New and adequate topology for Offgrid Battery Inverters

[joao.caracas]

Offgrid inverters (convert 12,24 or 48V DC to 220V AC) are definitely not new stuff. Thousands of cheap (crappy) chinese models are available and also some good ones. Our company is starting a project to develop a offgrid inverter 24Vdc/220Vac 2kW, and i don't think reinventing the wheel is a good thing, but I would like to make some improvements on this inverter.

Almost all crappy inverters are based on one or several paralleled push-pull converter coupled with a full bridge inverter, and than my question, with today technology which would be the "best" topology for the DC/DC converter?

It needs to be isolated, high efficiency (as always...), Input voltage 20 to 28V, DC bus 370V, Ac voltage 220V 60Hz...

Do you think a single LLC DC/DC converter would be a good choice for this power?
I have the possibility of making a bidirectional inverter, would CLLC be a good choice of bidirectional dc/dc converter?

Why dont make a single current fed push-pull converter?

 

[Warpspeed]

2Kw at 20 volts needs 100+ amps AVERAGE.
Only reasonable approach to that would be a forward converter, either full bridge or push pull.
Today's technology has not changed the fundamental problem.
For minimal conduction losses you absolutely must use something that draws a fairly constant current from the source, without intermittent high crest factor current peaks.

 

[smijesh]

Primary current for your Invert er will be 183A if you operate in 12VDC input(Efficiency 90%). It will be difficult to switch high current with high efficiency. So try with ISI topology (interleaved source invert er).

 

[joao.caracas]

smijesh, I did not know this topology, I will analyze. The keys are operated at 120 degrees, such as an inverter?

Warpspeed imagine que I use a full-bridge topology based, so there are 2 MOSFET in series with this current, If i use 3 mosfets (IPB027N10N3 2.7k Ohm) in parallel for each switch, the conduction losses are: 2x (2.7m / 3) x 100² = 18W. Is this correct?


I was thinking on this CLLC topology:



I do not need it to be bidirectional right now, but I intend to try this for future purposes. After doing a "basic" offgrid I will need to reverse the Hybrid grid tie.
This converter has the advantages of having a ZVS (or ZCS) for the mosfet and ZCS for the rectifying diodes. What do you think of using this? Another possibility is use a half-bridge with a voltage doubler on the secondary side to reduce conduction losses.

 

[Warpspeed]

Warpspeed imagine que I use a full-bridge topology based, so there are 2 MOSFET in series with this current, If i use 3 mosfets (IPB027N10N3 2.7k Ohm) in parallel for each switch, the conduction losses are: 2x (2.7m / 3) x 100² = 18W. Is this correct?

Yes that is pretty much it. Just Ohms law with regard to Rds on, assuming full maximum duty cycle.

Push pull requires higher voltage rated mosfets, which have a higher on resistance, but there will only be one in each conduction path. This is usually a pretty good tradeoff at twelve volts.
At twenty volts a full bridge is often better because lower voltage rated mosfets can safely be used, and even though there will now be two in each conduction path, the conduction losses can still be favourable.

Above twenty volts, a full bridge is definitely the way to go.

Our biggest hurdle by far will be conduction loss. Switching loss will be negligible in comparison.
All these fancy resonant mode circuits are brilliant at reducing switching loss at very high switching frequency, but its not really a solution relevant for our particular rather mundane low voltage high current switching problem here.

The coupling capacitors in your circuit are a simple solution to several problems, and are more applicable at higher operating voltages.
Finding suitable capacitors of large enough capacitance and high enough ripple current rating, pretty much rules them out for this application.

A much more appropriate solution would be monitoring the transformer current with a Hall effect device and terminating each half cycle before core saturation can be reached. Or in other words use current mode control.

This also provides some further up stream protection against user abuse, pretty typical for inverters because you have absolutely no control about what "Joe the idiot" is going to plug into the output.

 

[joao.caracas]

Thats true, the series capacitors will be a problem at this current.
I'm also afraid of using any Push-pull topology at this power and frequency because of the leakage flux.

 

[Warpspeed]

I'm also afraid of using any Push-pull topology at this power and frequency because of the leakage flux.

You can do a pretty good job on a push pull transformer primary with a pair of interleaved foils. The killer for leakage inductance is usually the wire lengths outside the transformer between foils and drains.
Its essential to keep those lengths minimally short.
Voltage spikes can be clamped or snubbed. They look nasty, but the actual energy contained in a very narrow high voltage pulse may not be that great.

Push pull still looks better at 12v.
But at 24v a bridge may be a better choice.
Above 24v a bride has all the advantages.

 

[joao.caracas]

Is there any online store for cooper foils?
Here in Brazil is very difficult to buy it.

 

[joao.caracas]

correction: Copper foils.

 

[Warpspeed]

Is there any online store for cooper foils?
Here in Brazil is very difficult to buy it.

You could ask a local transformer manufacturer, they should have foil in stock for fitting electrostatic screens between windings.
Check your phone book for copper merchants that stock copper pipe, sheets, rods and copper strip and angles. They may have foil as well, or can get it for you.

As a last resort, maybe a craft store that supplies artists and bohemian hippy types with weird stuff, like beads and coloured plastic, may have rolls of really thin brass and copper sheet.

One example of suitable foil available on e-bay:
https://www.ebay.com/itm/Copper-She...140499a&pid=100010&rk=3&rkt=4&sd=110468698857

The foil I have here is .005 inch or .127mm
A single layer of that 35mm wide will easily carry 18 amps rms.
Two pieces wound on together would carry 36 amps, and so on.

 

[Velkarn]

LLC is not suitable for low input voltages, try to calculate resonance elements end you will see it

 

[joao.caracas]

Velkarn I realized that ... the result of magnetization inductor is approximately 1uH and the resonance of the order of tens of nH. You suggest any topology for this converter?

 

[Easy peasy]

We have done a modified LLC for 24V to 350V at 3kW, with very high efficiency, it can be done, we used flip fets in // for the LV side, the best part was that our control allowed a low current turn off at any power level (quasi sine wave currents) so the turn off event of the LV mosfets is very soft (low RFI) and very efficient, ZVS for turn on.
And yes 3.2kW out at 21VDC in = 153 amps DC coming in to the unit - continuously...

 

[joao.caracas]

Easy Peasy, this sounds great! Which efficiency was possible to achieve? Can you share the design guidelines or the modifications with us? Is there any paper or document on this?

 

[Easy peasy]

define "share" please

- - - Updated - - -

http://totalmobilepower.com/Efficiency.html

 

[joao.caracas]

All you feel you can share, without disregarding all the hard work you had to create it, without breaking any patent or trade agreement . In short, everything that can be shared for educational reasons.