Prius IGBTs for a Coilgun

[13]

Hi guys,
Recently a friend and I took apart a Prius inverter module for the IGBTs that drive motors.  As much as we have succeeded in getting the parts we have not succeeded in getting them to work.  In fact, they seem to be backwards!  We hooked up a negative lead to the possible gate and the IGBT turns on and when we hooked up a positive lead the thing turned off.  We would really like to use the IGBTs in parallel to maximize the current capability (we need around 4kamps).  Does anyone know how to work these?  See the picture attached.

[13]

Here is the Photo

[johansen]

almost looks like there is a gate driver on the igbt itself.

have you traced the rest of the circuitry?

[johansen]

still looks fishy. are 3 out of those 5 pins connected to the emitter?

[rob77]

why such a beast for a coilgun ? are you building it single stage ? single stage coil gun will be very inefficient.

[T3sl4co1l]

OBTW, should merge thread with this one https://www.eevblog.com/forum/beginners/weird-igbt-please-help/msg1010764/#msg1010764

[13]

still looks fishy. are 3 out of those 5 pins connected to the emitter?

I don't know, but I would say putting a gate driver does not make ANY sense at all.
High voltage high current devices almost all (N type only) use N substrate connected to drain/collector, which means unless they used triple well with some clever trick, it is not possible to build a gate driver on the same die, and even if they do so, they do not have enough driving capability due to there is no vertical structure. From the photo, it does not look like it has enough space for a lateral gate driver output pair (not all the outer space can be used. Keep in mind the field rings on the perimeter).
Second, what do they get from building a gate driver on die? Faster switching? No, external caps will make things slow. Higher immunity? No, Kelvin connection does the same thing.
So, my conclusion is, the 5 pins should be all gate driving signal. Be it 5 gates, no emitter, or a combination of N gates and 5-N emitters remains unknown to me.

PS. From Google search, I am able to find exact module with its supporting board, which has separate gate drivers. See link here: http://www.eetimes.com/document.asp?doc_id=1281167

Thanks for all your replies! I looked around the board and found some sm packages that look like the could be gate drivers.  However, toyota decided to use custom part numbers since I could not find the parts.  I attached some more pictures I think one of the leads is for the emitter and another for the kelvin lead (as it's labeled?). As of right now I am trying to build a two stage coil gun with the four 450v 12,000?F capacitors (a bit over kill) so using an IGBT to turn of the caps long pulse would be great.  Does anyone know why the IGBT seems to work when a negative voltage is applied?
P.S. I also have two more bricks of IGBTs from a different prius.
Thanks,
13

[rfeecs]

There is a detailed look at the inverter module here:
http://techno-fandom.org/~hobbit/cars/ginv/index.html

This page has a lot of discussion of the IGBT.  It looks like the OP's picture was lifted right from this page:
http://techno-fandom.org/~hobbit/cars/ginv/i2compon.html

[13]

There is a detailed look at the inverter module here:
http://techno-fandom.org/~hobbit/cars/ginv/index.html

This page has a lot of discussion of the IGBT.  It looks like the OP's picture was lifted right from this page:
http://techno-fandom.org/~hobbit/cars/ginv/i2compon.html

Thanks!
I'll try and get the brick to function in my coil gun circuit and post it when I do.

[13]

Hi,
I had some luck with getting the IGBTs to go but they seem to be losing around 5 volts from start to finish.  I set my power supply to 18v and tried to pass it through the IGBT.  I know that I will get some loss, but 5v seems way too much for an IGBT.  Should I be using a higher voltage to test a large IGBT like the one I have?

Thanks,
13

[zl2wrw]

Hi,
I had some luck with getting the IGBTs to go but they seem to be losing around 5 volts from start to finish.  I set my power supply to 18v and tried to pass it through the IGBT.  I know that I will get some loss, but 5v seems way too much for an IGBT.  Should I be using a higher voltage to test a large IGBT like the one I have?

Thanks,
13

What are you using for a gate driver?
To get Vce(on) = 5V I'd expect you'd have to have a high Ic (unless you were not turning the IGBT on properly in the first place)

If you only put say +5V between the gate and emitter, you won't turn the IGBT on properly (these are NOT logic level devices). A real gate driver will use something like +15V on the gate to turn on and -15V to turn off.

[johansen]

http://techno-fandom.org/~hobbit/cars/ginv/IGBT-flipped.jpg

http://techno-fandom.org/~hobbit/cars/ginv/i2compon.html

it appears to be a sense igbt.. the additional 2 leads are a diode for temperature sensing.


   

I believe we may be looking at the world's largest SENSFET !

   Motorola's tradename for the device which is able to indicate the
   current that's flowing in it. They are used for motor control when
   current sensing resistors become impractical due to dissipation and
   Ldi/dt issues.  An extra mosfet transistor is formed on the same die
   as the main device, thus it models the exact same current as the main
   device except that it has an die area 1/10000 the size and takes that
   fraction of main current.  A resistor of 200 ohms is connected from
   the source terminal S of this transistor which is brought out.  Also
   brought out is the Kelvin terminal, which I think is terminal E and
   goes to the emitters of the bipolar output transistor right on the
   die so that lead resistance and the accompanying offset voltage are
   avoided.  These two terminals usually feed into a differential
   amplifier.  The MC34129 current mode PWM chip interfaces quite well
   to these devices so you might find a tech bulletin on this chip to
   explain it more fully.

   Sensfets have five legs, this has seven. I think K and A are a diode
   being used as a die temperature sensing device.