EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Bancroftjo on December 26, 2015, 09:08:26 pm
-
I am investigating gate drivers for future projects I want to make. I bought an IR2117 high side gate driver and have been bread-boarding it but am getting unexpected results. I have attached a photo of the schematic and screenshots from my oscilloscope. But basically my setup can be described as follows.
I am feeding the logic side with a 12Vdc supply and am providing a 0-10V square wave to the IN pin for testing. I have a N type IGBT on the high side with 50Vdc on the collector and a 220 ohm resistor from the emitter to ground. There is a 33 ohm series gate resistor. I have bypass capacitors on the Vcc pin and I have a 0.1u capacitor between Vb and Vs. I am using a 1N4001 diode between Vcc and Vb. I believe I have followed the suggested schematic in the datasheet.
My expected result from this experiment was to see Vg follow the signal on Vin. The square wave I am applying from a signal generator goes from 0V to 10v and has a 20ms period and is 50% duty cycle. The results I am seeing is the gate being sent high (see attachments) for 5.95ms then dropping back down to ground, I was expecting it to go for the full 10ms.
Am I missing something, or is this correct behavior for this type of gate driver? I have only ever implemented discrete gate drivers and have never used a dedicated IC. My end goal is to use gate drivers to control N type devices on the high side of a 150Vdc, 50A, 10kHz circuit.
I would like to know if my schematic is correct, and if so is this correct behavior for this gate driver. I would also appreciate any functional description of these gate drivers or any other recommendations.
The oscilloscope screenshots show the gate voltage versus the signal generator voltage, and another shows the gate voltage versus the emitter voltage. The schematic is drawn as I have the circuit bread-boarded.
IR2117
http://www.irf.com/product-info/datasheets/data/ir2117.pdf (http://www.irf.com/product-info/datasheets/data/ir2117.pdf)
FGA180N33
http://media.digikey.com/pdf/Data%20Sheets/Fairchild%20PDFs/FGA180N33AT.pdf (http://media.digikey.com/pdf/Data%20Sheets/Fairchild%20PDFs/FGA180N33AT.pdf)
Thank You
Jon
-
While the waveform is not what I expect to see the bootstrap voltage is very low. Cap should be much larger, that diode is slow, and the 200 ohm load resistance may keep it from charging fully. The chip is likely going into under voltage shutdown.
-
I've never seen a bootstrape gate drive without a low side switch. In fact, I can't really see it working without out.
You need to make a floating high side supply.
-
Assuming the off-time is sufficient to recharge the capacitor (i.e., tau = 0.1uF * 220 ohms = 22us << t_off), then the problem is probably supply current.
Datasheet page 3:
I_QBS = 50uA typ, 240uA max
V_BSUV- = 8.2V typ
Your diode drops about 0.6V, so Vbs = 11.4 at the start of the pulse. The gate switching event takes some charge, 170nC typ, which drops the capacitor voltage by V = Q/C = 0.17 / 0.1 = 1.7V, taking it down to 9.7V.
It's noteworthy that, for most applications where you'd actually be stressing that monster IGBT, you'd need about an order of magnitude stronger gate drive (i.e., several A peak output) and a proportionally smaller gate resistor (like, 1 ohm).
The driver output turns itself off below about 8.2V, so in the 1.5V remaining, which took about 6ms, we should suppose the supply current draw was actually I = C * dV/dt = 0.1uF * 1.5V / 0.006s = 25uA. That's in the ballpark of 50uA, anyway.
TLDR: your IGBT is fuckoff too large, and your pulse duration is too long. Run it at >10kHz and it'll be fine, or use a bigger cap. (Oh, and don't use 1N4001, you'll break something. Use UF4004 or something fast like that instead.)
Tim
-
I implemented the changes you all suggested today and had much better luck!
I changed the diode to a 1N4148, which didn't have much effect except sharpening the rising and falling edges at the full 10kHz.
I then changed the bootstrap capacitor to 1uF and accomplished the full 10ms pulse length, then tried with 100% duty cycle and it collapsed after 88ms, much longer than we'd ever need.
Additionally I upped Vcc to +15V and was able to maintain the gate drive for 150ms. I also tried a FDH333 diode and saw minimal difference, except it rounded off the rising and falling edges a little.
In my final circuit I settled for the 1uF bootstrap capacitor, a 1N4148 diode, and +12V logic supply. I also measured Vce(on) to be 0.8V
I switched the gate resistor to 10 Ohms and saw faster rising and falling edges.
Thank you for all the advice! And for those who asked, no we will not be using this IGBT in our final design it is just what I had on hand yesterday, we are looking at one with 94nC total gate charge instead of the 170nC of the one in this schematic.
Thank You
Jon