switch-mode sine wave inverter design

[helenfish]

hello,everyone

i need some help on design of a dc to ac inverter,which is used to convert 12v dc to 230v pure sine wave. could anyone give some ideas to me? may i use SMPS to build this inverter? i need some ideas on how to convert dc to pure sine wave also. thanks ya

[jackinnj]

Originally posted by helenfish
hello,everyone

i need some help on design of a dc to ac inverter,which is used to convert 12v dc to 230v pure sine wave. could anyone give some ideas to me? may i use SMPS to build this inverter? i need some ideas on how to convert dc to pure sine wave also. thanks ya

you can use a watch crystal (32768 Hz) to derive a 60Hz square wave by dividing down with a CMOS counter, then use a 4th order low pass filter to remove the harmonics -- amplify this signal with a class-D amplifier to drive a transformer -- what could be simpler?

[pinkmouse]

Just buy one- unless you need to learn about inverters or you need serious, (2KW+) power, (which is then not a beginners project), then you can buy one way cheaper than you can build.

[N-Channel]

Helen,

The first part, the DC-DC Converter should be pretty straightforward- N-Channel MOSFTEs with very low Rds(on) ratings, and high voltage MUR-series rectifiers at the outputs. Most likely, you will do a center-tap pushpull on the primary side of the boost transformer, but your output will be a conventional bridge rectification scheme, using the with MURs. Good candidates for the MOSFETs and output rectifiers would be would be the IRFB4215 (60Vdss and Rds(on of 9mW , and the MUR1560 (15A, 600V), respectively.

If you're doing MOSFETs for the H-Bridge, a good one here would be the IRF32N50K (500Vdss, 32A Id). I have not worked with IGBTs so I could not recommend a good one here, but I'm sure Eva or one of the other Switchmode Gurus might be able to help here.

The real trick here is the 50Hz Sinewave generator. You could just do a sinewave oscillator set for 50Hz, then adapt the drive signals for N-Channel H-Bridge, but with the MOSFETs (or IGBTs) running in their linear regions, you will be dissipating alot of heat as your power levels go up.

Another approach here would be to use a PWM signal at, say, 30kHz, modulated at 50Hz to drive the H-Bridge, and then form a low-pass filter designed to recover the 100Hz from the modulated output. This will drive up the cost of the curciut, though, as inductors and capacitors capable of these voltages (~330Vdc) and passing heavy currents will neither be small nor cheap.

BTW, and perhaps I should have asked this first, but, what power levels are you talking about?

Hope this helps,

Steve

[Eva]

Such a project is similar or even harder than developing a 4KW class-D subwoofer amplifier for car use.

The main problem are the switching and conduction losses associated with the *high*voltage* class-D output amplifier, where each output transistor has to hard-switch 350V or more over a conducting diode in each clock cycle, thus generatng heat and EMI. Another problem is the size of the output filter, because it's a big compromise between voltage ripple, current ripple and size. Efficiency won't be much above 80% for such a design (barely better than the theoretical efficiency of a linear class-B amplifier playing a sine wave at full output swing ).

Consider a quasi-sine converter (square wave with controlled duty cycle) if your application can handle it.

[jackinnj]

EVA -- yes you quickly burn watts --

here's an idea for slew control using a sine wave and some clamp diodes -- the opamp must be rather capable to do this and you burn energy in the linear region for the switching devices -- but dv/dt is much lower -- it is akin to what happens in some of the Linear TEch controllers:

[mzzj]

Originally posted by Eva
Efficiency won't be much above 80% for such a design (barely better than the theoretical efficiency of a linear class-B amplifier playing a sine wave at full output swing ).

I have got around 93-94% effiency at 2.2kW out from my class-D BCA full-bridge running from 300v rail. And this was ~100vrms to 4.5ohms load.

I would expect better effiency with around 20ohms load as it would allow near maximum pulse width. 100vrms from 300v fed full-bridge is not the most efficient way to do things.
+Use of lower switching frequency wouldnt hurt either.

[Eva]

What switching devices were you employing? How much?

If they were MOSFETs, were you switching directly across its (conductiong) body diodes?.

Anyway, 93% efficiency at 2.2KW means 165W of heat that you have to get rid of.

[mzzj]

Originally posted by Eva
What switching devices were you employing? How much?

If they were MOSFETs, were you switching directly across its (conductiong) body diodes?.

Anyway, 93% efficiency at 2.2KW means 165W of heat that you have to get rid of.

Some oldish STW20n50 mosfets. 2 parallei. Some more up-to-date mosfets(like coolmos series) would be sufficient without parallei devices.

body diode is not a problem in BCA-topology as its never conducting, and i was using zero-recovery sic-diodes for freewheeling diodes. Some ultrafast-recovery ordinary Si-diodes would be more cost efficient and they are still at least decade better than intrisic boby diode in mosfets.

Yeah, 165w is a lot of heat without fans. Luckily one(or two) CPU-cooler can handle that amount easily.

Just bought 10pcs A64 orginal boxed coolers with 2 euros a piece, these should take 100w without too much sweat.

[mzzj]

Originally posted by N-Channel
I have not worked with IGBTs so I could not recommend a good one here [/B]

IRF´s Warpspeed and Warpspeed2 or Fairchilds SMPS and SMPS2 series IGBT´s would be most obvious candidates.

For example warp-series IRG4PC50W should do 1-2kW full-bridge pretty easily.