Burning (turned to a wire) IGBTs in a three phase inverter b

EDAboard.com Forum Index -> Power Electronics -> Burning (turned to a wire) IGBTs in a three phase inverter b

Author

Message

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

19 Jun 2009 23:56 how to prevent igbt from failure

Hi.

I'm building a three phase inverter, using IGBTs for power swithches. To drive the gates Im using the IR2110, I have snubber capacitors, IGBTs are rated for 600v, the system voltage is 300v (3 ph 220 rectified), I have 18v zener diodes to clamp Vge. All the components are near and traces are small. Very detailed info (schematics + signals) of the system is attached in the file "system seting.pdf".

The bridge works nicely under 180v (I used 3 BK power supplies to get this voltage), but when I got to the rectified 300 VDC always a branch of IGBT burns turning them into a wire (resistance between colletor-emitter = 0). The 300Vdc source is fused at 5 amp plus 4amp fuses in each branch. The load I'm usign for the test is a 0.5hp (2.3amp) induction motor.

Please help. I'm out of ideas.

Thanks in advance.

Alex.

Sorry, but you need login in to view this attachment

hock

Joined: 28 May 2001
Posts: 1291
Helped: 23
Location: India

20 Jun 2009 9:10 ir2210 schematics

check for cross conduction.
what is the peak off state voltage acorss the off state device at 100 v and at 150 V ?
are you using capacitive snubber across the switch?
hock

FvM

Joined: 22 Jan 2008
Posts: 5154
Helped: 766
Location: Bochum, Germany

20 Jun 2009 10:26 how does a 3 phase pwm igbt inverter work

The best way to understand the circuit failure would be to acquire the event with your oscilloscope!

I don't see a particular problem in the circuit, but I don't understand your drive scheme. As far as I understand, your not operating high and low side switches of a half bridge in the usual PWM way. As a consequence, the bootstrap voltage is continously dropping during the positive half wave. That's not the way, IR2210 is intended to operate. Hopefully, the undervoltage lockout should prevent from IGBT damage by desaturation, but it's difficult to predict the circuit behaviour in this situation.

I also would prefer higher resolution waveforms to see the switching behaviour. Apart from unsuitable control timing, overvoltages from excessive circuit inductances must be expected causing the failure. Waveforms and a drawing of circuit topology or a photo can give an insight.

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

20 Jun 2009 15:47 ir2210 aplication note

Hi

Thanks for the rapid response.

hock:
Im using a 0.15uF capacitive snubbers in each branch. I measured the voltage in one of this capacitors and the peak voltage is 125v (last figure in the PDF), I think this peak is caused at the switch off.

FvM:
Im sorry, I forgot the control signal explanation. In the attached image I show what signal I'm sending to IGBTs. The PWM is 10khz and 75%.

Well about the high side foating supply going to under voltage, I had probed the control signal form the MCU and the ouput of the IR2110, and I did not see the high side gate voltage to drop under 11v with no load. (The condition I'm testing is the worst case). The algorithm charges the bootstrap capacitors at the begining.
In the first figure of page 7 the high side gate singnal appears to drop and conduct at the same time of the low channel, but when probing directly the Vge of the high side switch its working fine, this would be the second figure on page 7 in the pdf.

I've attached the pcb layout and a picture.

Monday I'll get you some pics of the 10kHz PWM, and the comparison between the high control signal and the high floating voltage.

I only have acces to 10:1 probes, can I monitor the circuit with this?

Thanks in advance

Alex

Sorry, but you need login in to view this attachment

flatulent

Joined: 19 Jul 2002
Posts: 4875
Helped: 324
Location: Middle Earth

20 Jun 2009 16:00 three-phase inverter igbt diode

Another source of additional clues can be had by using a clamp on current probe. Compare the currents at the transistors with the voltages.

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

23 Jun 2009 2:00 ir2210 layout

Hello Again.

I've captured more signals. I've added the control signals, as FvM requested. Also I made some experiments by vaying the current (duty cyce of the PWM, avarage voltage), and measured the voltage at the snubber capacitor.

I think the snubber measurments show the problem, It seems that the 0.15uF capacitors snubber are not enough.

At a 0.08A current the overvoltage was 7.5v (6% of the bus voltage 120v)
At a 0.2A current the overvoltage was 17.2v (14% of the bus voltage 120v)
At a 0.4A current the overvoltage was 25v (20% of the bus voltage 120v)
At a 1.3A current the overvoltage was 54v (45% of the bus voltage 120v)

And so on. With this trend the 300v tests would generate really high over voltages, because the current increases by more tha 2.

Do you think this could be the problem?

What kind of snubber circuit do you recomend to minimize the over voltages?

Thanks in advance.

Alex

Sorry, but you need login in to view this attachment

FvM

Joined: 22 Jan 2008
Posts: 5154
Helped: 766
Location: Bochum, Germany

23 Jun 2009 6:47 3 phase inverter bridge pic

Quote:

Do you think this could be the problem?
What kind of snubber circuit do you recomend to minimize the over voltages?

It's at least a sufficient explanation for catastrophic failure, as observed. There may be other problems, too. I fear, that the bus wiring involves too much leak inductance.

You're talking about snubber capacitors, but I don't see a bus capacitor in the circuit. Do you have any?

I think, it's difficult to measure voltages, e.g. an actual Vce of an IGBT in a mains supplied circuit without a differential probe. Thus it's hard to determine, if the shown waveforms are "real".

Google
AdSense
Google Adsense

23 Jun 2009 6:47 Ads

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

23 Jun 2009 14:02 igbt switch scheme in three phase inverter

Hi FVM

I have only two little bus capacitors. One 0.1uF at the output of the power supply, then a couple of 14 gauge 15cm long cables go to the inverter board, where another 0.1uF capacitor is placed between the bus terminals.

Do I need to put more capacitance to the power supply? (1mF or more)

Thanks once again.

Alex

GetDeviceInfo

Joined: 30 Mar 2009
Posts: 103

05 Jul 2009 5:24 b phase invertor

Just thinking out loud;

My experience has been with negative voltages for turning 'off' the gate.

Turn off times can exceed dead times which result in 'shoot thru', which may well be the condition your experiencing.

Again in my experience, asymmetrical gate control typically found the diode conducting along with Rg 'on', leaving Rg 'off' sunk to the negative voltage.

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

05 Jul 2009 22:51 dead time inverter igbt

Thanks GetDeviceInfo.

I think my problem was that I did not have bus capacitors. I added two 820uF (1.6mF total) capacitor to the DC BUS, and the voltage spikes and rising stopped. The measurements I used to get to this conclusion are in the following images.
The test was made with two BK power supplies at 100vdc. In the first figure yo can see how the dc bus voltage is above 120v. By adding the capacitors the bus voltage remained at 100V.

I hope this solved the problem. What do you think.

Thanks

Alex

Sorry, but you need login in to view this attachment

GetDeviceInfo

Joined: 30 Mar 2009
Posts: 103

08 Jul 2009 18:34 how does an igbt three phase inverter work?

and so, how did that work out?

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

28 Jul 2009 18:15 igbt inverter bridge images

Hi Again

The capacitors solved the rising voltage problems in the DC voltage bus.

I attempted to the controller (postion PI with velocity PI in cascade), but I had many problems with the bootstrap capacitors getting discharged.

So I had changed the PWM scheme, now I'm leaving the low IGBTs on all the time. The PWM modulation is only sent to the high side mosfet. This solves the discharging of the bootstrap capacitor.

Now the problem is cross conduction. The low side IGBts are conducting when they are supposed to be open.

With the oscilloscope I've measured the Vec, when the low IGBT has 15v at the gate i have Vec=1.5v (with a little noise), but when the IGBT should be off (Vec=Vbus) some of the PWM signal can be seen going from 0v to Vbus.

What can I do to minimize cross conduction in my circuit.

Thanks in advance

Alex

FvM

Joined: 22 Jan 2008
Posts: 5154
Helped: 766
Location: Bochum, Germany

28 Jul 2009 19:20 igbt gate burning

Quote:

So I had changed the PWM scheme, now I'm leaving the low IGBTs on all the time. The PWM modulation is only sent to the high side mosfet. This solves the discharging of the bootstrap capacitor.

I don't understand, how this is supposed to work. The low side switch should be controlled by an inverted high side signal,
a dead time logic should delay the risind edge of each gate drive.

To guarantee correct operation of the boostrap circuit, a minimum on-time must be provided in pwm generation for the low-side driver.

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

29 Jul 2009 4:39 ir2210 driver burning

Thanks FvM.

I'm getting confused here with the control signals. The next image shows what control signals Im sending to the inverter bridge.

Active: means a solid 15v to the lower IGBT gate.
PWM: a 10Khz with a max 75% duty to high IGBT.
off: the IGBT should be an open circuit.

I've inserted a dead time when hall sensor state changes. All the IGBTs are sent to off state for some ns.

Something that happens with this method is, that the DC bus current consumtion increased.

Thanks Once again.

Alex

Sorry, but you need login in to view this attachment

FvM

Joined: 22 Jan 2008
Posts: 5154
Helped: 766
Location: Bochum, Germany

29 Jul 2009 12:06 igbt gate sequence 3 phase

In my opinion, a drive pattern as I suggested in this thread should be used http://www.edaboard.com/viewtopic.php?t=356065. I think, correct bootstrap function of the highside driver can't be guaranteed without switching the low side in PWM operation. As another problem, your drive scheme doesn't achieve proportional PWM action at low loads, as reported in the said thread.

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

30 Jul 2009 14:38 burning test for igbt modules

Thanks FvM.

I've made the change in the PWM scheme. Now my position PI works fine, I need some time to adjust Gains, and add an acceleration ramp. Thanks once again.

I thought the PWM scheme I was using was usefull, because I saw it in almost al the BLDC app notes, but now thanks to you I know that is not the best PWM scheme.

I was wondering:

If I make a 15v power supply using a transformer and connected to the Vs and Vb pins of the IR2110. Can I replace the bootstrap capacitors? (of course taking out the charge diode from the circuit) would this get me a continuos high side operation? would this give more problems (noise, spurious switching, cross conduction)

I think there is some cross conduction in my circuit, how can I reduce it?

Thanks in advance

Alex.

mananshah93

Joined: 01 Aug 2009
Posts: 3
Location: India

03 Aug 2009 5:42 igbt damage

Hi! to all....
I am also designing the same thing using 6 IGBT and 3 IR2110 drivers...
I found this post while surfing ...

I have some queries...
1. PWM sequence...how you have decided pwm sequence...I saw there hall-sensor is used in the project....can u explain me if hall-sensor is neccessary for this???
And algorithm for pwm sequence......

2. I want to use RCD snubber circuit like shown in the attached file...but I don't know how to calcullate R,C and D....my details for the supply and motor are----Vbus=230V ... R=5 ohm between two windings...
now I have checked the circuit with low voltages (~15V)... so I don't know other parameters...can u help me out??

Sorry, but you need login in to view this attachment

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

04 Aug 2009 22:15 igbt dead time

Hi mananshah

I'm attaching a document that has many formulas for snubber circuit calculation.

About PWM sequence I was using the sequence found in all microchip application notes. No I've changed thanks to FvM, this PWM scheme gives you a better torque at low speeds. I'm using hall sensors because I want to control the rotor position. If you only need to control speed you can make sensorless control, (well you still need ADC capability).

Hope this helps.

I have a question:
the 230v bus voltage, is the rated motor voltage?
does the rated motor voltage is RMS or DC?

Thanks

Alex

Sorry, but you need login in to view this attachment

mananshah93

Joined: 01 Aug 2009
Posts: 3
Location: India

05 Aug 2009 5:37 snubber circuit for igbt inverter

Hi! Alex
the voltage which is going to highside igbt is 230-240 DC coming from AC 230V and rectified to produce DC.

Actually, I do have centrifuge 5810 from eppendorf, in which this type of driver is used, but it has gone bad...now I am trying to make my own...
But I don't know the whether the motor is used is AC Induction or Ac synchronous or 3-phase brushless DC motor... the motor is covered , I can only see the 3 wires coming from inside...no details are there ...how to know which motor is there? I have checked the resistance between each winding is 5 ohm......

So, I have measured the voltage at collector og highside IGBT is 230-240V DC...

I have other question tooo..
I want to solve this mystery for moving ahead...please tell me(if u know), is there any difference when you give pwm sequence for your igbt drive to drive 3-phase AC induction motor or BLDC or Synchronous AC motor???

I am confused because I don't know the motor is which type???
I heard that 120 degree phase diff. is necessary for AC synchronous??
tell me if u know...

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

05 Aug 2009 20:34 bldc hall sensor timing advance circuit

Hi Again

I'm posting the signals of the final PWM scheme. I hope I've understanded right this time.

mananshah:
Well before doing a drive you need to kwon the target motor. It's true that the 3phase bridge topology is the same for Induction, synchronous and Brushless motors, but the control algorithm will change.

Burshless motors and synchronous motors have permanent magnets to provide a magnetic field. I you rotate your motor and measure the voltage between phases, you have a BLDC or a synchronous motor. There is little difference between the construction of theres motors. Synchronous are mainly designed to work with sinusoidal voltages at 60hz. While some BLDC motors can work in a bigger frequency range (I've seen some of 0-300hz), and some times they admit trapezoidal voltages.

If you dont get any voltage you have an induction motor. As its name indicates the magnetic field is induced in the rotor, just like voltage is induced in the secondary winding of a transformer.

Hope this helps

Alex

Sorry, but you need login in to view this attachment

alexplace2001

Joined: 17 Jun 2009
Posts: 21
Location: Mexico

05 Aug 2009 21:29 igbt dc bus inverter

Hi Again.

I'm checking some measurments I took today, from the IGBT inverter bridge. The tests where made at a 50Vdc bus voltage. With the motor as inductive load. The dead time is 8us, automatically generated by the PWM module on the dsPIC.

The Image "print_11" are the gate signals of one IGBT branch.
Yellow: low side Green: High Side.
(You can see the cursors for the green channel from 50v to 65v.)

In a red circle you can see some noise (I think) Image "print_12", has a zoom of this "noise". There you can see, that it's under 50v, so no conduction of high side IGBT should take place.

Moving on to the Vce of the low side IGBT we have the waveform of the Image "print_14".
Yellow: Vce for low side IGBT Green: Gate signal.
Well theoretically when 15v applied to the gate I should see almost zero (a couple of volts) in the Vce. This is true for sections 1 and 2, but for sections 3 and 4, something else is happening.
Image "print_15" has a zoom in of the noisy area circled in "print_14"
As you can see there are some strange wave forms there.

I suppose this is cross conduction, but im not shure.
What can I do to solve this problem?

Thanks in advance.

Alex.

Sorry, but you need login in to view this attachment

sziang

Joined: 09 Nov 2009
Posts: 1
Location: Malaysia

09 Nov 2009 3:34 Re: Burning (turned to a wire) IGBTs in a three phase invert

Hi There,

From the looks of it, it looks like back emf generated from the motor. I am not really sure thou. I did sensorless PMBLDC motor driving and that slope looks just like the back emf that I am detecting to provide positional feedback.

Sorry but didn't really help you on your problem.

Rgds
sz