What will happen if I inject rated output voltage into secondary of a single phase transformer?

I have a mains rated transformer that takes 220V and gives me 12V. Now what if I connect it to a H-bridge drawing power from a 12V car battery like this?

simulate this circuit – Schematic created using CircuitLab

The gates are connected to two IR2110 drivers which in turn is controlled by an ATmega328P emitting PWM signals for 50Hz sinusoidal wave. Will this fry my IRF540's, my IR2110 driver, the transformer, or would I be pushing out 220V AC?

• I've done exactly this, back-driving a mains-transformer as a cheap way to get an isolated 550V power supply (I actually needed +-275V, but that's irrelevant) with low effort. I used a L298 H-bridge, but it's basically the same thing. If you only need a few milliamperes, it works quite well. – Connor Wolf Jan 15 '15 at 9:50
• Make sure that C1 is very small as since the IRF 540 are switched hard on they become voltage sources and a very high of current will pass through the capacitor. Try not to resonate the LC circuit as you do not feed a parallel LC circuit from a voltage source. – user114989 Jun 25 '16 at 12:58

I'm assuming you mean this type of circuit to drive the mosfets: -

It is able to drive N ch devices because it uses a bootstrapping technique.

You won't get 220V AC but about 155VAC. This is because you need a 17V DC power supply to be able to reproduce a 12V RMS sinewave in a H bridge. (12V x sqrt(2) = 17V).

There will be other losses too but I don't expect things to melt. If you are happy that the reverse diodes in the IRF540s are capable of the peak current then this should also be OK. You'll need a fairly large reservoir capacitor across the supply feed from the battery because the kick-back from the transformer pushes energy back to the supply rails.

I'd advise you to try it on 12V DC and see what output level you get then move the supply to something more like 17V if you truly need 220V AC output. I'd also be wary about the 10nF capacitor across the transformer winding connected to the H bridge - try it without this component first.

• I do have a similar driver circuitry but I am planning on using 47uF ceramic MLCCs as C1, can I get away without C2? I used 1N5819's instead of '4007's, would that be a problem? And is it a good idea to use a boost converter to derive 17V from car batteries before sending it into the H bridge? – Maxthon Chan Jan 15 '15 at 13:24
• Also, about the reverse diodes, why would I need this? And would antiparallel 1N5819's good enough? – Maxthon Chan Jan 15 '15 at 13:29
• You should check the data sheet for the 1N5819 against the UF4007. Reverse diodes are needed when the H bridge disconnects from the load - this also occurs each time the H bridge switches due to anti-shoot-thru circuitry. – Andy aka Jan 15 '15 at 13:33
• What should I compare? Blocking voltage is not a concern here as I am not dealing anything near 20V. UF4007 have a reverse recovery time of 75ns and 1N5819 datasheets did not specify that. – Maxthon Chan Jan 15 '15 at 14:07
• If you do not have Trr then I wouldn't use that diode. Having said that, the device is a schottky diode so this won't be specified. Providing max voltages and currents are better then there should be no issue here. – Andy aka Jan 15 '15 at 14:43

This is close to the form of a basic DC-AC inverter circuit. However in this arrangement the upper MOSFETs should be P-Channel types, (note that you need to drive the Gate against a steady Source voltage). It can be done with all N-Channel parts but the circuit needs to be different, (see link)

Obviously if you turn on two of the MOSFETs from one side they will be fried. So even with a PWM output it is best to give a small delay before tuning on the alternate MOSFET pair. Some of the MCU parts have such a delay available when outputting a pair of PWM signals.

Here is a link to an example inverter circuit: http://www.bristolwatch.com/ele/arduino_power_inverter.htm - (Scroll down page)

• The OP is using a driver capable of driving 4 N ch mosfets - see my answer for clarification. – Andy aka Jan 15 '15 at 8:55
• The MOSFETs of the same side is prevented from be turned on simultaneously before the driver kicks in (I have several GAL chips programmed into dual H-bridge drive signal verifier to specifically address this issue) – Maxthon Chan Jan 15 '15 at 13:27
• OK, the IR2110 can drive two NMOS parts in this particular configuration. The extra Vs connection was not shown in the first example. – Nedd Jan 19 '15 at 20:40