How to drive an IGBT with an optocoupler?

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
What's the proper schematic for driving the IGBT (IRG4BC20UD) from a microcontroller (MSP430-g2553) pin through the optocoupler (FOD3148)? I am going to design a full bridge inverter to drive 9W CFLs (220V/50Hz)? PLEASE HELP!
 

ronv

Joined Nov 12, 2008
3,770
I'm not sure what you are asking exactly, but I'll start something.
You need 4 of the optos. One for each IGBT.
Choose the resistor for the diode in the opto for 10 ma. Be careful, your micro can only drive 48 ma total at 3 volts. I'm thinking your supply is 3 volts?
Add the gate resistor suggested in the IGBT datasheet.
Use 12 volts to power the opto.
Be careful to provide some dead time between turn on and turn off of the H.
 

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
Hello, Thanks a lot for the response. Actually I am trying to develop a full bridge inverter. For the inverter, I need replicate the below circuit for each IGBT. But, my circuit is giving me no output, I believe there is something or everything wrong with the connections. Please help me with your expert opinion.20150116_213633_Richtone(HDR).jpg
 

JohnInTX

Joined Jun 26, 2012
4,667
You should wire the LED with a single series resistor - I get about 90 ohms at 3Vdd and 10ma - so remove the 100 ohm nearest the uC. A '1' on the output port should source about 10ma at 2.7V. 2.7Vout-1.8Vfmax = .9V .9/10ma=90 ohms series resistance.

The NOTES part of the data sheet says to connect pins 1 and 4 to LED common (Item 9) but the pinout says NC - go figure.

You show no load on the IGBT so when it turns on, it will smoke maybe?

ronv pretty much sums up the rest.

Good Luck.
 

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
You should wire the LED with a single series resistor - I get about 90 ohms at 3Vdd and 10ma - so remove the 100 ohm nearest the uC. A '1' on the output port should source about 10ma at 2.7V. 2.7Vout-1.8Vfmax = .9V .9/10ma=90 ohms series resistance.

The NOTES part of the data sheet says to connect pins 1 and 4 to LED common (Item 9) but the pinout says NC - go figure.

You show no load on the IGBT so when it turns on, it will smoke maybe?

ronv pretty much sums up the rest.

Good Luck.
Hi John,

Thanks a lot for your response. I have updated the circuit according to your suggestions but I am not sure I understood completely. Can you kindly tell me how did you come up with the 10mA/2.7V at the output port? I find it very hard to understand datasheets. Please help! I also don't understand the datasheet note about pin 1 and 4 need to be connected to the led common as it shows no connection in the test circuit.

20150118_135920_Richtone(HDR).jpg
 

JohnInTX

Joined Jun 26, 2012
4,667
Can you kindly tell me how did you come up with the 10mA/2.7V at the output port? I find it very hard to understand datasheets. Please help! I also don't understand the datasheet note about pin 1 and 4 need to be connected to the led common as it shows no connection in the test circuit.
Sure.
First, I took ronv's advice about using 10ma for the LED. From the Electrical-Optical Characteristics table in the opto datasheet, it requires no more than 7.5ma to guarantee turn on. I picked the 10ma because that is where VforwardLED is spec'd. So 10ma it is. VforwardLED at 10ma is at most 1.8V so I used 1.8Vf and 10ma for the LED.
Next, I looked at the MSP430G2x13 datasheet from to see what the port pin would drive. Fig 9 indicates that with 3Vdd at 10ma out, one can expect about 2.7V at the pin. If the pin sources 2.7V and Vf=1.8V then the voltage across the R is 2.7V-1.8V = 0.9V. At 10ma this means R= .9/10ma=90 ohms.

Note that these values are picked to ensure that it works. You could reduce the current to 7.5ma for a lower total IO load on the processor - probably something to consider. The first goal though is to get it working, then tune it. It would be a waste of time if you picked typical values then spent a lot of time finding that the particular part you are using isn't 'typical'.

Regarding pins 1 and 4 - I don't know why the discrepancy between Note 9 and the rest of the datasheet. Since all of the test circuits AND the pin descriptions indicate No Connection I would leave them open for now. Connecting 1 and 4 to LED common may be for better noise immunity. On things like this, I usually just go with the most likely (NC on pins 1 and 4 in this case) BUT before finalizing the design I'll call or email a Field Applications Engineer or factory tech support to resolve the discrepancy.

The load goes in the center bar of the H made by the 4 IGBT's.
Agreed. To test one channel (not the H bridge) don't connect HV where you have it. Connect it to what you called Vo on the drawing. Vo then becomes the collector of the IGBT. Turn on the LED and the IGBT should turn on, pulling Vo to ground.

That's pretty much it.
 
Last edited:

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
Thanks a lot John. After your reply I have updated the circuit in breadboard but unfortunately my IGBT smoked out as you mentioned before, I tried to load the IGBT properly which obviously didn't work out. I am really confused why I couldn't go any more than 13 volts for the IGBT collector voltage wheres the maximum voltage the power supply could provide is 25 volts. Please help me with your expert advice.

unnamed.jpg
 

JohnInTX

Joined Jun 26, 2012
4,667
A couple of things.
The NMK0515SC is rated for 15V at 250ma. The drawing in post #5 shows pins 3 and 4 reversed (hooked up backwards). Is the power supply for the IGBT separate?
Are you sure everything is hooked up according to the drawing? 13V through a 1K resistor is only 13ma, even if the IGBT is shorted/backwards so something is wrong there.
 
Last edited:

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
A couple of things.
The NMK0515SC is rated for 15V at 250ma. The drawing in post #5 shows pins 3 and 4 reversed (hooked up backwards). Is the power supply for the IGBT separate?
Are you sure everything is hooked up according to the drawing? 13V through a 1K resistor is only 13ma, even if the IGBT is shorted/backwards so something is wrong there.
Hi John,

Thanks again! I showed it the wrong way in the drawing before but while hooking it up I did it correctly. Actually I changed my dc-dc converter to (NME0515SC ) which is not giving me 15 v, I am thinking this is because the datasheet has an inductor order part and maybe I am not loading it in it's 10% of the rated load which is a must to get it working. I am also not getting anything out from the output of the optocoupler. I think as I am giving around 2.7 v in the input of the optocoupler at pin 2 from the MSP430, do I need to use a BJT to supply the base and get it working? I found the IGBT/MOSFET Gate Drive Optocoupler that they have a transistor in the input. I have checked my connections and they seem to be ok to me now but still not getting anything. Would you suggest me to change the optocoupler? I am struggling to get it working. I really appreciate your help and time.

3Untitled.png
Tahsina
 

JohnInTX

Joined Jun 26, 2012
4,667
The datasheet for the DC DC converter is a bit fuzzy on the 10% requirement. It does not specifically say (that I can find) that the min output current must be 10% of full load. It does use 10% as a minimum load for the various output specifications. I would therefore load it to 10% minimum for your tests. How about an LED running at 7ma as a power on indicator for now?

The extra LC on the output is not necessary. You can add them to reduce ripple but don't worry about that for now.

As far as the opto, why not test it by removing the IGBT stuff and just rigging another LED to the output? Use Fig 22 in the opto datasheet as a reference (for the output side) and put an LED and resistor to pull 20ma or so from pins 6+7 to ground, anode towards ground. When your processor turns on the opto, the LED should light, troubleshoot from there.

Good luck.
 

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
The datasheet for the DC DC converter is a bit fuzzy on the 10% requirement. It does not specifically say (that I can find) that the min output current must be 10% of full load. It does use 10% as a minimum load for the various output specifications. I would therefore load it to 10% minimum for your tests. How about an LED running at 7ma as a power on indicator for now?

The extra LC on the output is not necessary. You can add them to reduce ripple but don't worry about that for now.

As far as the opto, why not test it by removing the IGBT stuff and just rigging another LED to the output? Use Fig 22 in the opto datasheet as a reference (for the output side) and put an LED and resistor to pull 20ma or so from pins 6+7 to ground, anode towards ground. When your processor turns on the opto, the LED should light, troubleshoot from there.

Good luck.
Hi John,

I re-implemented the circuit according to your guidelines. The dc-dc is working fine and producing 15V accordingly. What I found is the MSP430 pin is producing around 1.8V. Earlier you told and I also found in datasheet that the MSP430 should produce 2.7V at output pin but I am getting 1.8V. Any observation on that? I am producing a spwm signal from the pin 1.2 of chip 2553.

Thanks a lot!

Tahsina
 

ronv

Joined Nov 12, 2008
3,770
Are you sure you have the 90 ohm resistor in series with the diode?
Check that pin at the output of the mico with the resistor disconnected.
 

JohnInTX

Joined Jun 26, 2012
4,667
Any observation on that? I am producing a spwm signal from the pin 1.2 of chip 2553.
What are you measuring the voltage with? Be sure you are measuring from the pin to ground, not across the LED. Be aware that if you are driving the output as a PWM, the voltage measured with a DVM will be roughly the expected output voltage * the duty cycle i.e. 2.7V 50% duty cycle gives .5*2.7V=1.35V measured. Other than that, I don't know.
 

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
Are you sure you have the 90 ohm resistor in series with the diode?
Check that pin at the output of the mico with the resistor disconnected.
Hi,

I have tried the way you suggested but could not get it working. However, I will try few other things tomorrow to make it work. Thanks a lot.
 

Thread Starter

Tahsina Hossain

Joined Jan 16, 2015
50
Hi guys,
I finally got my dc-dc and opto working together (circuit 2 in the image) but my IGBT is not switching (circuit 1 in the image). any observations?

allabout.jpg
 
Top