

The Buffer Capacitor.
One most critical part, if not the most critical, is the capacitor used for tuning the transformer to the vibrator dead time period. If the transformer is not tuned properly, the vibrator contacts will arc and be destroyed, and/or the current will be excessive. The value of this capacitor can be found with an oscilloscope observing the output waveform, and an ammeter in series with DC supply.
The ideal value is one that causes minimum primary current, no arcing on the contacts, and the ideal waveform. There should be no overshoot at the rising part of the wave, and the edge should not be rounded off.
Because of the large value that would be required if the buffer capacitor is placed in the primary circuit, it is connected across the secondary instead, and the value is reduced by the square of the turns ratio. Being across the secondary means the capacitor is subjected to high peak voltage. It therefore must be of a type suitable for this kind of work. Polypropylene 2000V is the best choice. Ordinary polyester types seem to be satisfactory provided the voltage rating is at least 630V DC and they are not exposed to more than 250Vrms. Self healing types of 250V AC capacitor are not recommended because over time the dielectric burns away resulting in reduced capacitance. It goes without saying that if you have a commercially made inverter still with a paper capacitor in this position, that it will be most likely to be leaky and the vibrator will be damaged unless it is replaced.
For AC inverters like this, the buffer capacitor value has to be a compromise because of the different kinds of load that might be plugged into it. Generally, the value ends up being higher than expected. In this particular circuit, the ideal no load buffer capacitor was .015uF, but it was necessary to increase this to around .25uF to prevent contact sparking with some loads. The idling current is about 100mA higher as a result.

Vibrator inverters do not like low power factor loads. It can be seen that a capacitive load effectively detunes the transformer/buffer capacitor circuit by adding more capacitance. Inductive loads detune the circuit by adding inductance, making the buffer capacitor less effective. Most low power factor loads are inductive, so extra capacitance must be added to use them with a vibrator inverter. Some commercially made inverters recommend operating an incandescent lamp along with loads such as fans or tape recorder motors. Although wasteful, it does improve the power factor by swamping out the inductive component. Capacitive loads also exist; for example in appliances where a capacitor is used as a voltage dropper. Apart from the power factor issue, if the inverter is 100c/s, then the reactance of the capacitor decreases, passing more current than originally intended.
Ordinary fluorescent lamps with an iron cored choke are unsuitable with 100 c/s inverters because the inductance of the choke increases at the higher frequency. The tube has insufficient starting or running current.
50c/s vibrator inverters are fine with fluorescent lamps, provided they are power factor corrected.
Basically, vibrator inverters should only be used with resistive loads.

It is fairly obvious that if 12V is fed into the inverter, and 240V comes out, then any change in supply voltage will be multiplied by 20. So, if the supply drops to 10V, then output will be 200V. If supply is 14V, then the output will be 280V. Also to be taken into account is the regulation of the transformer itself.
In the real world, a "12V" supply is likely to be 11-14V depending on battery charge status and supply cable voltage drop. Ideally, the output voltage will be 230-250V.
In the worst case scenario, with 11V input, and a 50W load on the output, the voltage output should be not much less than 220V RMS. This is essentially what determined the turns ratio chosen for this particular design. Fortunately, an 18VCT to 240V transformer works out ideally, and is commonly available.
Having chosen this transformer, what about light loads (such as a 25W lamp), or when the supply is up around 14V? In this situation, the output might be a bit excessive. One way to deal with this is to have a slightly lower tapping on the transformer secondary to switch in when appropriate. This is used by many commercially made inverters.
The other method is to switch in series resistance in the supply, and this is what I did here. I found a 0.33R 20W resistor worked perfectly and easily handles the current. I could have wound some turns on the toroid and subtracted the voltage so developed, but it was just easier to install the resistor.

The circuit.

The simplicity of the circuit is obvious.

Turning now to the circuit design of this particular inverter we can see that few solid state designs can be so simple, as well as work with either input polarity.
The design can actually be simplified further. Essentially, all that is required is the vibrator, buffer capacitor, and transformer. However, to make it practical for as many uses as possible, some refinements have been added.
The incoming 12V supply is fused at 7A. This protects the vibrator and transformer should the inverter be overloaded. A non technical person might, for example, plug in a 1kW radiator.
Should one of the vibrator contacts stick, the fuse will blow before the supply wiring or transformer primary is damaged.
RF filtering is provided by the 1uF and 1.5uF capacitors. Because of the square wave nature of current draw, and the opening and closing of the vibrator contacts, RF interference occurs, unless filtering and suppression is provided. The 1uF is located where the supply cable enters the inverter and suppresses RFI leaving the inverter through the supply. The 1.5uF is connected right across the vibrator socket. Its main purpose is to suppress RFI caused by the driving coil. While these two capacitors appear to be in parallel and thus could be replaced by one, in the RF sense they couldn't be further apart. Earthing location and connecting lead inductance play an important part of RFI suppression with vibrator power supplies. These two capacitors are non polarised types to allow for either input polarity.
Unless they have designed for vibrator use, conventional laminated transformers sometimes perform poorly, so a toroid was used in this design. Not only is regulation and efficiency improved considerably, but RFI is reduced, contact arcing suppression is easier, and the inverter weighs less.
Connected across the vibrator contacts are 120R 5W resistors. Their value is not hugely critical, and varies from 100R to 330R in typical circuits. Along with the buffer capacitor, they help eliminate contact sparking by damping the relatively high peak voltages developed across the contacts. RFI is also reduced. Many circuits don't include these resistors, and they are not always required - it does depend on other factors, but their inclusion can only be beneficial. In this design, they weren't necessary as far as spark suppression was concerned, but there was an improvement in the reduction of the amount of RFI produced.
As described previously, a .33R 20W resistor can be switched in to the supply if the load is light, or supply voltage high. In practice, the switch is set to "Low" with loads of less than 25W, and the supply voltage is high.
The buffer capacitance is two series connected .47uF polypropylene types. Having two capacitors in series like this increases the overall voltage rating, and by earthing the junction of the two, the AC output is balanced with respect to earth. Furthermore, both secondary connections are bypassed to earth for RFI purposes. When powering an AM radio, the clean output of this inverter is obvious.
A neon pilot light shows the inverter is producing output.
The 240V proceeds to a normal 3 pin socket. Efficiency with a 40W load is 83%.

This type of inverter is ideally suited to resistive loads. This photo was taken before the cabinet was fitted with a handle.

For housing the inverter, I cut and bent up an aluminium box which was sprayed in green hammertone. Louvres were punched on the sides of the cover. Front and rear panel labels are made from Scotchcal which is a photosensitive aluminium which is also self adhesive. It is now no longer available.
The inverter thus has the classic 1950's look with a commercially made appearance.
A few metres of 12V flex connects to a 2 pin polarised plug for the supply.