I am using an ADUM3221 to drive the gate of a large MOSFET (an IXTN600N04T2) at about 500 Hz. It appears to work but the ADUM3221 is getting very hot. What am I doing wrong?
The ADuM3221 isolated gate driver is designed to be used with an external series gate resistor. This resistor helps to reduce ringing at the gate, tune drive strength, and also spread the power dissipation of the gate drive action between the resistor and the gate driver. Are you using an external series gate resistor? If so, how large is it? The IXTN600N04T2 comes with an internal 1.32 Ω resistor in the gate, so this is a good start.
The ADuM3221 has two output channels. Both of these are referenced to the GND2 node, and as such the sources of any MOSFETs being driven should be at the same potential.
Are you driving both outputs to different MOSFETs? If the outputs are being combined to drive one load, minor differences in the output propagation delays can lead to shoot through. This can be dealt with using specific topologies of external series gate resistance, but is generally not recommended.
The IXTN600N04T2 is a large MOSFET (as you mentioned). The approximate power to switch the device would be:
Qtot * V * Fs
I'll assume you are driving the MOSFET to 10 V.
560 nC * 10 V * 500 = 2.3 mW
This is pretty low dissipation, thanks to the low switching frequency. This number gets shared between the internal and external resistances in the charge/discharge path, so the IC sees even less than this. I don't expect much self heating in the ADuM3221 due to this.
Dear Mr Schnell,
Many thanks for your quick response. Yes you were spot on: I was not using gate resistors and one of the two MOSFETS was damaged and had a low resistance path to ground. I was also using a long cable between the gate driver chip and the MOSFET. I have now fixed all three issues. I have no idea of the best value of gate resistor – I am using the MOSFETs to control voltage using PWM so I am trying 330R which should keep the current down to 36 mA or so even with a dead short. Hopefully this won’t reduce the efficiency of the MOSFET too far
Thank you again,
I'm glad you found the issue. For the ADuM3221, we often use values of around 5 to 10 Ω total series resistance for fast turn on/off. The slower your MOSFET gate turn on, the higher the losses at the power device will be. At 500 Hz, you might be able to get away with a slower turn on/off like the 330 Ω you are thinking about, but lowering the resistance will help a lot with efficiency.
It sounds like you know this already, but I'm adding this for anyone else reading this thread:
You can measure the power lost per switching transition by measuring the Vds and Ids of the power MOSFET and multiplying the two waveforms together. There will be a power triangle that comes out during switching transitions. The slower the rise/fall, the larger the area of the triangle. Faster edges are obtained with lower gate resistances (but it also means higher power dissipation within the gate driver IC). This figure shows an idealized IGBT, but for a MOSFET it looks similar. Ic = Ids, and Vce = Vds.
Sorry to bother you again.
I am still having problems. I replaced the defective MOSFET and the gate driver and it ran a couple of times and then the ADUM3221 failed and was overheating even when the gate was unconnected.
Digging into it a little more, the ADUM3221 shows a maximum gate charge around 50 nC at 10 volts (page 9, Fig 8 in the datasheet) and (as you point out) Qtot for the MOSFET is around 590 nC at 10V. Do I simply need to find a heftier gate driver? Or am I misunderstanding something (my experience using PWM with MOSFETS is minimal)
Fig 8 in the datasheet is a recommended max load due to self heating of the driver. The external resistor is specified because power gets shared between the internal and external resistances. If you increase the external resistor, the max allowable load goes up.
The ADuM3221 definitely shouldn't be failing unloaded at 500 Hz. Do you have any parts of the schematic that you can share? One thing to consider when using the ADuM3221 is that the two outputs have to be reference to the same ground, you can't run a half-bridge with a single ADuM3221.
Decoupling is also very important. The driver experiences very short, but very high peaks of current on edge transitions. If the decoupling capacitors are too small, or too far from the part, the gate driver will see quick dips in the voltage, which can affect its internal logic.
Many thanks again for your help. I messed up on the decoupling capacitors: I missed out the 10uF on the output side. I’ll try that. Circuit diagram like this:
What I surmise might have happened is:
1.. Control works correctly for a short time
2.. Lack of decoupling capacitor stresses ADUM3221
3.. ADUM3221 fails
4.. Overheating subsequently a symptom of the failure
To confirm – both outputs share a common ground