Constant voltage transformers with battery-to-mains inverters: any pitfalls?

[tggzzz]

I recently had about 3 minutes to research an item in a local auction, so I placed the minimum bid. Fortunately nobody else wanted it, so I got it at a very acceptable price. The only thing I missed was its weight: 20kg - but that's not important. It is a 630VA 240V isolating constant voltage transformer (a.k.a. ferroresonant transformer), which will be useful in one way (as a simple isolating transformer) and possibly in another - the subject of my question.

If you are in the middle of a field and want to use some mains powered equipment, then the obvious way is to use a 12V car battery and a DC->mains inverter. But that raises the problem that the mains equipment is probably incompatible with a square-wave inverter, possibly with a modified-sine inverter, and might only be compatible with an expensive sinewave inverter.

Now constant-voltage transformers operate at resonance, which implies they filter out other frequencies - so they might be able to turn an inverter's output into a decent sine wave. And indeed:

• eyeballing the output shows a far more sine-like output than the input; the peaks aren't flattened edited for clarity to...
• with the mains input, eyeballing the output shows a far more sine-like output than the input; unlike the input, the output's peaks aren't flattened
• low-power low-voltage tests indicate that harmonics of 50Hz are suppressed by >25dB, which is more than I was expecting

So, does anybody have experience of using DC-mains inverters plus CV transformers? Are there any pitfalls to avoid?

[MagicSmoker]

I don't have specific experience supply a ferroresonant transformer with a "modified square wave" inverter, but aside from higher circulating current in the tank capacitor it shouldn't be an issue. In fact, it should be a halfway decent way to get something closer to a sine wave out of the inverter.

[tggzzz]

I don't have specific experience supply a ferroresonant transformer with a "modified square wave" inverter, but aside from higher circulating current in the tank capacitor it shouldn't be an issue. In fact, it should be a halfway decent way to get something closer to a sine wave out of the inverter.

I can imagine some mechanisms, but I don't know whether they are a problem in reality. The principle issue it that there is energy in the harmonics, so where does it go and where is it dissipated.

[MagicSmoker]

I can imagine some mechanisms, but I don't know whether they are a problem in reality. The principle issue it that there is energy in the harmonics, so where does it go and where is it dissipated.

This will depend somewhat on the exact nature of the inverter, but most of the harmonic energy will be dissipated in the ferroresonant supply (ie - the core, which already operates with high losses because it swings between + and - saturation, and the parasitic resistances of the resonating winding and capacitor). Some energy will simply circulate between the ferroresonant transformer primary and the reservoir capacitance that feeds the H-bridge output stage in the inverter (assuming that is the circuit employed), much like normal primary magnetizing current.

[PChi]

I had some experience about 30 years ago. They worked OK and produced a reasonable sine wave being driven by a half bridge square wave (bipolar transistors). I had a go at trying to analyse what was going on but just resorted to guess work. There were some winding taps which could be used to adjust the output voltage along with the tank capacitors which were connected to the secondary.
It didn't work when connected to a colour CRT monitor because the CV transformer limited the output power and the monitor degauss circuit took a high start up current.

It was old technology then but I do remember that one customer bought 24 V 1000 W units. They had tried a couple of PWM types previously but the EMC emissions of them were too high.

I think that PWM technology has moved on a fair way and has obsoleted the CVT approach.

[tggzzz]

Thanks for both the responses.

I'm sure that inverter technology has moved on, but there's always the bottom end of the market! Sine wave invverters are availble, but only at a price.

The other question is that there are SMPSs/inverters in test equipment that tacitly assume the input is a sinewave - because that's all that was common when the equipment was designed. I can imagine such equipment "behaving poorly" with a square or modified-sine input.

Using the resonance behaviour of the CV/ferroresonant transformer will largely restore the sine wave by filtering out the harmonics. The constant-voltage behaviour is not important in that application.

[SeanB]

Remember the ferroresonant unit will attempt to draw a massive turn on surge, and when running it will draw the full load current on the primary side, though the power factor will be somewhat horrid with a very high harmonic content on the input current. Most inverters will struggle with this load, and you are better off using it as an isolation transformer in the lab, not out in a field.

Out in the field you probably are using something with a SMPS power supply, so just connect it to the inverter direct and there is a good chance it will work fine. Only things that do not like the square wave is things with an auto tap changer and those with a voltage selector, as the voltage falls in the middle of the switching range for the power supply, so on 110V it is going to be too high after the doubler and on 230V too low. Devices with a universal input ( 100-255VAC) will likely work fine, and most active PFC will work with this.

I remember those old ferroresonant UPS devices, just took one apart recently, and there was 30kg of copper in the 2 transformers and the filter chokes. This one used a semi resonant approach, using a 50uF capacitor on the output and a series inductor to make a LC tank to do harmonic filtering. Input was a simple SCR controller with a 0.7mH 80A inductor to provide a battery charge control, using wire that was more bus bar than enamelled copper wire. Got a lifetime supply of 2N3773 transistors as well, on the 4 massive heatsink assemblies, though the diodes they used there across the transistors were definitely not man enough, as they had cremated themselves and cooked the PCB tracks off the board.

[oldway]

In my point of view, we can not simply supply a CVT by the square wave output of an inverter because the input impedance of the CVT is capacitive at high frequencies. (square wave has a high harmonics content)

It is as feeding a partially capacitive load with your square wave inverter.

The high-frequency harmonics will therefore virtually been shorted, which could damage your inverter.

In old CVT inverters with fast scr's , we had an inductance in the dc bus blocking these harmonics.

On the other hand, the cvt transformer has a low efficiency and high no load losses. This is definitively not what we need for a battery powered inverter.

[MK]

They produce mains with significant 3rd harmonic, which reduces the load on any bridge rectifiers that follow, they have a lazy s shaped VI curve which may upset your inverter, just a guess, as I have only over used them off the mains. Worked on many from 30-750 VA before switchers became popular. They are inherantly noisy too, just dont use one for your hifi, unless you are listening very loud...

[tggzzz]

Remember the ferroresonant unit will attempt to draw a massive turn on surge,

When using it on the regular mains with no secondary load, I didn't notice the lights dim More seriously, that sounds like a good use case for a high power series resistor, a differential probe, and a storage scope.

and when running it will draw the full load current on the primary side, though the power factor will be somewhat horrid with a very high harmonic content on the input current.

My limited experiments have made me wonder about that, but I've done no concrete tests.

Most inverters will struggle with this load, and you are better off using it as an isolation transformer in the lab, not out in a field.

That's the kind of suspicion that made me ask the question in the first place.

Out in the field you probably are using something with a SMPS power supply, so just connect it to the inverter direct and there is a good chance it will work fine. Only things that do not like the square wave is things with an auto tap changer and those with a voltage selector, as the voltage falls in the middle of the switching range for the power supply, so on 110V it is going to be too high after the doubler and on 230V too low. Devices with a universal input ( 100-255VAC) will likely work fine, and most active PFC will work with this.

Much of my relevant kit is old, not a universal input, and does have a switch for 110V and 240V.

Thanks for your comments.

[SeanB]

650VA is not that big a surge, but the one I have is a 2kVA unit, it does have a pretty big THRUmmmmmmm on turn on. The other one had an ammeter on the input, and it would draw around 5A from the mains just running there doing nothing, so I added a 120mm fan to move air across the case to keep the room heater cool, though it was big enough that convection cooling worked well on it.

[MustardMan]

I note this is an old thread (October), but just in case somebody reads threads this old...

I worked for many years with "cheap" inverters, particular back when 'modified sine wave' - just say square wave and drop the BS -  inverters were just about it, and they were *expensive* (like $1k for 1kVA). Pure sine wave inverters of this power output and higher were so expensive your hair would turn grey (and now mine has!).

I sometimes used CVTs to fix the square wave output. Exceedingly inefficient. However, the output would drive things like audio power amps and stuff with virtually no audible noise. Feeding the square wave in resulted in a terrible buzz (not hum, no ground loops were involved). They were also the only solution when dealing with some mains operated equipment that uses a capacitive voltage dropper - a square wave would kill these in minutes, usually seconds, and sometimes cycles!

I played with CVTs from 100W to 2500W. The 2500 came from an inverter and was built as a UPS. The inverter was square wave, and the CVT gave it an almost perfect sine wave output. Losses were horrible, at no load the thing still drew significant current. At high (max) load it was in the order of 80% to 90% efficient. I did graphs showing how the output voltage coped with different loads, including overloads... the voltage collapses.

This was before SMPS were readily available, and especially for old colour TV sets, the CVT was virtually a necessity!

In the present, pure sine wave inverters are cheap ($200 for 1kVA), and nearly everything uses an internal SMPS design which can cope with square anyway (and many can cope with just straight DC - at 90+ volts).

Good CVT: Very clean output. Can 'replace' almost a full missing mains cycle. Almost indestructible.

Bad CVT: Extremely inefficient, and get very hot. Can not work with an overload (but will not be damaged).

Unless you are really on a budget, use a modern pure sine wave inverter (but beware, the 'cheap' ones do NOT live up to their surge ratings).

Cheers,
MM.

[tggzzz]

Thanks for your comments. Solid practical experience is valuable.