Dc-Ac pure sive inverter design

[quasi]

Building this type of inverter ain't easy. Especially when you can buy them off the shelf for less that you can make it for, unless you can get the parts for nothing.

And any mistakes will be costly usually in FETS.

However this is the design principle I suggest.

Convert first the 12v to +/- 350 volts rails using a 3525 with a centre tapped winding on a ferrite core and push-pull mosfet switching. These FETS (or FET groups) will need a current rating of around 200 amps (say 4 x 60volt 50amp FETS per side). This will need feedback to maintain the voltage under load.

Then build what is basically a class D amplifier switching a PWM sinewave with FETS on your +/- 350v rails. This will give you around 230 volts RMS (losses accounted for) into your load. These FETS need to be say 500volt 20amps to allow for surges and stuff and you will only need about one per rail.

Sounds easy....but the quality of components especially the price of good high value and voltage capacitors and the FETS will really be about twice the price of a commercial unit.

If you want to tackle this then I suggest a module at a time. I,e build a simple SMPS using a 3525 first and understand how the whole thing works. Then build a high power version to generate the rails and capacity to feed the class D amp.

There are circuits around for class D amps but you will need to upspec them to handle the power.

Oh ...you will not need to worry about distortion figures.

Good Luck.

[johnson591116]

..................................Good lock

[soundbag]

................


Thank you.

soundbag

[jackinnj]

there was a good inverter article in EW a few months back -- and there is a reason for "approximating" a sine-wave pointed out by the author -- the remote equipment test he had been making had a real hard time digesting a square wave

take a look at Linear Technology's LT3439 switching chip -- you might be able to get some samples sent to Nigeria -- even if you don't use the chip the PDF on this and their other quiet gate drivers should show you how previous designers have approached the problem.

[soundbag]

Thank you Jack,

My problem is not dc-dc, but dc-ac.

Is there no other way of inverting dc-ac, than dsp, pic, etc.
For an inverter of say 500watts?

Pls, Jack let me know, the dsp thing is very high for 500 watt
inverters.

soundbag

[gmarsh]

My suggestion - buy an off-the-shelf 12V inverter that plugs into a car cigarette lighter, and pull it apart. Steal their design, add some improvements for a higher power output, and call it your own.

As much as you don't want to use PICs, they're probably one of the easiest ways of generating the proper gate drive signals you'll need for the output MOSFETs/IGBTs. The alternative is to create two analog circuits - one producing a 60hz sine wave and one producing a triangle wave at the switching frequency - and compare them.

This isn't that hard to do, anyone who has built a HIP4080 class-D amp is familiar with the method. But going with a microcontroller will be a simpler solution IMO.

All you need to do is select a PIC/AVR/H8/8051/etc which has a couple onboard timers capable of PWM. Store a sine wave cycle (or 1/4 of one) in memory and make the chip fire a new duty cycle value at the timer unit every time a PWM cycle completes. You'll likely need a second timer of some sort in order to provide dead time control, or you can do that with external logic.

The microcontroller outputs (and maybe a bit of external dead-time logic) then drive a pair of gate driver IC's, which in turn drive a pair of gate-drive transformers for your output. Actually, a HIP4080 or HIP4081 could do a wonderful job - it provides both gate drive and dead time control in the same chip.

You could even get creative and use the same microcontroller to generate the gate drive for your 12VDC -> HV DC converter.

[jackinnj]

Originally posted by soundbag
Thank you Jack,

My problem is not dc-dc, but dc-ac.

Is there no other way of inverting dc-ac, than dsp, pic, etc.
For an inverter of say 500watts?

Pls, Jack let me know, the dsp thing is very high for 500 watt
inverters.

soundbag

The inverters I've seen just use a multivibrator to bang a square wave -- so on the transition you have a lot of noise -- the Linear Tech chip I cited drives the gate in a manner which softens the transitions (for want of a better term) -- the Linear Tech application note is useful for its discussion of "prior art"

gmarsh has a good idea in that an off-the-shelf inverter will give you a nice transformer and case and plug-in to the cig lighter -- everything else you throw out!

you don't need a full sine wave, in this application however -- because a full sine wave is going to provide less "on-time" and lower efficiency. you might want to peruse the apnotes on microchip's website.

the Linear Tech chips are included in the parts library of their Switcher CAD software --

[soundbag]

Thank you,

I will work on these suggestions for now.

If you know about any good book on "modern inverter design"
I will appreciate it pls.

I will get back to you soon.

soundbag