This pair of inverters was purchased off ebay from a seller in Tasmania. It is not clear what exactly they were intended to power. The 1H60668 is clearly not something for ordinary domestic use. The battery connector is a type that was used with AWA Carphone power supplies. The ebay seller unfortunately did not send the documentation with the inverters shown in the ad, although he insisted that he had done so. All I had were the ebay photos of part of the documentation. Sometime later, I discovered the inverter was shown in AWA's "Mobile Radio-Telephone" catalog from the mid 1960's. Alas, it is still not clear what it is meant to be used for, and is merely listed amongst other two way radio equipment.

From the AWA Mobile Radio-Telephone catalog.

It was known that output was 40W at 250V 50c/s, from this and the ebay documentation, and that the input is 12V.
Perhaps the most unusual aspect is the inclusion of a mains input socket. Normally the mains input feeds the output socket, but when the mains power is not present, the inverter switches on, and the output socket is switched over to the inverter's output transformer. In this regard, it actually functions as an uninterruptable power supply. No on/off power switching is provided which supports the UPS intention. However,  there is no facility for battery charging when mains power is present.

All I have of the instructions.

The inverter is built on a steel chassis which clips into the enclosure, in the same way as the AWA Carphone power supplies. It was obvious by the presence of more than one transformer, and only two transistors, that it was a self oscillating design. It appears to be from the mid 1960's, given the components used.

Electrically, there is nothing particularly unusual about the design. With the exception of the changeover contactor and mains input socket, it is just like the first generation solid state inverters that were popular in the U.S. during the 1960's.

Oscillation occurs by virtue of the feedback from the output transformer to the smaller oscillator transformer which drives the transistor bases.
The 470 ohm resistors provide a small amount of forward bias so the transistors conduct enough to get the oscillation started. Because of slight differences in the transistors, and other component tolerances, one transistor will always start conducting before the other. In theory, with perfect components, this kind of inverter would actually not work!
The 2.2 ohm resistors limit the base current and allow for slightly different transistor characteristics. The 1N1763 diode prevents the opposite transistor, to that which is conducting, being fed with reverse base voltage.
The oscillator transformer is unusually not made by AWA, but rather by TMC, which is the Telephone Manufacturing Company, an organisation that was involved with Pye at one time. This was the most difficult part of the circuit to trace out, but it had to be done. Two of the transformer tags, pins 6 and 11 are merely used as tie points. The transformer housing is reminiscent of the popular Rola "Isocore" design.
Between the transistor collectors is an RC network to suppress switching spikes. A limitation of self oscillating inverters is that output frequency depends on the load and supply voltage. Indeed, as the tests showed, this inverter showed some variation. It would be thus unsuitable for frequency critical loads such as timers or clocks.

Chassis view. At bottom left is the oscillator transformer. To its right is the mains contactor.

The transformer primary is driven in the usual way by alternate switching of the transistors. Protection is via a 10A fuse but this is too high to protect the transformer from overload. It would likely only ever blow if a transistor short circuited. No specific reverse polarity protection is provided. The transistors are a long obsolete PNP germanium type, RCA  type 3800. This is equivalent to 2N277 and rated at 50V 25A, which is more than enough for this small inverter!
These transistors were also used in AWA's 32V television inverters.

The transformer secondary is tapped, and according to the literature, provides either 220, 230, 240, or 250V. Why this selection of voltages needs to be provided is not clear, when any 40W or smaller load is not likely to be particularly voltage sensitive anyway. The documentation does mention these voltages being peak, which is correct for a square wave. The rms is the same in this case.

There is a 240V contactor which does the changeover function. When the mains is applied, the coil pulls in the contacts, disconnecting the 12V supply and the transformer secondary, but connecting the input socket to the output. A 500mA fuse protects the mains wiring, but is not used when the inverter is operating.

A view of the transistors and output choke.

A small choke is included in the output from the transformer to smooth the waveform.

The effectiveness of the choke is clear. These waveforms show the input and output sides of the choke with a 40W resistive load.
 

Parts list. Note the had written addition referring to the RC network between the transistor collectors.

Bodgie Mods.
Present in both inverters were some capacitors which didn't quite look as they were put there by AWA. Closer investigation showed quite a dangerous arrangement with them. Inverter s/n 94 had a .47uF 630V Styroseal capacitor connected between the mains neutral and the chassis.
Additionally, there was another .47uF, this time a greencap, from the live side of the transformer secondary to the chassis.
When operating off the DC supply, with no mains connected to the input socket, this simply results in a balanced output. However, with the mains connected it is a potentially dangerous situation. Depending on the mains polarity, which incidentally was not standardised when this inverter was made, what is intended as the neutral could be the live! The reactance of the .47uF capacitor is low enough to provide a nasty shock if there is no earth connection to the chassis. Worse still is if the capacitor should break down. If the contactor had been wired with double pole switching, then it could have been possible to get away with the scheme - but it wasn't.

The large red capacitor and the greencap at the bottom left were added by a previous owner.

Inverter s/n 99 had only one capacitor added; this time a decrepit wax dipped paper type of .22uF, again between neutral and chassis. Even more likely is a breakdown with this type of capacitor.

In this inverter, an old wax paper capacitor has been added. The blue wires are connected to the mains.

Such capacitors would have been added to round off the waveform to reduce RFI, which suggests the inverters were used to power radio receivers of some sort. I removed them and tested the inverter with an HMV mid 1960's valve radio, and found no RFI problems. It was necessary to earth the 12V supply however, otherwise a buzzing could be heard through the radio.
What is visible of the parts list does not show if there were other capacitors on the output side of the transformer, and by the hand written mention of the "smoothing circuit placed between two collectors" it seems the 22R and .47uF network was a later modification. It, however, appears to be a legitimate update to the design by AWA.

In order to use one of the inverters, and seeing as the two round pin polarised connector would never likely to appear, I installed binding posts for the 12V input.

Frequency variation under load. These waveforms were taken between the transistor collectors.

With the 40W load, frequency was 54.3Hz and under no load it increased to 61.1Hz.
 
 

Load	Input Current	Output Volts	Efficiency
0	530mA	289V	0%
15W	1.75A	254V	68%
40W	3.7A	216V	86%

As can be seen from the above table, efficiency improves considerably at full load. Regulation is adequate for the loads likely to be used. Very low power loads (under 15W) are likely to be those where the supply is rectified to provide DC, and so the apparently excessive AC output voltage would actually be to their advantage.