IGBT Gate Driver help needed

[lompilomp]

Hello,
I'm working on making an inverter, funny enough not to move a vehicle but rather to turn ~350Vdc into ~220Vac, single phase, 60hz. Seems simple enough.

My main goal is to have the whole system be as simple and straightforward as possible, with the least amount of components to have a working, reliable inverter.

I intend to use an EGS002 board https://www.lz2gl.com/data/power-invert ... eet_en.pdf, which pretty much does 99% of the work for me. These boards are dirt cheap, around $3 a piece when you buy a couple. They do pretty much everything you'd need a 50/60Hz inverter brain to do, monitors output voltage (through a step-down of course) and maintains the output as smooth as possible. Even has 2 IR2110S gate drivers https://www.infineon.com/dgdl/Infineon- ... c80333167e which are capable of delivering 2.5A peak, switching between 0 and 12V+.

So I currently have 6 Toshiba MG300J2YS40 300A, 600V Half-Bridges that I got for dirt cheap https://datasheetspdf.com/pdf-file/5650 ... 00J2YS40/1. From my understanding, just two of them making a full bridge should be enough for a ~12-15kW continuous load at 220v. I intend to play around with these a bit, but ultimately want to be ABLE to drive either multiples of them or simply bigger ones. (I have 4 other absolutely MASSIVE half-bridges, but we'll burn that bridge once we cross it )

The main concern is what sort of current is required for driving IGBT's with larger gate capacitances. I know that it's related to switching frequency, but I have no idea the frequency that the EGS002 board operates at. Some places online state 23.4kHz, which could be right, as 23,400/60 gives you about 390 pulses per cycle. I might actually be able to test this as I have an oscilloscope BUT have actually never used it and don't really know how but I am definitely gonna have to learn to use it for this project.

I've seen inverters that use the EGS002 and instead of using gate drivers use a very crude "totem pole" configuration of TIP41C and TIP42C transistors https://www.st.com/resource/en/datasheet/tip41c.pdf which should be able to handle 6A and 10A peaks. Those inverters use MOSFETS rather than IGBT's, turning 48Vdc into AC and passing it through a comically oversized transformer to turn it into 220Vac. And so my main concern here is the switching speed, and the 'lag' that adding such a configuration would add to the already 'slow' switching speeds of the IGBT's.

I see that the openinverter design implements a 2.0/2.5A IGBT driver as well, that seems to have somewhat similar specs to the IR2110 that the EGS boards have. What's the largest IGBT people have used on that driver successfully?

ANY input is appreciated, I have most of the components ready to start testing but before I start to work on a "design" I would love to have someone with experience point me in the right direction.

Thanks,
Brian

[johu]

The openinverter design uses a 5A driver these days, the Si8261BCC with 4.7 Ohm resistors.
Even with the 2.5A version I ran the 1200V/400A bricks in Polo for 8 years. Also 600V/600A bricks have been used successfully.

There are other options for a similar price that offer more protection features, others will chime in I'm sure

EDIT: these were used at a humble 8.8 kHz, 23 kHz will overload the little DC-DC converters.

[lompilomp]

The openinverter design uses a 5A driver these days, the Si8261BCC with 4.7 Ohm resistors.
Even with the 2.5A version I ran the 1200V/400A bricks in Polo for 8 years. Also 600V/600A bricks have been used successfully.

There are other options for a similar price that offer more protection features, others will chime in I'm sure

EDIT: these were used at a humble 8.8 kHz, 23 kHz will overload the little DC-DC converters.

I see the formula you put in the wiki P=Ug²/Rg*(tr+tf)*fPWM

Is the drive power completely unrelated to the gate capacitance?

I ask because oddly enough, the massive bricks I have have a slightly lower tr+tf, and so according to the formula would require less power? It's confusing because the gate capacitance of these 300A bricks is 27,000pF while the massive bricks' is 87,000pF...

Either way it seems like I'm gonna need ~1w of converter plus the headroom.

Do you think there's any chance that the igbt would work properly WITHOUT a negative gate drive or is that a big no-no?

Thanks for the help!

[johu]

I see the formula you put in the wiki P=Ug²/Rg*(tr+tf)*fPWM

Is the drive power completely unrelated to the gate capacitance?

I ask because oddly enough, the massive bricks I have have a slightly lower tr+tf, and so according to the formula would require less power? It's confusing because the gate capacitance of these 300A bricks is 27,000pF while the massive bricks' is 87,000pF...

Either way it seems like I'm gonna need ~1w of converter plus the headroom.

Do you think there's any chance that the igbt would work properly WITHOUT a negative gate drive or is that a big no-no?

Thanks for the help!

Yes the formula is a bit rough, that why I also wrote "If this gate resistor is less then 4.7Ohm of this driver this means that tr and tf will be a bit higher as well". So you could further approximate: if they state a 1.5 Ohm gate resistor you could multiply tf/tr with 4.7/1.5.
The DC-DC converters that were shipped with the latest version were MGJ2D051509SC (2W, -9/+15V). Unipolar switching is a big no-no especially for the low side gate. It is capacitively coupled to the collector which sees high dV/dt when the high side turns on. This dV/dt couples to the gate and turns it on.

[arber333]

The DC-DC converters that were shipped with the latest version were MGJ2D051509SC (2W, -9/+15V).

I really liked those MGJ2D051509SC. They were able to withstand SC for unlimited period apparently. I actually rebuilt quite a few driver boards after IGBT blasts and those DCDCs were OK each time even though drivers were toast.

I recommend you go by Prius inverter route for your inverter. At least so you get comfortable with the concept. DIY inverter allways means a lot of transistors burned before one comes to the same conclusion OEMs reached long ago.
If you insist on DIY then i recommend you still buy one cheap inverter be it Prius or Outlander so you can observe how the current path is built. I figured that has a lot to do with ringing or any losses you see in inverters...

[lompilomp]

I connected them raw to the egs002 module just messing around (12v 1.5A current limited of course).

To be honest I have never used an oscilloscope before and it doesn't help this one is ancient as well haha. So I don't really know what I'm looking at, but here's the AC output of the H-Bridge. It seems quite rough but from what I understand it needs a capacitor and inductor as an LC filter. And it probably doesn't help that the EGS board was running blind with no feedback.

Anyways, I was looking at an interesting IGBT driver. Its called the "EXB841" and seems to be one of those weird All-in-one deals, with its own DC-DC converter for negative gate switching and everything. They cost very little, around $4 a piece. Says it can handle 4A. Wonder how those would do?

[lompilomp]

OH it finally just hit me that the driver for the top gate needs to have a reference to the CE voltage in the middle of the Half bridge, so you HAVE to have either an isolated DC-DC converter or a gate driver with a built-in reference like the Vs in the IR2110?

After looking at the datasheet of the EXB841, am I correct in saying that it wont work for the high-side gates?

Well, at least I figured that out before doing something stupid.