For modern cheap 12V DC to 230V 50Hz AC inverters, it seems to be common practice to feed the 12V to a center tap on the primary side of the transformer and then use MOSFETS to alternately ground the two ends of the winding. Why is this the preferred topology?

To my thinking, given that the 400V output of the transformer is immediately rectified and pumped to a capacitor, which then feeds the high-voltage mosfets that produce the sine-wave (or modified-sine-wave) that drive the output, I don't understand the need to invert the 12V input like this. Why not have a single primary coil, a single mosfet, and just turn the mosfet on and off to produce a 0-12V square wave on the primary coil, then feed the resulting 0-400V output from the transformer's secondary into a capacitor?

I'm sure there's a reason almost every inverter does it this way, but I'm an ME not an EE so I lack the knowledge to understand it. In fact I'm not even sure I understand how a center-tapped primary even behaves. Does it affect the turns ratio? Do you get 12V induced on the unpowered primary winding?

 

Hello,

That is not true. There are many that use a Bridge and a single coil input transformer.
Have a look at the attached PDF.

Bertus

 

Hello,

That is not true. There are many that use a Bridge and a single coil input transformer.
Have a look at the attached PDF.

Bertus

Thanks for that. Very useful reference! I will read it carefully soon. From a quick skimming they use an H-bridge with four transistors to drive the primary bidirectionally. Why not use a single transistor to drive the primary with just one polarity?

 

Hello,

In a bridge configuration, you basicaly have double the voltage available.

Bertus

 

Also, you need to drive the magnetic field in opposite directions, or it will become magnetized and reduce its saturation current. A single sided driver like you propose would not reverse the field. Either of the other schemes does.

Bob

 

In a bridge configuration, you basicaly have double the voltage available.

Not 100% sure I understand that. I'm assuming the average current going to be I = P/V - so for a 500W 12V inverter 500/12 = 41.66A on average. So, can you clarify which components get half current by virtue of being driven forwards and backwards? Is it the transistors and the transformer primary?

 

Also, you need to drive the magnetic field in opposite directions, or it will become magnetized and reduce its saturation current. A single sided driver like you propose would not reverse the field. Either of the other schemes does.

Bob

Thanks, that really makes sense. So by driving bidirectionally you make better use of the transformer real-estate available basically?

 

Thanks, that really makes sense. So by driving bidirectionally you make better use of the transformer real-estate available basically?

Not related to that.
You need to drive it bidirectionally to avoid saturating the magnetic core.
Thus the average DC current into the primary must be near zero.
So it's not possible to efficiently drive the primary with only one transistor.

You could drive a single primary winding with 0-12V using a push-pull, two-transistor driver, but it would need to be coupled through a large capacitor to block the 6V average DC component.

One configuration that does only require one transistor is the flyback converter, but that is generally not as efficient.

 

Not related to that.
You need to drive it bidirectionally to avoid saturating the magnetic core.
Thus the average DC current into the primary must be near zero.
So it's not possible to efficiently drive the primary with only one transistor.

You could drive a single primary winding with 0-12V using a push-pull, two-transistor driver, but it would need to be coupled through a large capacitor to block the 6V average DC component.

One configuration that does only require one transistor is the flyback converter, but that is generally not as efficient.

Aha! The flyback converter was almost what I was envisaging (though I was missing the freewheeling diode in my imaginary circuit).

In your second paragraph, where would the capacitor be placed? In series between the transformer primary and the transistors?

Lastly, a pure-sine inverter I disassembled looks like it has 6 mosfets near the transformers on the input side. Any idea what kind of topology it uses? (pic attached) -

Would the two transformers be connected in parallel?

 

The flyback converter was almost what I was envisaging (though I was missing the freewheeling diode in my imaginary circuit).

A flyback converter does not use a freewheeling diode as that would clamp the output voltage to essentially zero.
It's the inductive kick that generates the output voltage.

where would the capacitor be placed? In series between the transformer primary and the transistors?

Yes.

Lastly, a pure-sine inverter I disassembled looks like it has 6 mosfets near the transformers on the input side. Any idea what kind of topology it uses?

Not without a schematic.

 

The centre tapped primary is more common on low supply voltages.
The argument goes as follows:
With a centre-tapped primary, at any instant in time, only half the primary is passing any current, and the other half isn't, but you only have two transistors. If it was all passing current, your wire could be twice as thick and half the resistance, so half the loss.
With a bridge, you have four transistors, but all the primary is in use. Two transistors are conducting at any instant in time, generating twice the heat loss as one transistor.
Given the available Rds(on) of transistors of various voltage ratings, two transistors and a split-primary works out more efficient at 12V and four transistors and a single winding works out more efficient at higher voltages.
You have also got to think about how to drive the two transistors on the positive supply (if there are four transistors), or how to suppress the voltage spikes when half the split-primary is switched off.

Lastly, a pure-sine inverter I disassembled looks like it has 6 mosfets near the transformers on the input side. Any idea what kind of topology it uses? (pic attached) -
Would the two transformers be connected in parallel?

They might well be in parallel, but the most common circuit has two DC-DC converters in parallel (so the outputs connect together after the rectifier diodes) then a PWM circuit to give the sinewave, filtered by the large toroidal choke in the top right corner.

 

They might well be in parallel, but the most common circuit has two DC-DC converters in parallel (so the outputs connect together after the rectifier diodes) then a PWM circuit to give the sinewave, filtered by the large toroidal choke in the top right corner.

Thanks, I think your reply was the final piece in the puzzle for understanding topology choice (though I didn't understand why you could make the wire twice as thick if the primary is all in use - is it because you need twice the primary turns to do the same job with a center-tapped transformer?).

So, given mine is a 12V inverter, a center-tapped transformer is more likely. Doubling the circuit for double the power capacity would explain the two transformers and four of the bottom left mosfets. I can see 8 diodes above the transformers, so probably two discrete rectifier bridges. Then there's two large and two slightly smaller capacitors above those. No idea why they doubled up on what seems to be identical capacitors - if outputs are connected together why not just have one huge 400V capacitor?

Any ideas what the bottom left two mosfet-looking things bolted to the case could be? Can't read part numbers because they're covered in rubber.

 

The first circuit is a very simple and cheap inverter with an unregulated squarewave output voltage.
An inverter circuit producing a pure sinewave regulated voltage is much more complicated and expensive.

 

(though I didn't understand why you could make the wire twice as thick if the primary is all in use - is it because you need twice the primary turns to do the same job with a center-tapped transformer?).

One set of turns for each MOSFET.
Only one MOSFET is on at any given time, so the other part of the winding isn't conducting any current. It's just occupying space on the bobbin.
If it only needed one winding, then it could be thicker, half the resistance, half the loss.

Could the two mystery devices be diodes in reverse across the supply so that it blows the fuse if connected up back-to-front?
Or 5V regulators for the logic?

 

One set of turns for each MOSFET.
Only one MOSFET is on at any given time, so the other part of the winding isn't conducting any current. It's just occupying space on the bobbin.
If it only needed one winding, then it could be thicker, half the resistance, half the loss.

Could the two mystery devices be diodes in reverse across the supply so that it blows the fuse if connected up back-to-front?
Or 5V regulators for the logic?

A user on another forum suspects it uses the flyback topology, and each set of three FETs are in parallel driving one transformer. The large copper traces seem to support this theory since they appear to connect to all three FETs?

 

If it is an isolated flyback transformer, then where is the LCR circuit on the MOSFET drain?
[As output neutral must be connected to output earth it should be isolated.]
As the input current to a flyback peaks at >4x average, then I doubt it would be a flyback. Average current would be output power divided by 12V, and peak current >4 times that. That's a lot of current.

Given the latest information, I'd go for two transformer with split primaries, in parallel, each leg being driven by 3 MOSFETs in parallel.
The transformers might have separate rectifiers.

 

If it is an isolated flyback transformer, then where is the LCR circuit on the MOSFET drain?
[As output neutral must be connected to output earth it should be isolated.]
As the input current to a flyback peaks at >4x average, then I doubt it would be a flyback. Average current would be output power divided by 12V, and peak current >4 times that. That's a lot of current.

Given the latest information, I'd go for two transformer with split primaries, in parallel, each leg being driven by 3 MOSFETs in parallel.
The transformers might have separate rectifiers.

Hmm, that's a good point. This is a 500W inverter, so 41.6A @ 12V input average current at rated load. I've taken several images from different angles if it helps, they're in this album here (pics 14, 21 and 25 show the transformer traces a little better) -

http://imgur.com/a/p9ecHGy

You can actually see 8 discrete diodes just near the two transformers, between them and the big capacitors.

So you think it's two transformers electrically paralleled, with +12 going to the center tap of the primaries, then 3 mosfets grounding one end of the paralleled primaries and the other 3 mosfets grounding the other end of the paralleled primaries?

 

If it is an isolated flyback transformer, then where is the LCR circuit on the MOSFET drain?
[As output neutral must be connected to output earth it should be isolated.]
As the input current to a flyback peaks at >4x average, then I doubt it would be a flyback. Average current would be output power divided by 12V, and peak current >4 times that. That's a lot of current.

Given the latest information, I'd go for two transformer with split primaries, in parallel, each leg being driven by 3 MOSFETs in parallel.
The transformers might have separate rectifiers.

Well, we were almost right.

I've given in and disassembled the inverter. Turns out it's a push-pull design. The two center MOSFETS don't appear to do any switching because they don't connect to the transformers. They just connect between the negative battery wire and the negative ground planes visible on the top. I'm assuming that means they are just there for reverse-polarity protection.

Photo:

http://imgur.com/6GR7uDC