Designing dumb EGS002 LF inverter

[xvel]

I want to build an iverter (230v) based on EGS002 from the ground up, input voltage 52.8v - 62.4v, power <= 2500VA (I have couple of toroidal transformers from 600VA to 2500VA) I cobbled it together with IPP086N10N3 transistors and it worked initially until I tried a PC with 650W be quiet pure power 12 PSU, I even had 3.6ohm resistor in series but it did not help, lm393 from EGS002 board was removed so there was no fast overcurrent protection but even if there was, I read that this alone can fry mosfets when using toroidal transformers and then I would anyway would not be able to power such PC from it.

Now I want to choose some mosfets and make the overcurrent protection work in such a way that it will save them from frying but allow for such loads with big inrush currents to start up without tripping it. First, are 100v mosfets enough for 62.4v input? Will I be able with carefull layout, keeping everything close to make sure that there will be no spikes above 100v? Or is it better to play it safe and just use 150v mosfets? There is also requirement to use builtin IR2213 to drive the mosfets, for now I thinking of IRFP4468 or IRFP4568, will IR2113 be able to drive them? What dead time will be good? As of overcurrent protection, is there simple dumb solution for it to work as I explained before? From the IRFP4468 datasheets I see that it can withstand 400A for 10ms which should be just enough to withstand ATX PSU inrush current of probably 60A peak (toroid has ~6.5 ratio), IRFP4568 SOA chart is junk and impossible to read but looks 400A 10ms also, probably is a little lower. I'm ok with adding some small resistance in series if that is necessary to make transistors happy but still want somewhat functioning overcurrent protection. I would appreciate any help with this.

 

[solar8484]

It would be best if you could post schematics and photos of the power board and transformer. There are many variables that can blow up FET's but one basic thing you need to have is a good choke inductor between the power board and the transformer that does not saturate under the target max load. In case you haven't seen it, @tinyt has a great thread on his build and you should study it thoroughly (https://diysolarforum.com/threads/another-inverter-build.53006/).

 

[xvel]

As I said, it was just cobbled together on an universal board, it's partially disassembled now, there is no schematic, it was basically circuit from eg8010 datasheet (page 9) difference was that lm393 was removed, but there was still IFB connected to 50mohm shunt on DC side and additionally 3300uf capacitor connected before shunt (which doesn't really make sense from the capacitor effectiveness perspective), there was also inductor on the primary side of 80uH made from 5 stacked yellow powder cores (24mm od, 14mm id, 8mm h) which had likely little too much turns as it got to 80C during operation. On the secondary there was 2uf of capacitance. Transformer was new custom 230v to 2x35v 600VA toroid with 35v windings as primary in parallel. Idle power was 8W and that was including powering EGS002 12v and 5v rails from 60v source using linear regulators. I had also 3.6ohm series resistor on the 230v side for testing, despite that at the moment I tried to power up a PC with 650W PSU the mosfets shorted. Before that it worked for a couple of hours powering smaller things and another PC with Corsair SF450 PSU which I turned on on the inverter 2 or 3 times with "success". Max continuous load during testing was about 300w for half hour, transistors were barely warm the whole time as was the transformer.

But this isn't really relevant as I want to build the new one from the ground up, just want to make it right enough to withstand biggish ATX PSU power up spike somehow, I don't plan to power anything more/different demanding than that from it.

 

[solar8484]

there was also inductor on the primary side of 80uH made from 5 stacked yellow powder cores (24mm od, 14mm id, 8mm h) which had likely little too much turns as it got to 80C during operation.

It seems your choke is saturating with the PC PSU (a high peak current non-linear) load that could lead to high di/dt and dv/dt which in turn could be inducing FET turn on's resulting shoot-through short-circuit in the H-bridge and blowing up the FET's. In general, you need high permeability (125u ideally) cores for the choke to avoid saturation under load.

 

[xvel]

But choke that is not saturating at that surge current (which is 20-40x continuous current for such small inverter) would need to be of similar size as the LF transormer or am I missing something?

 

[solar8484]

But choke that is not saturating at that surge current (which is 20-40x continuous current for such small inverter) would need to be of similar size as the LF transormer or am I missing something?

In general, inductors can start to saturate at current levels as low as half of the rated current @ 40K temp rise. Given the 80C temp you observed, it seems to be operating well above its rated current.

Have you measured the saturation current of the choke? If not, this post may help ( https://diysolarforum.com/threads/another-inverter-build.53006/post-790239).

 

[xvel]

tinyt made and measured inductor here: https://diysolarforum.com/threads/another-inverter-build.53006/post-836488
8 cores similar to this: https://datasheets.micrometals.com/MS-225125-2-DataSheet.pdf and 9 turns, he claims 27uH at 200A which doesn't seem possible (the rest of the measurement also doesn't make sense to me), that's 1800 ampere turns, such core has 7% initial permeability at 1800 ampere turns and with nominal 101.1uH (8x 156nH/N² 9 turns) it should be ~7uH. What's going on here, am I wrong with my calculations? (this isn't my topic, I don't really know what I'm doing )

 

[solar8484]

The number of turns is reduced by number of stacked cores squared. So, 8 stacked cores only needs 1/64 the number of turns required for a single core to achieve the same inductance. You should measure the saturation current as tinyt did.

Keep in mind that using lithium based batteries (very low internal resistance), large capacitor banks (high burst energy) and toroidal transformers (very low leakage inductance) make for a very unforgiving setup that can violently exploit design/build flaws. Given that you are learning, if possible, I suggest you start with a more forgiving (but less efficient) setup using AGM/SLA batteries (higher internal resistance), few/one capacitor (low burst energy) and EI transformers (higher leakage inductance) to avoid blow ups while you verify/improve the basic inverter design/build.