Hello!

I have been trying to figure out where and how to simulate an equivalent circuit for simulation. I have no clue how to do it and I need this simulation to test the behavior of an IGBT. The simulation will help me see whether to connect 9 IGBTs in parallel to a single gate driver connected to the output PWM pin of a DSPIC or to connect a demultiplexer to the PWM output in and later connect one gate driver to each IGBT.
I would deeply appreciate any insight for this dilema. Thanks in advance.

 

I'm not sure exactly what you're asking but I use the free and popular LTspice for software simulation. Many folks here use it as well, so it's easy to pass around the simulation files to let experts show exactly the changes they recommend to you.

As with any simulation, the challenge is to have accurate descriptions of the behavior of the components of interest, and to stay within the confines of the simulation model. In other words, you can't be asking questions about heat and temperature if the model doesn't have the appropriate parameters.

 

I'm not sure exactly what you're asking but I use the free and popular LTspice for software simulation. Many folks here use it as well, so it's easy to pass around the simulation files to let experts show exactly the changes they recommend to you.

As with any simulation, the challenge is to have accurate descriptions of the behavior of the components of interest, and to stay within the confines of the simulation model. In other words, you can't be asking questions about heat and temperature if the model doesn't have the appropriate parameters.

Thanks for the insight. The thing is that I don't even know how to simulate that IGBT model in ltspice. I tried using a general model but for some reason when I try to run the simulation, an error shows up.
I found that schematic I posted about the equivalent circuit for simulation on the IGBT data sheet.
As I said, I am looking forward to connecting 9 IGBT to a DSPIC controller PWM output pin as all IGBT will open/close at the same operating switching frequency of 30 kHz. I have two vague ideas of how to do so, but I don't even know how to include the exact IGBT model for the simulation or what that equivalent circuits for simulation is and how it could help me run a simulation.
Thanks again for the response.

 

Perhaps something here may help you.

 

I will check it out. Thank you!

Perhaps something here may help you.

 

Semiconductor manufacturers provide SPICE models of some of their semiconductors, so you might check with the manufacturer of the IGBT that you plan to use.

 

Semiconductor manufacturers provide SPICE models of some of their semiconductors, so you might check with the manufacturer of the IGBT that you plan to use.

I already checked and I didn't find any. I have invested great time researching on many things without having found necessarily what I need.

 

Semiconductor manufacturers provide SPICE models of some of their semiconductors, so you might check with the manufacturer of the IGBT that you plan to use.

That is why I decided to make my question public, as up to this point, I wanted to find something different from my own conclusions.

 

See 9 IGBTs in parallel

 

See 9 IGBTs in parallel
View attachment 142628

Thanks a lot for the demonstration! I will search around for the specific igbt and gate driver model to run the simulation.
Now, would you recommend to connect all 9 IGBT in parallel to the gate driver in real life or does that sound potentially impractical?
What other mechanism could be considered to control all 9 IGBT simultaneously?

THANKS A LOT TO ALL FOR TAKING TIME TO HELP ME FIND SOLUTIONS!

 

Semiconductor manufacturers provide SPICE models of some of their semiconductors, so you might check with the manufacturer of the IGBT that you plan to use.

I checked back on the manufacturer website and I didnt find any PSPICE file for the IGBT model I am pretending to use. Either that specific model or a similar one. What can I do about it?
All PSPICE files are for discrete IGBTs whereas I need to run a simulation of a module (brick) itself.

Thanks again for you help!

 

Thanks a lot for the demonstration! I will search around for the specific igbt and gate driver model to run the simulation.
Now, would you recommend to connect all 9 IGBT in parallel to the gate driver in real life or does that sound potentially impractical?
What other mechanism could be considered to control all 9 IGBT simultaneously?

THANKS A LOT TO ALL FOR TAKING TIME TO HELP ME FIND SOLUTIONS!

I will explain my scheme a little. Used 9 transistors (q[1:9] ~ q[1]...q[9]) and nine gate resistors of 10 ohms (R2[1:9]~ R2[1]...R2[9]).
The gate resistors match the switching times, despite the scatter of the thresholds of the individual transistors. I think that in the modules and done. I used a driver with an output current of +/- 3.5A. The speed was acceptable (when running at once nine transistors). But there are drivers with a large output current.

 

RE:""Semiconductor manufacturers provide SPICE models""
As Yes as No. Its true that most of datasheets contain a Spice models, but them are in most of cases construct for P-Spice not for LT-Spice. The reason is that LT uses physical layer data to modelate (crystal sizes, ion concentration etc etc, so the model results are VERY accurate, and it is freeware. Thus, model raw data may be used by evil-ful concurrent to steel the semiconductor design. In contrast, P-Spice uses only electronic measurement data what gives no big benefit for design-thieves, but it is hardly less accurate in areas far out of the normal exploit range, and surplus to this, it is high-price software. Thus, the libraries for P is somewhat 100-fold richer as for LT, BUT in the command line style it is possible to feed in the LT those P-designed library data. So, my advice, feed it the P data if LT is unavailable and use a LT.
Be warned, LT sometimes may create a notable load on computer, the worst case scenario is half-day long hypnotic state for i-core-8 machine with 16 GB RAM. However the most cases even small laptop may do the job at few seconds. LT has a brilliant user forum at Yahoo/groups where exist some gurus always trying to help novices, and probably most the questions may be covered by just history research.

 

Carefully look at my example. I used the model for Pspice and only changed one parameter to better match the datasheet.

 

I will explain my scheme a little. Used 9 transistors (q[1:9] ~ q[1]...q[9]) and nine gate resistors of 10 ohms (R2[1:9]~ R2[1]...R2[9]).
The gate resistors match the switching times, despite the scatter of the thresholds of the individual transistors. I think that in the modules and done. I used a driver with an output current of +/- 3.5A. The speed was acceptable (when running at once nine transistors). But there are drivers with a large output current.

Thanks a lot again for the insight. In other words, it should work as the simulation proofs so.
I got one more question regarding selecting the most appropiate gate driver for the IGBT model I am planning on using: As I am looking forward to connecting 9 large IGBT modules (bricks) to be controlled simultaneously, does the size of the gate driver matter? In other words, a single IC of output current larger than 1 A should work fine for the 9 large bricks? In addition to considering temp and other specs, of course, but I am a bit worried about size.
What would it be better, to connect 9 gate drivers to each IGBT in parallel to the PWM output pin or to connect 9 IGBTs in parallel to a single gate driver?

Sorry for asking too many question, I am learning on this process a lot THANKS to you all

 

Hi
Every gate needs a resistor. Nine in all so you need to look at the gate driver resistance tolerances and work out a parallel equivalent. If it scales up the you are good to go. Use also one heatsink to keep the igbts at exactly at same temp. The latter is very important.