https://github.com/metanurb21/esp32_igbt_driver
ESP32 PWM IGBT gate driver with rotary encoder and OLED screen driving an induction heater.
https://github.com/metanurb21/esp32_igbt_driver
esp32 full-bridge igbt induction-heating
Last synced: 3 months ago
JSON representation
ESP32 PWM IGBT gate driver with rotary encoder and OLED screen driving an induction heater.
- Host: GitHub
- URL: https://github.com/metanurb21/esp32_igbt_driver
- Owner: metanurb21
- License: gpl-3.0
- Created: 2025-02-18T17:21:13.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2025-02-28T02:23:22.000Z (3 months ago)
- Last Synced: 2025-02-28T10:17:46.796Z (3 months ago)
- Topics: esp32, full-bridge, igbt, induction-heating
- Language: C++
- Homepage:
- Size: 9.24 MB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# ESP32 Driving IXDI & IXDN IGBT gate drivers used for a DIY induction heater.
This is a work in progress.
I wanted to build a high power induction heater capable of melting metal, copper, iron, aluminum etc.
Getting it just right turned out be a bit of a project as you can imagine. I picked up a little here and a little there from scouring the internet.
I kept finding quick little builds with a few cheap caps and a coat hanger with 12v supply claiming to be producing 600+watts. The math didn't add up. The popular trend of heating a thin box cutter blade red hot was just not doing it for me. I was soon to find out that to liquify a pound of copper or steel, I had to get $$-Serious-$$.
Given my inexperience with circuitry and schematics, I wanted it to be as simple as possible until I found my way. I'm always playing around with micro controllers and I am pretty comfortable coding, so I figured if I could do most the signal work with code, it should keep the circuit design to a minimum. An extra bonus doing it this way is it lets me change up the type of PWM signal I want without necessarily creating a new PCB. I initially built the inverter for a another Tesla coil build I have planned. "Ramped QCWSSTC anyone?" Here is what I came up with so far.
## The Board Design
I made the schematic in [flux.ai](https://www.flux.ai/metanurb21/esp32-pwm-ixd630-igbt-driver-v2?editor=schematic). Here is a link to [The board Gerber files.](https://github.com/metanurb21/esp32_igbt_driver/blob/main/files/metanurb21-esp32-pwm-ixd630-igbt-driver-v2-Gerbers-Version2c165305.zip)
[The components list](https://github.com/metanurb21/esp32_igbt_driver/blob/main/files/pick_and_place.csv)
[The ESP32 driver code for Arduino IDE](https://github.com/metanurb21/esp32_igbt_driver/blob/main/code/esp32__igbt_gate_driver.ino)
This setup is probably not the perfect PCB or design, like I said, I am just learning. This setup does not operate a [phase loop lock](https://en.wikipedia.org/wiki/Phase-locked_loop) to auto tune, it's all manual. I have an LED that glows when I am in resonance and I can hear and see the current being drawn. My buddy calls it "Tune for maximum smoke" :wink:. The PLL approach is my next build attempt. It does however work pretty well from 10khz to 100khz. I have two coils of different sizes and they resonate at around 82khz and 52khz. The current code is set to range from 10khz to 100khz. To change that, you just change some of the variables. I decided to go with a rotary encoder instead of an analog read from a potentiometer on the ESP32. It's a lot more accurate, doesn't bounce around at all and does not suffer from interference. I have ran this thing for hours and hit it pretty hard with decent voltage and amps. So far so good. Pay attention to what's going on and keep all the components cool. This is done with a combination of fans and water blocks with a pump circulating cool water around all the hot spots.
Liquid cooled things:
- The IGBT'S.
- The IXDN gate drivers.
- The coil and the tank capacitor.
Air coolled with heat sinks:
- The circuit board and anything else you want.
## The components you will need to build and acquire.
There is no single or correct way to build an induction heater. I have built 3 or 4 different boards with different approaches. Each version required slightly different setup and components. This really stared for me from an article by [Teslascience Hacks](https://teslascience.wordpress.com/autotuning-9kw-induction-forge-version-1-2/). Check it out if you want another way to built one. The things you have to get right are the IGBT gate signals, adequate protection from voltage and current spikes and the proper balance of values to make the heating coil resonate so that it produces the maximum and most efficient electro magnetic coupling (power) to the intended work piece.
## To Build
Apart from the driver board, you will need to build:
A high power IGBT full-bridge inverter. There many schematics available from a google search.
A copper pipe heating coil. 1/4" or 3/8" pipe.
A coupling transformer [I used 5 of these](https://www.easternvoltageresearch.com/ferrite-core-5000-1/) with 10 turns of Litz wire (the red multi stranded magnet wire rope you can make).
A GTD (gate drive transformer). You can make or buy from [eastern voltage research](https://www.easternvoltageresearch.com/gate-transformer-kit-5000-1/).
A liquid cooling network for the IGBT's and the the tank capacitor. Common sense and some basic building and fabricating skills required.
Various different value power sources depending on the approach you take. 5V, 12V, 15V.
## To Acquire
Other items you will need are: [A 30A Variac is recommended](https://www.ebay.com/itm/375079366063). [Various capacitors](https://www.ebay.com/itm/234559065702) and [snubber caps](https://www.ebay.com/itm/396177418199?_skw=snubber+capacitopr&itmmeta=01JMDDCM8SQPJGYN1PYTKN86BS&hash=item5c3e03a7d7:g:VE4AAOSwAYdno4WP&itmprp=enc%3AAQAKAAAA0FkggFvd1GGDu0w3yXCmi1c9mWgMXZX5fwHpwCae7OcaRXv1trcV3pJHvxD7N%2BYRzYReG5U4lVD76NHHILRzqmooxb7L2cbUsKP2JHcCTopAn1pdaqD1sa0glHSgaajgOATGAaY3qhzo%2BHvpVgDuLP%2FbjWafyUHbSjMkNS7ixHWHxHD6KvJA9vZjk2rkU%2FkQ3LNl9oBzikCl3rA7CmxaXTL2N42wPFiJ0QYJF6qiNYs4mD%2FDD%2FP0W9Pats5fIZlX6m4r4gOI%2FzUm9CFR3ZX5Mfk%3D%7Ctkp%3ABk9SR6bGsq2jZQ). Some copper bus bars, A pond pump. Some high-heat rubber pipe, A large full bridge rectifier, Heat sinks, fans, water blocks, large gauge wire, magnet wire, some high heat fiberglass pipe insulation, a kiln crucible and I would suggest a prinkling of some ceramic fiber insulation here and there to protect things. It gets kinda scary your first time!
Most of these things you can find on eBay, Amazon and your local hardware store.
The larger capacitors, snubber caps and IGBT's you will need to source from eBay or Alibaba etc.
For great quality stuff geared towards high voltage enthusiasts, I use https://www.easternvoltageresearch.com/
For the large current water cooled capacitors, I recommend [Dawncap.](https://www.dawncapacitors.com/home). These things can get really pricey if you go top name brand. I would suggest for getting started, you buy from Dawncap. I have bought a few different ones over the past year, and I can attest that they hold up and get the job done. The caps I have are:
[DAWNCAP DCC-5H 2.6UF 700V 950A 400 KVAR Water cooled large current Big Current Resonant Capacitor](https://www.aliexpress.us/item/3256805434541389.html?spm=a2g0o.order_list.order_list_main.21.21ef1802Jx87JR&gatewayAdapt=glo2usa)
[DAWNCAP DCC 3.5UF 1200V AC 120A Water &Wind cooled large current Big Current Resonant Capacitor](https://www.aliexpress.us/item/3256805388886798.html?spm=a2g0o.order_list.order_list_main.26.21ef1802Jx87JR&gatewayAdapt=glo2usa)
[DAWNCAP DCC 2UF 600V 650A Water cooled large current Big Current Resonant Capacitor](https://www.aliexpress.us/item/3256805387045331.html?spm=a2g0o.order_list.order_list_main.31.21ef1802Jx87JR&gatewayAdapt=glo2usa)
:open_mouth: 950A!!!, getting real serious ya. All have served me well.
For IGBT bricks, it's nice to have quality bricks like Infineon etc. You can get them used on eBay or you can get new cheaper ones on Alibaba. Again, you get what you pay for but while experimenting and blowing crap up, It's fine to go cheaper and then you can replace with the quality products when you are ready. [Some cheaper ones here](https://www.aliexpress.us/item/3256805870685571.html?spm=a2g0o.order_list.order_list_main.277.21ef1802ginfNj&gatewayAdapt=glo2usa). One note, it's not always better to go bigger, like 400A or 600A. The bigger they get, the slower the gate times. We want to keep things as fast as possible since we are dealing with very high frequency switching.
Once you've blown everything up once or twice and you are finally up and running, you should be looking something like this. Or better!. :)
## Driving the IGBT gates.
The idea is to rapidly switch on and off the gates of each of the dual IGBT bricks in an H-Bridge configuration. We are taking 120v or 240v AC from the wall, converting it to DC using the rectifier, using the micro controller to switch the gates using a PWM signal.There is one input on the board to receive the PWM signal which go to the IXD gate drivers. There are two IXDN gate driver chips on the board and are in an inverted and non-inverted configuration. They will create two signals 180° out of phase from each other alternating the on and off states of the gates in the H-Bridge configuration. There are two outputs from the controller PCB which are connected to a 1:1:1:1:1 GDT. One primary and two sets of four secondaries. These are connected to the IGBT's gate keepers. Here is a snapshot of me probing the gates on my oscilloscope. Nice and square. No major spikes or ringing. Could use a tad more dead time (overshoot), but that will get handled further down the pipeline.
## The Gate Keepers
The IGBT gates are directly connected to parallel 3Ω 5W resistors and shottky reverse diodes which receive the input from the GDT. This helps with giving a little dead-time to guard against shoot through. That's when the signals overlap and two gates get turned on at the same time creating a short to ground on one side of the bridge. Not good.
## The Coupling Transformer
The high-side of the IGBT's are connected to the coupling transformer via a large 3.5UF 1200V AC 120A DC blocking capacitor. This coupling transformer is fashioned from stringing together five large ferrite toroids taped together with kapton tape. Wound around this (for my setup) is ten turns of litz wire. This is a wire rope made from twisted together thinner 22awg enameled copper wire. You can do this with a screw gun or drill. This type of wire setup allows for much larger currents without getting overly hot. Initially I tried with 3awg thick jumper cable wire thinking it would be fine. It melted. :grin: This creates high frequency AC through a coupling transformer around one leg of the copper coil.
## How the coil works
With a good impedance match between the TX (experiment with the number of litz wire turns), the copper coil and the resonant tank capacitor, we create high frequency AC current flowing through the coil generating a changing magnetic field, which in turn induces electrical currents (eddy currents) within the work piece, causing it to heat up due to its electrical resistance. At some point the metal reaches it's [curie point](https://en.wikipedia.org/wiki/Curie_temperature). This is temperature at which a material loses its permanent magnetism and tends to glow white hot. Note: the work piece does not have to be a ferrous metal (metals that contain iron as their main component). The image at the top of the page is melted copper. It takes a bit more continues power after this to reach the melting point though. You can google the melting temperatures of your intended victim. If you can crank a few more amps into it and keep the resonant frequency after curie point, it begins to melt. The goal is not always to melt things of course. This controlled heating process can heat treat metals to different depths depending on what you are trying to achieve. As you can see below, even the fiberglass protective shielding on the work coil is starting to char. :laughing: Must be doing something right.
## Short Video
[](https://www.youtube.com/watch?v=emqwikQgqFo&t=5s)