This inverter came from eBay at a very reasonable 'buy it now' price. With it came a quantity of what were called "vibrations bulbs" and a couple of other miscellaneous parts. I had gathered from the photos that there was some kind of fault, and someone had attempted to repair it. Input is 12V, and output is 230V 40W at 50 cycles.

It is typical of Australian made vibrator inverters from the era, using a 50 cycle Van Ruyten vibrator. Smaller inverters used radio type vibrators operating at 100c/s, and several of these have been described elsewhere on the site. Van Ruyten inverters have been described elsewhere on this site; here and here.

This note stuck to the side implied problems. "A vibrator out of an old mantel radio" is actually not directly suitable. I hope the "230W" is a mistake; it should be "230V" of course.

The inverter arrived soon enough, and there were a few interesting things that caught my attention.
Construction of this inverter is exactly the same as previous models I've worked on. A chassis made up of various brackets supports the vibrator and transformer. Depending on the model, different size transformers can be accommodated without any chassis modification.
The vibrators are always the Van Ruyten UY-5 pin base type, which are available in 12, 32, and 110V. A Ring Grip domestic light switch mechanism and three pin output socket are also used in all the models. In the 100W models, a high/low output switch is provided. In the 40W model this is not required, and the same hole is used for a separate grommet for one of the power input leads.

First Impressions.
The cabinet had been repainted blue, and it seems underneath this was a coat of black. This was over the original grey hammertone. There had also been three holes drilled into the side of the cabinet, possibly to mount the inverter on something.
The screws holding the front panel to the cabinet had been replaced with modern Phillips head self tappers. I found three of the originals loose inside the cabinet.

Written on the vibrator was "Defective needs repair". Something was rattling around inside it.
I noticed that one of the capacitors had been disconnected by cutting one of its leads. The capacitor in question is a 0.1uF across the driver coil contacts. I dreaded the damage that might have occurred running the vibrator without it connected. The earthed input battery lead had obviously been replaced. (As there are no polarised components, this inverter works with negative or positive earth).
The wire between the switch and vibrator socket had droplets of liquid on it. I suspect the insulation was decomposing into something acidic, since there was rust on the casing where the wire was against it.

Note the capacitor with lead cut at the vibrator socket. Note also the droplets on the wire running along the bottom.

As far as the electrical design is concerned it was as expected, except for one thing. There is absolutely no RF filtering on the input or output. Not even just one bypass capacitor. Also, despite the name plate stating 40W output, the hand written labelling on the transformer states 50W. By the look of the transformer size, I have no doubt it would easily handle 50W, if not a bit more. The vibrator is rated for at least 100W, so in this inverter it would have an easy life.
The total lack of any filtering implies this inverter was not meant for any radio or audio equipment. What would an unfiltered 40W inverter be used for then? A 240V soldering iron comes to mind, but not much else. I don't think anyone would buy an expensive inverter just to run a fan, and the inductive nature of such doesn't suit this inverter anyway. It would tolerate a power factor corrected 20W or 40W fluorescent lamp. This seems to suggest it might be for something specific. The format of the model number also differs from the other two 100W units I've worked on.

For Fluorescent Lamps.
The answer to the mystery came from an advertisement in Radio, Television & Hobbies for April 1959.

As stated, the inverter can power a 40W tube or two 20W tubes. A double 20W fluorescent fitting is electrically equivalent to a single 40W type since the two tubes are connected in series, with the total voltage drop and running current being the same.
Of course, the inverter does not run the tubes directly since it has no control gear. Instead, a conventional 240V fitting is plugged into the inverter, since it contains the ballast, starter, and power factor correction condenser. On that note, it's essential the fitting used is of the high power factor type; that is it does have a power factor correction condenser. Some fittings don't, and will damage the vibrator since they appear as an inductive load.
The 50W rating of the transformer is also explained. A 40W fluorescent lamp actually requires about 50W due to ballast losses.

The extra slots in the chassis allow for larger transformers in other models.

Input fusing for the supply is taken care of by a 7.5A 3AG fuse. This is located in a bayonet in-line fuse holder in the active battery lead. Supply is switched with a bakelite tumbler switch and then fed to the reed of the vibrator. Each set of fixed vibrator contacts feeds the outside connections of the transformer primary, alternately switching the 12V between them. The transformer centre-tap is connected to the earthed side of the supply. This differs from the usual radio type of supply where the reed is earthed and the centre-tap is fed with the supply. It's exactly the same in terms of operation of course. The vibrator drive coil is switched by its own contact. Across this is a 0.1uF capacitor to prevent sparking caused by the high back EMF, produced when the coil's magnetic flux suddenly collapses.

There is some primary timing capacitance provided by the 2uF across the whole primary winding, and two 1uF's across the vibrator contacts. The latter appear to be connected more for the specific purpose of assisting with the prevention of contact sparking. Most of the timing capacitance is across the transformer secondary, in the form of a 0.56uF capacitor. Timing capacitance has been discussed at length elsewhere on the site, but briefly its purpose is to control the reversal of voltage in the transformer primary, when the vibrator contacts open. Correctly used, the voltage across the vibrator contacts right at the time when they open and close, will be at a minimum. This prevents contact sparking, as well as contact material transfer. Merely selecting timing capacitance to prevent visible sparking will not always result in the correct value, which also prevents contact material transfer. Thus, the waveform must be observed on a CRO to determine this. In the absence of a CRO, selecting the capacitance on the basis of lowest input current, with the supply operating into no load, is the next best method. However, it is impossible to accurately set for a 65% slope this way. All that can be done is to increase the capacitance slightly from the ideal value, but just how much is the unknown.

The 230V output from the transformer feeds a three pin socket. The earth pin is not connected to anything; not atypical of many inverters. In terms of safety it doesn't need to be, since neither side of the secondary is earthed. However, for inverters powering audio or radio equipment, it can result in a reduction of RFI, where the appliance is fitted with a three core mains lead.

Capacitor Replacement.
First thing was to replace the AEE Microcaps. These came out in the 1950's and are a paper type. The ones in the inverter were dated 1966. Although the inverter is only operating from 12V, the peak voltages are somewhat higher than this, and breakdown can still occur. Also was the fact that the moulded covering was cracking, and would allow moisture to enter was not a good sign. I reconnected the yellow Philips capacitor as I've never seen a failure of this type before, being a very reliable polyester type.
For the timing capacitor, I'd wait until checking the value was suitable before replacing it. Experience has shown that the choice of value used in some inverters has not always been ideal.
Next was to deal with the vibrator before powering up.

All capacitors replaced.

Screw had been removed from driver coil.

It was pretty obvious the vibrator had been got at. The driver coil had been removed at some point and put back incorrectly. The wire to the coil is meant to be routed out through the slot, not on the inside of the mechanism. Also, it was so far down that the reed was jammed up.

What was the point in pushing the coil so far down it hit against the reed?

One of the contacts was way out of adjustment.

The vibrator is a dual interrupter type, and it was obvious that one contact was not making when its mate on the same side already was. Another thing was the two vertical sections that support the driver coil and its contact were not parallel.

Note that the two sides are not parallel. The drive coil wire should emerge from the adjustment slot otherwise the reed hits against it.

As far as I could tell, the alignment of the two sides had come from the factory like that. The stack bolts were all very tight. This is no surprise as from other examples of these vibrators I've seen, the quality control was a bit lacking.
However, the driver coil positioning, and loose lock nuts for all the contact adjustments, is indicative of someone having a play.
I am never surprised what I see when people play around with vibrators, not really knowing what they're doing, and not using any proper adjustment procedure. The fact it was assumed the vibrator needed to be worked on in the first place is always suspicious.
These Van Ruyten vibrators are series (separate) drive, and despite the sometimes poor quality control during manufacture, the contacts are very durable and wear is not normally an issue. I guess it's the moving parts that always attracts attention!
Our would-be vibrator restorer had replaced the foam can insulation with paper. The original foam used with these vibrators has usually decomposed.

The paper was dated from at least 2002.

I used thin foam rubber to replace the can insulation. The paper would have been satisfactory from an electrical insulation point of view (the mechanism is live), but foam rubber is better for acoustic insulation. As per other Van Ruyten vibrators, I used foam rubber for replacement. At least one manufacturer once used corrugated cardboard which would be another option.
The paper found inside seemed to mention something about alternative energy, so it would appear whoever owned this inverter might have had an interest in such things and was going to use it with a 12V alternative energy system.

With the vibrator set up visually, it was time to test the inverter and check the adjustment. Alas, it didn't vibrate. The reed simply swung over and stayed there, even though the drive contact was now open. It was obvious why the Philips 0.1uF had been disconnected.
This capacitor had broken down, short circuiting the driver coil contact. A closer look revealed the voltage rating was inadequate to start with, at only 160V.

309V across a 160V capacitor. No wonder it broke down.

With a 400V capacitor to replace it, the vibrator began to work. Measuring the voltage across the capacitor showed a peak of 309V. Obviously, the designer thought that since the supply was only 12V, that the lowest voltage rating (160V) would suit. Not so. In fact, a neon bulb across the driver coil contacts lights up quite brightly, demonstrating just how much back EMF there is.
Having got that far, it was time to check the adjustment of the power contacts.

Vibrator in operation. No sparking evident. Drive coil wire was subsequently re-routed correctly.

Vibrator Adjustment.
For the Van Ruyten vibrators, adjustment is very easy since it's all done by set screws bearing against the fixed contacts. I took the easy way out and rather than get out my vibrator testing panel, I adjusted the vibrator in situ. Disconnecting the timing capacitor and loading the inverter with a 25W light bulb makes the switching time more obvious. It must be pointed out this method is satisfactory with this inverter - it may not be with others. The important point is, the loading prevents damaging voltages across the vibrator contacts, or across the transformer insulation. As it was, there was virtually no sparking at the contacts. The fact it works with no timing capacitor in this situation does not mean the inverter should normally be used this way. With no load, destructive voltages do appear. Furthermore, no timing capacitor will eventually result in material transfer across the contacts, pitting them and causing them to stick.

With no timing capacitor, the primary waveform with the inverter feeding a resistive load looks like this.

I set the duty cycle to 78%; i.e. 7.8ms 'on' time for each contact, and ensured the 'off' time was fairly equal. If the duty cycle is not even, there will be a DC component in the transformer which reduces efficiency. Given that this is a dual interrupter type with each set of contacts internally paralleled, how does one adjust each set so they close and open at the same time?
The way to do this is to isolate the other contact of each pair by having its gap wider than normal. Then the vibrator is adjusted as normal. Remove the vibrator from the test fixture, and now visually adjust the 'other' contacts so they appear to open and close with those previously adjusted. To get a finer adjustment, a current limited 30V 2A bench supply is connected from the reed to each contact pair. By moving the reed by hand, and causing the contacts to close, a spark will be visible. The idea is to get adjustment such that the 'other' contacts spark at the same time as the previously adjusted contact. It's then necessary to check adjustment with the CRO again and repeat the procedure if necessary.
In this particular inverter, the exact synchronism of the contacts is not as important as the 100W inverters, since even if the vibrator was only working on one set of contacts it would still be well within ratings.

Timing Capacitor.
Originally this was a paper Ducon 0.56uF 600V type. The unloaded waveform showed this value was not grossly excessive, or too little. Replacing this capacitor with a decade capacitance box, it was found the 'ideal' value was 0.13uF.

Ideal timing capacitance is 0.13uF for this vibrator and transformer.

The 'practical' timing capacitor was about 0.27uF. This is a slightly higher value than the ideal, and is what should be used in practice to allow for vibrator aging. It's determined so the slope is about 65%. If the transformer was feeding a rectifier to produce a DC output, the capacitor value would be left at that.
In the case of an AC inverter, as has been mentioned many times in the other articles on this site, the timing capacitor needs to be higher than the ideal or practical values, to accommodate for the inductive nature of some loads. If an inductive load is used, the timing capacitance is effectively reduced, which will be damaging for the transformer and vibrator. It is more of a compromise than a perfect solution, since with no load the capacitance is more than it should be. However, in practice, it works well enough without adversely affecting vibrator life. If the inverter is used with a very inductive load such as a fluorescent lamp, such as this one, it must be power factor corrected so it appears resistive. Vibrator inverters do not like inductive loads!

No load waveform with 0.56uF, and at right when loaded with a 40W light bulb. Note the ringing when loaded.

In this regard, the original value of 0.56uF was satisfactory, and I simply replaced it with a polyester type. There is some ringing in the waveform caused by transformer leakage reactance. This is not abnormal, and can be eliminated by including a damping resistor in series with the timing capacitor. Seeing as this inverter was not intended to provide a perfect waveform for delicate equipment, I did not persevere with this.

Inverter powering a 40W incandescent bulb.

Driving a Fluorescent Lamp.

\
20W fluorescent lamp powered off the Van Ruyten inverter. 10,000uF capacitor connected across the 12V input so as not to upset the regulated power supply being used to power it.

Naturally I had to test the inverter with a fluorescent lamp. I didn't have a 40W fitting, or high power factor double 20W fitting to hand, so to start with, I tried a conventional 20W fitting with a glow tube starter. Current draw initially was 6A; this being because of the power factor correction capacitor and nothing else. Once the tube started and was drawing current, the consumption fell to 4A. Another 20W fitting was tried; this one with a filament transformer (i.e. "rapid start"). There was some difficulty starting the lamp unless one side of the 230V output was earthed. This is not surprising, since rapid start fittings don't provide a high voltage kick to get the tube started. It must be pointed out that rapid start fittings are in the minority, and the inverter would have been designed for switch start types. Current consumption with the rapid start fitting was about 4A before starting, and stayed at this figure once started. The waveforms were not checked, but it would appear the filament transformer provided some cancellation of what would be excess capacitance until the tube started.
For further info on fluorescent lamp operation, see the 100W Van Ruyten inverter article.
The waveforms were not checked, so it is not known if the power factor correction capacitor was actually the optimum value in both instances. Efficiency is poor when operating one 20W tube, but would improve with two 20W tubes or a single 40W tube. From the small amount of information in the advertisement, it does seem to imply single 20W tube operation was not intended.
When used with a vibrator inverter, it is important that the lamp starts as quickly as possible, since the timing capacitance presented to the vibrator is otherwise excessive.

Also supplied with the inverter were these vibrators and other parts.

Considering four other vibrators were included, this certainly made the purchase good value for money. There was a shunt drive synchronous mechanism which I've never seen before, as was a non-synchronous mechanism which I vaguely recognise as a military type. It's quite small and suspended by what are really rubber bands. More familiar is the 12V Oak synchronous or dual-interrupter mechanism. The 'complete' vibrator is a 6V V5211 split reed synchronous. Unfortunately, its driving contact has broken off. One could configure it to operate as shunt drive if necessary. An aluminium clamp was provided which fits the can. Two mystery objects are a home made two pin socket, and two three legged support brackets. Lastly is an ancient surface mount UY-5 base valve socket - which fits the Van Ruyten vibrator. Presumably this was to connect up the VR vibrator for some kind of 'testing'.

It would appear that the owner, convinced the inverter needed a new vibrator, went in search of such named objects. Unfortunately, none of them are suitable for direct use in this inverter. For one thing, only one of them is 12V. Also, the output will be 100 to 115c/s instead of 50c/s, which means the timing capacitance will be much too high. Also, the current ratings of the contacts will be pushed to above their limits, with the possible exception of the unmarked 12V Oak mechanism - if it is dual interrupter and not just synchronous. I have not checked the contact timing to confirm this.
In summary, it just shows why the apparent simplicity of a vibrator power supply gets so many into strife. I doubt the original Van Ruyten vibrator was faulty at all, and it shouldn't have been tampered with. More likely, it was got at because it wasn't vibrating, and it took a while to realise it was the shorted 0.1uF instead. Too late now, because the vibrator had already been put well out of adjustment.