A non PWM <5% distortion sinewave inverter.

[Warpspeed]

Is your +/- 300 volts derived directly from batteries or solar panels, or is it regulated?

My system is still under construction, but the basic idea is to rectify the incoming 240v three phase supply with phase controlled SCRs to generate a rock solid +/- 300v dc supply bus, with plenty of holdup capacitance.

This inverter will sit across that dc bus to generate 240v nominal to run my house.

Now the unusual part of all this, is that I can then feed some dc power from batteries or solar into that +/- 300v dc bus, to power all my small constant loads.

When my dc/dc converter jacks up the +/-300v bus to perhaps +/-310 volts, voltage feedback in the SCR rectifier throttles down the SCRs totally, and all my power comes from solar/batteries.
So something like a few watts to a few hundred watts of 240v load will be then totally solar powered.

If someone turns on a huge sudden load, the solar system would be easily overpowered, and the dc bus voltage will fall from 310v to 300v, but not too quickly if there is enough holdup capacitance.
Voltage feedback in the SCR rectifier will swing into action within milliseconds,and take up the load.

Something like my refrigerator has a 140 watt running load, easily supplied by solar, but a 3.9Kw starting current. Its about 1KJ for one second inrush.

This way, I could run just the refrigerator off one solar panel, and use grid power just for sourcing the startup inrush.
I can get 3600 fridge starts from the grid for 1KWH cost, and otherwise run my refrigerator for free.

So far I have the 9KW three phase SCR rectifier module complete.
The 1.5KW prototype inverter is about 99% there, a much bigger one will come later I did make a second set of boards to build another.
Next comes a 1 to 1.5Kw dc/dc converter. 30V nominal in, +/- 310v out.
That is going to be be of a very simple and unusual topolgy that is going to surprise a few people.

Sunny Skyguy,
No snubbers needed. But it does need a lot of capacitance on the dc supply bus.
The reason being that when one of the inverters "subtracts" from the ac output voltage, power is recirculated back into the dc bus. The bus bounces around a fair bit unless it is well tied down with plenty of microfarads.

Alertlinks,
This type of inverter is quite happy with the type inrush loads caused by refrigerators and airconditioners. Transformers can handle huge short overloads quite well, and IGBT modules are available rated at hundreds of amps.
If you have the raw dc power to feed it from, the inverter itself is going to cope easily.

 

[D.A.(Tony)Stewart]

The cheapest capacitance with low ESR is a battery with a cost far less than $10/ Farad.

V*Ah *3600= Watt-seconds=Joules= 1/2C (Vi²-Vf²)
where Vi is the initial float voltage when fully charged and Vf is the final voltage after battery has discharged. So one can solve for C based on the Wh capacity of kg weight by rule of thumb.

ESR is rated from CCA for lead acid where 12.5V nom 100% charge drops to 7.5V at 0 'C or a 5V drop with CCA Amp rating. thus a 500A CCA rating has a minimum full charge ESR of a new battery at 5V/500A= 10 mΩ

 

[Warpspeed]

Re: A non PWM &lt;5% distortion sinewave inverter.

The cheapest capacitance with low ESR is a battery with a cost far less than $10/ Farad.

That is definitely true at low to mid voltage.
At high voltages, the very large number of battery cells required can become a bit of a maintenance problem.

- - - Updated - - -

Thanks for sharing, I'll surely try this.

Some further thoughts on hardware.

In my clock board schematic as drawn, the address counter is a ripple counter (not synchronous) and the EPROM is 120nS, the fastest one I could find in this now pretty old 2K x 8 memory technology.
Larger capacity EPROMS are now generally available much faster.

The output latch updates the data 300nS after the address counter is clocked. That is plenty of time for stable EPROM output data, but only if you use a fast HC4040 ripple counter and a fast EPROM.

Another thing to worry about is what happens as the +5v builds up and collapses, accidentally turning on two IGBTs together is a giant no no. So that is why there is that 4.2v under voltage detector chip.

My paranoia about simultaneously turning more than one IGBT on in the same inverter still will not go away.
Its a major source of unexpected random failure in inverter circuits.

The very best way would be to positively interlock the three opto isolators in a way completely independent of the +5v supply. Here is a suggestion of how that may be done :