The motor controller power supplies I took these off of was on a bearing roller grinder. The rollers made on that grinder are used in wind turbine shaft bearings. Each roller weighs between 30 to 60 pounds. I also got a bunch of motor braking resistors that came off the same machine. Have to pull one out of the warehouse but I think they are either 2,000 or 2,500 watt resistors. If you want one of them let me know.
These are used in inverters, VFDs, and servo drives typically in a 3-phase H bridge configuration. They come in all kinds of configurations combining 2 or more IGBTs in a single package, additional diodes pointing different ways, drivers, etc. It's common to find the entire 3-phase H-Bridge with drivers and some amount of control built into a single package; Open up any smaller modern VFD or servo drive and it will likely have a single. combined module. Pop one of these big guys open and they are usually filled with a clear jelly like silicone goo. You can see that its a number of smaller transistors all paralleled together with bond wires to get that 200A Max goodness. Basic idea is rectify the incoming AC, filter it, then feed the H-Bridge with the DC bus. Use your control logic to chop the output back into 3-phase and feed it to the motor as required.
Because there are many transistor dies in parallel, these large devices are only specified for fast hard switching to ensure that they all turn-on/off at nearly the same instant to minimize the detriments of uneven power sharing due to the natural variation in gate/base thresholds amongst the dies.
@@ChrisSmith-tc4df That's one of the advantages of using the combined controller/driver/IGBT modules as the integrated control takes care of the intricacies of driving it properly.
It would have been nice to look at the input capacitance of this device and think about how difficult it is to drive with any speed. It would be interesting to see if your curve tracer can drive it...
Wow, most bricks like that have the freewheel diode across the transistor internally and the third connection is usually another transistor for a half bridge. ❤
The power supply might be sweating more than the transistor, unless one reduces the gate voltage. If it is just for the current one could test a power MOSFET like the IRFB3077PBF with 3.3mOhm resistance and 210A continuous current in a TO220AB case (with heat sink). Higher currents for short pulses may be provided with a capacitor bank like when you test the saturation behavior of coils for switching converters.
I've worked with these before, the circuit that drove the gate put out +/-15 V! to turn it on and off. I don't remember what the actual threshold voltage was but i suspect it was much less than this and the designers were merely being cautious. This is also why they drive the gate negative in the off state, there's less risk of grounding issues causing it to not be fully off.
The IGBT's i'm familliar with are in the same package, but there's actually two IGBTs in there to create a push/pull output that can be driven to either +v or ground. The contacts are C1 E2, and the common C2E1 in the middle (schematic wise, the C2E1 contact is one of the outer ones physically - the same one that's the flyback diode in this single IGBT module), then you have 2 gate and 2 emitter contacts on the little contacts on the side for driving each of the IGBTs in the brick
These actually often require a few amps gate current to switch fast enough, you don't want to operate at 600 V 200 A 120 kW loss for more than 100 ns or so during switching. The gate charge is about 1.9 micro columb and to be (dis)charged fast.
I’ve seen similar in big industrial Hypertherm plasma cutters. The one I looked at was destroyed by the emitter screw coming loose, taking out optocouplers on the main board.
1) 2:48 There's already what appears to be a snubber diode running from emitter to collector. I gather they put some built in protection, but allow you to increase that protection for higher inductive loads? 2) If you search for brake chopper circuit, the transistor actually appears across the DC bus with a brake resistor in parallel with the diode. It doesn't directly drive the motor as per your diagram.
Like you siad, these series of modules from Infineon are meant as brake choppers for traction applications/VFDs and the like, therefore they add the chopper diode internally to the die so you have it as a single package. This is great for cooling and makes a smaller footprint. Big difference between power MOSFETs and IGBTs is the lack of inherent body diode in IGBTs so they also have to be added separately, but they are (almost) always already integrated in IGBT power modules.
I assume these are the jobbies typically used in VFDs/VSDs. The ones I am familiar with take around 700 to 800V DC and chop it back to AC (with filtering) for the motor output. I believe several are run in parallel and as you say a lot of cooling air (they are very noisy beasts due to the fans). Largest drive I have seen is 1200KW output at 415V 3 phase AC (several modules in parallel) which at the time (2010) was also the largest made in the UK. More power!
I love your videos. I love your content. But my issue with this is you fidget with the thing you are showing. Please put it down and let the camera focus, let the video compression algorithm settle down. Pick up and reposition as necessary but please let us SEE what you are talking about. Thank you. Keep up the great work.
You can easily test the full current with a charged rather small capacitor as power source. If you have som 100 uF 400 V electrolytes or similar, you can do meaningful switching test. Pay attention to saftey, loopinductance and switching speed. You can use a simple aircore inductor as load.
There's a project for the 3D printer content guys. A 3000 watt breadboard. The holes would be 2.5 cm in diameter. Then, of course, an adapter so it could be used with an Arduino Nano.
IR camera can maybe see the back plate better, if you taped it over with black tape? Really would have loved to see it on your curve tracer, compared to a 3904 or something jelly-like
You could replace the master power switch in the breaker box in your house with that, typical house master breaker is rated about 200 amps 250v Useful if you want to install solar power privately and not back feed it to the power company Add a current sense and esp32 and a voltmeter chip Get the voltage stats, current of every circuit in the house, and when the current sense transformer see a mains power outage switch dat igbt off and switch igbt on for solar And can be operated over Bluetooth on your smartphone And that way the power company can back feed and trickle charge your solar backup, and solar panels only are used when mains don't work, power company ain't going to put solar panels on my house and claim partial ownership of my private property that's how those solar power subsidies technically work They pay for the solar panels, but you don't own it
The vfd drives these things are in have tons of protection in them designed to handle that very situation, as it does happen - they have very fast analog overcurrent protection that will open the main contactor immediately when a failure happens, the drive then just shows an error code until the IGBT is replaced. The failures are very mundane and uneventful actually. The only time there's a problem is when one shorts to the gate, and the gate isn't isolated anymore - then it also blows up the gate drive module(s) as well, and makes a mess on the control board.
@@Clancydaenlightened These are BJTs though on the high power side - they don't like to have the polarity of the emitter and collector reversed - most have an intrinsic flyback diode across E and C - they'd blow up immediately if you put AC across them
couldn't resist! just press attenuate.. a real buzz! yup! agreed! IGBT's are purdy cool! well, except when they'r burnt up! a dip in the pool! good luck mr. fix it ! hand gestures are spot on!
0:21 expensive igbt too A dream transistor for a Tesla coil full bridge if you can afford 4 of em They can run higher current too if pulsed via an interrupter circuit With microcontroller you can spit out musical square waves, or even use an 80s soundchip If you got a beefy circuit you can run in continuous wave and play hifi quality audio, plasma speaker has better frequency response, more dynamic range, and is truly omnidirectional
And even run it at 250v mains You can reach close to that 1200v in a Tesla coil primary with inductive spikes after rectification you have about 400 volts or so, and inductive kicks can reach up to 700-800 volts
You can get similar half bridge (2 IGBTs) modules new for about 20 if you look around - 40 for 2 of them (4 igbts) isn't that expensive. There are lots of variants of these all in the same package with the same connections - they're like industrial lego
@@gorak9000The ones you want for this application are more like 80+ USD per module You want low rds on and drains source voltage (collector emitter) And you want lots of current and voltage capacity And what's the switching speed and pulsed operation rating I don't want a 700 volts at several hundred amps going thru a cheap 20 dollar transistor with crazy LC spikes 20 dollar transistor is fine for some small load under a few kw But running a 10 ft 20kw drsstc is different
@@Clancydaenlightened The 20 dollar ones I'm talking about are Fuji Electric, made in Japan. 2MBI 200F-060 600V 200A I have a set of them in a drive for a medium sized (7.5kw) CNC lathe spindle. I was shocked at how cheap they were. I'm not sure if they are the original part that was used in the drive, or someone before me replaced them. When I got the machine, the spindle drive had blown up, and someone replaced some of the parts, but they didn't find all of the issues and couldn't figure it out. I found the remaining issues and replaced the parts, and now it's fully functional again
Peak 240,000 Watts... 240KW... almost 1/4 Megawatt!!! -- I had to think that out several times but that's right wow.. That can handle the continuous thermal capabilities at full load w/ a heat sink or does the spec limit the total watts based on thermal constraints . Always enjoy your videos and have a happy thanksgiving if you're here in the USA
240kw is about 326HP. Some electric car have more power than 500HP. Besides some good charging stations have 150+KW charging capacities. Without these IGBTs electric car industry wouldn't exist.
I wonder if the gates of those monsters is susceptible to static electricity like their smaller MOSFET cousins? Do IGBT's incorporate static protection for the gates?
There are usually about 10 transistor chips in parallel and similar number of diode chips. Each about 10 mm square. They are placed on a ceramic PCB and connected with a few parallel thick (0.4 mm?) bond wires. It is probably the back of the PCB we see at the bottom, sometimes there is a heat spreader but I don't think there is one in this case. The PCB is covered in a clear sticky soft potting compound for electric insulation. A number of plastic tabs press the PCB against the heatsink when mounted.
@@urlkrueger The one I have looked at have no balancing resistors other than bond wires, copper traces and internal chip resistance. I think the chips are sometimes made to have a positive temperature coefficient by selecting on geometry and doping patern. Also these transistors are made for switching applications, they are probably not thermally stable in the linear region other than at very low power. I don't think you should build a linear amplifier with them. They are typically switched in 1 to 20 kHz with switching times well below 1 us. Similar concerns about stability apply to a single chip since all power transistor chips internally have many parallel transistor structures on the same chip. So part of a chip can go into a thermal runaway.
@@urlkrueger Do a image search for "igbt module inside" or teardown and you will see many images of the inside of similar devices. I maybe exaggerated the number of chips since most of these modules have two transistor functions in a push pull half bridge configuration, not one as in this example.
They're too slow for RF. They should work at a few khz, but they tend to be slow, and you need pretty beefy transistors to drive the gates fast to avoid a lot of switch loss.
They're 100% potted. I'm also curious to see the inside of one. I actually have a dead one that I replaced - I tried to open it, I ended up stabbing myself in the thumb with either a screwdriver or a knife, and haven't attempted opening it again. I suspect the way in is with lots of heat and scraping, or milling it down.
I'm wondering if anyone has thought of using the BMF to charge a capacitor or a battery. If this gismo switches in 10s khz with such a high wattage, a snubber diode will create a lot of wasted power and heat
I G B T CURRENT GROUND BATTERY TERMINAL VIP POWER VOLTAGES I CURRENT QURYNT AMPS BATTERY DC..AC ONLY HZ 120V LIVE..ONE WIRE NO AMPS WATTS.NO CIRCUIT .AC..DC DISCHARGED HAS LOAD... AC NO LOAD PUSH JLQY PULLED JRQN
I left the gate drive header disconnected by accident once, and the stray EMI fields of power-up managed to charge all of the gates at once causing the array to crowbar and explode all of them into high velocity shrapnel that continued to rain down over several seconds. Pretty much the only thing left of the transistors were the heavy metal base plates with a gnarly crater in the middle of each of the bonded transistor dies.
The motor controller power supplies I took these off of was on a bearing roller grinder. The rollers made on that grinder are used in wind turbine shaft bearings. Each roller weighs between 30 to 60 pounds. I also got a bunch of motor braking resistors that came off the same machine. Have to pull one out of the warehouse but I think they are either 2,000 or 2,500 watt resistors. If you want one of them let me know.
These are used in inverters, VFDs, and servo drives typically in a 3-phase H bridge configuration. They come in all kinds of configurations combining 2 or more IGBTs in a single package, additional diodes pointing different ways, drivers, etc. It's common to find the entire 3-phase H-Bridge with drivers and some amount of control built into a single package; Open up any smaller modern VFD or servo drive and it will likely have a single. combined module. Pop one of these big guys open and they are usually filled with a clear jelly like silicone goo. You can see that its a number of smaller transistors all paralleled together with bond wires to get that 200A Max goodness. Basic idea is rectify the incoming AC, filter it, then feed the H-Bridge with the DC bus. Use your control logic to chop the output back into 3-phase and feed it to the motor as required.
Because there are many transistor dies in parallel, these large devices are only specified for fast hard switching to ensure that they all turn-on/off at nearly the same instant to minimize the detriments of uneven power sharing due to the natural variation in gate/base thresholds amongst the dies.
@@ChrisSmith-tc4df That's one of the advantages of using the combined controller/driver/IGBT modules as the integrated control takes care of the intricacies of driving it properly.
It would have been nice to look at the input capacitance of this device and think about how difficult it is to drive with any speed. It would be interesting to see if your curve tracer can drive it...
Holy shit, that's one hell of a transistor!
Wow, most bricks like that have the freewheel diode across the transistor internally and the third connection is usually another transistor for a half bridge. ❤
$117 on Mouser, less than I expected.
Once again you've given us a WOW moment!
The power supply might be sweating more than the transistor, unless one reduces the gate voltage.
If it is just for the current one could test a power MOSFET like the IRFB3077PBF with 3.3mOhm resistance and 210A continuous current in a TO220AB case (with heat sink).
Higher currents for short pulses may be provided with a capacitor bank like when you test the saturation behavior of coils for switching converters.
I've worked with these before, the circuit that drove the gate put out +/-15 V! to turn it on and off. I don't remember what the actual threshold voltage was but i suspect it was much less than this and the designers were merely being cautious. This is also why they drive the gate negative in the off state, there's less risk of grounding issues causing it to not be fully off.
The IGBT's i'm familliar with are in the same package, but there's actually two IGBTs in there to create a push/pull output that can be driven to either +v or ground. The contacts are C1 E2, and the common C2E1 in the middle (schematic wise, the C2E1 contact is one of the outer ones physically - the same one that's the flyback diode in this single IGBT module), then you have 2 gate and 2 emitter contacts on the little contacts on the side for driving each of the IGBTs in the brick
These actually often require a few amps gate current to switch fast enough, you don't want to operate at 600 V 200 A 120 kW loss for more than 100 ns or so during switching. The gate charge is about 1.9 micro columb and to be (dis)charged fast.
I’ve seen similar in big industrial Hypertherm plasma cutters. The one I looked at was destroyed by the emitter screw coming loose, taking out optocouplers on the main board.
1) 2:48 There's already what appears to be a snubber diode running from emitter to collector. I gather they put some built in protection, but allow you to increase that protection for higher inductive loads?
2) If you search for brake chopper circuit, the transistor actually appears across the DC bus with a brake resistor in parallel with the diode. It doesn't directly drive the motor as per your diagram.
Like you siad, these series of modules from Infineon are meant as brake choppers for traction applications/VFDs and the like, therefore they add the chopper diode internally to the die so you have it as a single package. This is great for cooling and makes a smaller footprint. Big difference between power MOSFETs and IGBTs is the lack of inherent body diode in IGBTs so they also have to be added separately, but they are (almost) always already integrated in IGBT power modules.
I assume these are the jobbies typically used in VFDs/VSDs. The ones I am familiar with take around 700 to 800V DC and chop it back to AC (with filtering) for the motor output. I believe several are run in parallel and as you say a lot of cooling air (they are very noisy beasts due to the fans). Largest drive I have seen is 1200KW output at 415V 3 phase AC (several modules in parallel) which at the time (2010) was also the largest made in the UK. More power!
I love your videos. I love your content. But my issue with this is you fidget with the thing you are showing. Please put it down and let the camera focus, let the video compression algorithm settle down. Pick up and reposition as necessary but please let us SEE what you are talking about. Thank you. Keep up the great work.
You can easily test the full current with a charged rather small capacitor as power source. If you have som 100 uF 400 V electrolytes or similar, you can do meaningful switching test. Pay attention to saftey, loopinductance and switching speed.
You can use a simple aircore inductor as load.
I thought you might end up leaving a scorch mark in the middle of your notepad. LOL!! Impressive device Great for large industrial motor control.
We need a bigger breadboard :-)
There's a project for the 3D printer content guys. A 3000 watt breadboard. The holes would be 2.5 cm in diameter. Then, of course, an adapter so it could be used with an Arduino Nano.
Pretty neat. I learned a lot today.
I used to use these 20+ years ago with magnetic pulse compression to drive copper vapour lasers. the alernative was to use a thyratron.
Nice. I assume we'll be seeing that monster in a Solid State Tesla Coil in the near future?
This little transistor seems to cost about 100 EUR in Europe (listed price at Mouser, not sure if VAT is included)
IR camera can maybe see the back plate better, if you taped it over with black tape? Really would have loved to see it on your curve tracer, compared to a 3904 or something jelly-like
6:25 must be a spectacular failure mode when it blows up
Will it fail open circuit or dead short @ 200 amps
You could replace the master power switch in the breaker box in your house with that, typical house master breaker is rated about 200 amps 250v
Useful if you want to install solar power privately and not back feed it to the power company
Add a current sense and esp32 and a voltmeter chip
Get the voltage stats, current of every circuit in the house, and when the current sense transformer see a mains power outage switch dat igbt off and switch igbt on for solar
And can be operated over Bluetooth on your smartphone
And that way the power company can back feed and trickle charge your solar backup, and solar panels only are used when mains don't work, power company ain't going to put solar panels on my house and claim partial ownership of my private property that's how those solar power subsidies technically work
They pay for the solar panels, but you don't own it
The vfd drives these things are in have tons of protection in them designed to handle that very situation, as it does happen - they have very fast analog overcurrent protection that will open the main contactor immediately when a failure happens, the drive then just shows an error code until the IGBT is replaced. The failures are very mundane and uneventful actually. The only time there's a problem is when one shorts to the gate, and the gate isn't isolated anymore - then it also blows up the gate drive module(s) as well, and makes a mess on the control board.
@@Clancydaenlightened These are BJTs though on the high power side - they don't like to have the polarity of the emitter and collector reversed - most have an intrinsic flyback diode across E and C - they'd blow up immediately if you put AC across them
@@gorak9000I know how a transistor works
I know the difference between a MOSFET, a bjt, an igbt, and a JFET
couldn't resist! just press attenuate.. a real buzz! yup! agreed! IGBT's are purdy cool! well, except when they'r burnt up! a dip in the pool! good luck mr. fix it ! hand gestures are spot on!
0:21 expensive igbt too
A dream transistor for a Tesla coil full bridge if you can afford 4 of em
They can run higher current too if pulsed via an interrupter circuit
With microcontroller you can spit out musical square waves, or even use an 80s soundchip
If you got a beefy circuit you can run in continuous wave and play hifi quality audio, plasma speaker has better frequency response, more dynamic range, and is truly omnidirectional
And even run it at 250v mains
You can reach close to that 1200v in a Tesla coil primary with inductive spikes after rectification you have about 400 volts or so, and inductive kicks can reach up to 700-800 volts
You can get similar half bridge (2 IGBTs) modules new for about 20 if you look around - 40 for 2 of them (4 igbts) isn't that expensive. There are lots of variants of these all in the same package with the same connections - they're like industrial lego
@@gorak9000The ones you want for this application are more like 80+ USD per module
You want low rds on and drains source voltage (collector emitter)
And you want lots of current and voltage capacity
And what's the switching speed and pulsed operation rating
I don't want a 700 volts at several hundred amps going thru a cheap 20 dollar transistor with crazy LC spikes
20 dollar transistor is fine for some small load under a few kw
But running a 10 ft 20kw drsstc is different
@@Clancydaenlightened The 20 dollar ones I'm talking about are Fuji Electric, made in Japan. 2MBI 200F-060 600V 200A I have a set of them in a drive for a medium sized (7.5kw) CNC lathe spindle. I was shocked at how cheap they were. I'm not sure if they are the original part that was used in the drive, or someone before me replaced them. When I got the machine, the spindle drive had blown up, and someone replaced some of the parts, but they didn't find all of the issues and couldn't figure it out. I found the remaining issues and replaced the parts, and now it's fully functional again
Peak 240,000 Watts... 240KW... almost 1/4 Megawatt!!! -- I had to think that out several times but that's right wow.. That can handle the continuous thermal capabilities at full load w/ a heat sink or does the spec limit the total watts based on thermal constraints . Always enjoy your videos and have a happy thanksgiving if you're here in the USA
240kw is about 326HP. Some electric car have more power than 500HP.
Besides some good charging stations have 150+KW charging capacities. Without these IGBTs electric car industry wouldn't exist.
I wonder if the gates of those monsters is susceptible to static electricity like their smaller MOSFET cousins? Do IGBT's incorporate static protection for the gates?
What test jig does the curve tracer require for a device like this 🙂
What are you making or repairing with the device?
I want one !!!!!
Part of me says "I want to see the inside" but another part says "Don't destroy this monster."
There are usually about 10 transistor chips in parallel and similar number of diode chips.
Each about 10 mm square. They are placed on a ceramic PCB and connected with a few parallel thick (0.4 mm?) bond wires. It is probably the back of the PCB we see at the bottom, sometimes there is a heat spreader but I don't think there is one in this case.
The PCB is covered in a clear sticky soft potting compound for electric insulation.
A number of plastic tabs press the PCB against the heatsink when mounted.
@@larslindgren3846 Interesting. A lot more complicated than a 3904 since I assume it must have load balancing resistors and such.
@@urlkrueger The one I have looked at have no balancing resistors other than bond wires, copper traces and internal chip resistance. I think the chips are sometimes made to have a positive temperature coefficient by selecting on geometry and doping patern. Also these transistors are made for switching applications, they are probably not thermally stable in the linear region other than at very low power.
I don't think you should build a linear amplifier with them.
They are typically switched in 1 to 20 kHz with switching times well below 1 us.
Similar concerns about stability apply to a single chip since all power transistor chips internally have many parallel transistor structures on the same chip. So part of a chip can go into a thermal runaway.
@@urlkrueger Do a image search for "igbt module inside" or teardown and you will see many images of the inside of similar devices.
I maybe exaggerated the number of chips since most of these modules have two transistor functions in a push pull half bridge configuration, not one as in this example.
Is the IGBT just good for switching, or would it make an audio amp, or RF amp?
I’ve seen them used for an RF heater
They're too slow for RF. They should work at a few khz, but they tend to be slow, and you need pretty beefy transistors to drive the gates fast to avoid a lot of switch loss.
mmm... you could drive some hellish big loud speakers with a pair of those!
You could probably turn it up to 11.
@@noggin73 probably goes to 110!
Open it up!!!
Yes, please!
They're 100% potted. I'm also curious to see the inside of one. I actually have a dead one that I replaced - I tried to open it, I ended up stabbing myself in the thumb with either a screwdriver or a knife, and haven't attempted opening it again. I suspect the way in is with lots of heat and scraping, or milling it down.
Please tell me you didn't just casually drop $100 at Mouser for a CotD video!
th-cam.com/video/_NcTRrbufyw/w-d-xo.html
I'm wondering if anyone has thought of using the BMF to charge a capacitor or a battery. If this gismo switches in 10s khz with such a high wattage, a snubber diode will create a lot of wasted power and heat
I G B T CURRENT GROUND BATTERY TERMINAL VIP POWER VOLTAGES I CURRENT QURYNT AMPS BATTERY DC..AC ONLY HZ 120V LIVE..ONE WIRE NO AMPS WATTS.NO CIRCUIT .AC..DC DISCHARGED HAS LOAD...
AC NO LOAD PUSH JLQY PULLED JRQN
I used to work at a UK manufacturer of IGBTs. 3.3kV 1.8kA was standard. When they failed at test, they failed hard... BANG!!!🫨😵💫
I left the gate drive header disconnected by accident once, and the stray EMI fields of power-up managed to charge all of the gates at once causing the array to crowbar and explode all of them into high velocity shrapnel that continued to rain down over several seconds.
Pretty much the only thing left of the transistors were the heavy metal base plates with a gnarly crater in the middle of each of the bonded transistor dies.