EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: FlatPad on May 19, 2021, 02:24:15 pm
-
I try to understand how to properly drive IGBTs, especially since i want to go to higher voltages.
Q29 is how i normally do it, works (sort of) with one single PSU and 12V
But there's no isolation and when the voltage supplied to the IGBTs get higher (as in 48V or 400v) they will stop opening right?
Now i want to go upwards in voltage and the final goal is to be able to do 400V somewhere in the future without electricuting myself or anyone else and not disturb the neighbours.
i'll use a 48V source i have until i know more.
Can i have some feedback, can i place an optocooupler like this?
Feels wrong to supply 48 or higher voltage to the optocoupler, but i don't really know what to do here.
I have removed alot of stuff to concentrate on driving the IGBT.
Hopefully my schematic will be shown here:
[attachimg=1]
-
IGBTs must never be allowed to operate in the linear region, even for microseconds. They do not share current well, so one particular area of the transistor will get hotter, carry more of the current, and go into thermal runaway.
You must use very fast drivers to turn them on rapidly and turn them off rapidly. Also, the drivers must deliver a lot of gate current to charge and discharge the gate capacitance quickly. So, you cannot drive them directly from opto-couplers, the rise/fall times are too slow. Also, the connections of your schematic seem all wrong. The high-side transistors need independent floating gate drivers for each half bridge.
Jon
-
Use a proper driver IC like the STGAP2D if you can find some.
-
Thanks both of you :)
Been playing with some ideas due to your guidance, also i looked at rise and fall times of some components, seems there are alot of stuff to have in mind to avoid problems later.
But hey that's what i'm trying to understand :)
I didn't know i had to switch fast, tho i guess i have been lucky before, likely because the components has been so damn oversized compared to what i've been using them for.
I have to try and let the smoke out of an IGBT just to see it explode :D
Sounds very doable :D
-
One consideration is that you don't strictly need full galvanic isolation on every IGBT channel. A bootstrapped half-bridge driver (x3 for a three-phase bridge) such as an IR2114S doesn't require isolated power supplies on each of the top IGBTs. Instead, you could isolate the gate-drive signals at the logic level and use a single isolated power supply. The IR2114S also has desaturation detection as a means of short circuit protection which would be well worth investigating.
The level-shifted bootstrapped drivers are useable up to a few kW and certainly up to 800/900V. And if you put the isolation barrier in the logic level side, it can make the PCB layout a bit less fussy and avoids potentially running the signal traces close to the HV.
Feel free to let the magic smoke out, but please be concious of how much energy is involved as you scale up in voltage, that magic smoke can often be accompanied by some magic shrapnel from the IGBT packaging.
-
I'm switching 120 amps peak in my motor application.
Both Gate driver(STGAP2D) channels being internally galvanically isolated avoids undershoot complications.
No need for isolated supply to the driver's output stages, just normal Bootstrap.
I use series resistors to prevent spikes from overcharging the bypass capacitors.
-
I agree with all of the above:
Use dedicated gate driver IC's. IR2103 is another one. It is very common for such IC's to have an internal dead-time delay to prevent shoot-through.
So these IC's do 3 things:
1. Galvanic Isolation.
2. Strong gate drive pulses.
3. Dead-time generation.
-
Thanks, however i'm trying to learn and understand what i do more than building an actual device to use.
Sure, at some point i will be building something for real, but i have to learn it first.
Right now i'm building a sequence that is supposed to control fall times and risetimes of transistors that are going to control the voltage to the IGBTs.
I feel i'm doing something that really isn't needed, but it satisfy my mind :)
I intend to show what i come up with.
To make this "sequence" i try to learn simulation in KiCad, there's progress but i struggle to find models for ngspice.
-
Rise and fall time is set by the Gate drive current. The Gate charge and discharge current can be set independently with 2 resistors and a diode.
Some driver ICs have separate outputs for turn On and turn Off.
The Gate plateau voltage needs to be known.
-
I feel i'm doing something that really isn't needed, but it satisfy my mind :)
I intend to show what i come up with.
To make this "sequence" i try to learn simulation in KiCad, there's progress but i struggle to find models for ngspice.
I see where you're coming from. It's always good to work through problems, simulate, experiment and get a good grasp of how different properties of the circuit are affected by component choices.
For most "small" inverters, the best option, practically speaking, when building is likely an IC since they have so many nice built in features that can really simplify the design. When it comes to simuilating and understanding, its probably a good idea to start with a discerete design (least if its not just due to the lack of good simulation models). There is a good description of gate-driver circuits in "Power Electronics" by Ned Mohan (other books exist and hopefully you'll get a few more suggestions), the book is generally geared around "high power" power electronics, but most of the principals are universally relevent. Mohan also covers de-saturation detection which is a good method of adding in some short-circuit protection.
I don't know where you're at generally in terms of power electronics understnading, but based on your original circuit, I'd recommend going back to basics with a simple boost converter (one low-side switch) and when you've got a descent and reliable gate-drive, maybe adding isolation then trying a buck-converter (one high-side switch), then synchronous buck (high and low side switch) and then you'll be good to go with an full inverter (which, with a motor as a load is just a three-phase synchronous buck converter).
Of-course you can go straight to a 3-phase inverter, but building up from a simple switch through a couple of the basic converter topologies be a slightly gentler route up the learning curve and helps build a few of the basic measurement skills.
-
Seems figuring out that sequence was harder than i expected, i might revisit that later.
For now i will control "dead time" from my programming of the microcontroller.
About driving the IGBTs, i played with some ideas about switching fast enough.
I framed the 3 diffrent ideas i came up with.
Made a rail that are supposed to be 20V higher than the highest voltage supplied, have to change some resistors to get the voltage about right.
I'd like to ask if either of them are better than the other or maybe really bad.
I my mind the comparator should be the best.
[attachimg=1]
-
A High-side Gate drive circuit needs to be floating and referenced to the Emitter of the IGBT.
-
I recommend the ADUM5230. It includes both an integrated isolated DC:DC converter, for the high side and gate driver. It's much easier than messing around with opto-couplers.
https://www.analog.com/en/products/adum5230.html (https://www.analog.com/en/products/adum5230.html)
-
Floating and referenced, i don't know how to translate that into my language and understanding.
Thought IGBTs needed to have a voltage about 20-25V higher than the voltage on the emitter?
I can't wait until i get them in the mail to start experimenting.
I guess that will make me understand the component better.
-
Floating and referenced, i don't know how to translate that into my language and understanding.
Thought IGBTs needed to have a voltage about 20-25V higher than the voltage on the emitter?
I can't wait until i get them in the mail to start experimenting.
I guess that will make me understand the component better.
The high-side driver's output stage is ground referenced to the Emitter of the high-side IGBT or so called "switching node"
It floats up and down with the switching waveform. So the Gate is being driven with respect to the Emitter regardless of voltage activity on the Emitter.
Depending on the manufacturing technology, there are often limits to how many volts the high-side driver is allowed to float by.
This might help, https://training.ti.com/high-side-devices-bootstrapping
It's possible to drive high-side transistors with respect to the low-side ground but it becomes difficult to do properly at high voltages.
-
This interesting driver appeared in my targeted advertising, UCC23513
It is designed to be driven like an LED, it actually has a HF oscillator that sends control signal to the isolated side.
The inputs of 2 devices can be cross connected to provide interlocking.