Disadvantages of Using half-wave instead of full-wave in Pure Sinewave Inverter?

[The Electrician]

However, could you highlight where on those two pages the manufacturer states that a laminated iron cored transformer is used for the output device, and that it is fed from an HF PWM source to emulate a sinewave?

Regards,
Xena.

The manufacturer doesn't discuss inner workings on those two pages or anywhere else, but you can get an idea of what's inside from images in the service manual: https://outbackpower.com/downloads/documents/inverter_chargers/fx_vfx_series/fx_fxr_service_manual.pdf

[T3sl4co1l]

Curious minds want to know how it is possible to drive a iron cored line frequency transformer with a PWM 15,625Hz square wave directly into the windings, and it does it with high efficiency and producing an averaged pure sine wave at nominal line frequency over its full range of loads, as an output.

Perhaps curious minds can set up an experiment and demonstrate it themselves.

Reporting back with real calculations and waveforms would provide valuable grounding to the discussion in this thread.

Tim

[Kim Christensen]

The manufacturer doesn't discuss inner workings on those two pages or anywhere else, but you can get an idea of what's inside from images in the service manual:

See attached and ignore spelling:

[mag_therm]

[
Back in the 1980s, push-pull driven steel laminated inverter transformers were actually common. But technology has superseded them. However, the TS lives in a country where this technology is not available. Or prohibitively expensive.

Curious minds want to know how it is possible to drive a iron cored line frequency transformer with a PWM 15,625Hz square wave directly into the windings, and it does it with high efficiency and producing an averaged pure sine wave at nominal line frequency over its full range of loads, as an output.

Perhaps not having read the original assertions, all those who say this is a thing have given caveats on the original.

We've not yet seen the honorable members ingenious circuits schematic yet, or detailed specifications of the transformer.

Regards.

To the contrary, I have confidence that the pwm into conventional iron cored 50 Hz transformer is working well enough.
I thought back today to what I was doing 50+years ago as junior engineer developing transistor inverters.
In those days we could only do base frequency (2 pulses per cycle) as opposed to 320 pulses in present discussion.
The Ott filter was costly, moreover while it was good for Scr inverters capacitive input Z was no good for BJT.

The solution I worked on was to connect a LC parallel tank directly to secondary of the push pull transformer.
The tank had a 400V AC oil filled capacitor. Tank Q was about 5 for fuse clearing.
The switching edges on the secondary were directly applied to the capacitor.
I recall the secret sauce was to wind the inverter transformer (EI) with sufficient extra leakage L. to eliminate the AC filter input choke.
~ 4 dog bones provided the extra leakage from standard ~ 0.05pu to I suppose 0.12 pu.

I remember the collector voltage waveform, which was a replica of the tank voltage, a pure sine.
The transistor commutation rectangular pulses was superimposed. and they would get fatter and thinner doing volt regulation,
and inherently move back and forward relative to the tank voltage  with load power factor changes.
However the secondary voltage was almost pure sine ( We used old HP distortion analysers)

Given that project, and the diminishing 16kHz flux swing that I mentioned in #43,
I am quite confident that an ordinary 50 Hz transformer will work with 16 kHz with today's Mosfets and their ap diodes.

[KerimF]

I built what it was supposed to be a 500W pure sinewave inverter. It seemed to work fine (as it should be) for no load and light load.
But by increasing the load gradually (up to 200W), the MOSFETs (2 x 4 STP75NF75) started to be really hot.

This didn't surprise me because I used to miss always one or more points in every new design

As you know, the main dissipation of a switched MOSFET occurs at the rising and falling edges.
In conventional square wave 50Hz (I use producing), even driving 4 MOSFET gates with 10 mA only used to work!
So, I had the impression (which I know now it is wrong) that 300mA will also do the job (in case of 16KHz). For instance, I was able to build the +/- 300mA driver by using 3 BJTs (2*BC337 + 1*BC327). I did it this way to minimize the cost.

So, I think now that I have to find an available 'low side' driver IC instead (rated 3A output, in the least).
I will likely find costly driver ICs for H-bridge instead (but it is better than doing nothing).

Meanwhile, I will try tomorrow to drive 1 MOSFET only (instead of 4) with 300mA. I expect the dissipation will be lowered (also the output attenuation). After all, switching a MOSFET at a rate of 16KHz with +/-75mA only, the rise and fall times will be relatively very long, and there was 4 MOSFETs. And the situation becomes even worse when they start to be hot because this increases their Rds. And this, in turn, increases their temperature (positive feedback?)

[johansen]

So, I had the impression (which I know now it is wrong) that 300mA will also do the job (in case of 16KHz). For instance, I was able to build the +/- 300mA driver by using 3 BJTs (2*BC337 + 1*BC327). I did it this way to minimize the cost.

So, I think now that I have to find an available 'low side' driver IC instead (rated 3A output, in the least).

you should be able to get a lot more than 300ma from those transistors, but if not, find something better. no need to spend money on a 3 amp capable gate driver, which is almost certainly going to be surface mount...

[Xena E]

Curious minds want to know how it is possible to drive a iron cored line frequency transformer with a PWM 15,625Hz square wave directly into the windings, and it does it with high efficiency and producing an averaged pure sine wave at nominal line frequency over its full range of loads, as an output.

Perhaps curious minds can set up an experiment and demonstrate it themselves.

Reporting back with real calculations and waveforms would provide valuable grounding to the discussion in this thread.

Tim

If by that Tim, you mean the original poster providing circuit diagrams and details of his build that he was asking for 'help' with, which I have only a very vague impression of, and that impression of what was built would not work to the claimed efficiency then yes, he should.

No one can form any opinion about this "project" without the details of what was done.

Extraordinary claims require extraordinary proof, there's been no proof and the story keeps changing.

I'll wait for Kerim F to post those details, and prepare to have my socks blown off. In the mean time stick to working with the idea that DC link bridge output inverters, are as good as things can get.

Regards.
X

[Doctorandus_P]

For commercial products, using two primary coils in the transformer is a big issue. Copper is more expensive and bulkier then FET's, so it both increases size, weight and cost.

For a one-off or small series, this is not so much of an issue, while using simpler electronics and parts availability are more important. Especially when you live in a part of the world where electronic parts are hard to come by.

[KerimF]

For commercial products, using two primary coils in the transformer is a big issue. Copper is more expensive and bulkier then FET's, so it both increases size, weight and cost.

Very true.

For a one-off or small series, this is not so much of an issue, while using simpler electronics and parts availability are more important. Especially when you live in a part of the world where electronic parts are hard to come by.

It happened that this is the situation on my side in these days. (Before March 2011, it was somehow easy to order anything from China, like power ferrite cores for efficient pure sinewave inverters, for example. And during those past golden years, I used to send just the schematics of my various designed controllers, drawn on Kicad for example. Then, the Chinese engineers drew their appropriate PCBs, by using their advanced tools, assembled the components on them and tested the finished boards before exporting them to me).

[SeanB]

Easiest way to lower cost is to use CCA wire, which should be very easy to get, as it is a lot cheaper than copper, though working with it needs crimped connections all through, or some rather aggressive flux to clean it for soldering. Yes not copper, but if all Chinese products and motors now are made with it.......