MOSFET Cross Conduction in Push-Pull Circuits
MOSFETs are used as high-frequency switches that can conduct high currents in many circuits, and they are imperfect devices that experience some losses. The majority of losses result from conduction, which then leads to overheating and failure of the part. Overheating typically causes a failure to short circuit, and when a MOSFET continues to be driven in its short circuit state the component will eventually burn up.
When pushed to their limit in a push-pull circuit, or in an equivalent half-bridge circuit, the two MOSFETs could be briefly conducting at the same instant. Due to mis-commutation and excessively long turn-on/turn-off times, the problem of cross-conduction can arise and lead to system failure. If your push-pull converter system is having problems with cross conduction, there are some simple things that can be done in the design to ensure MOSFETs will be less likely to fail.
What Causes Cross Conduction in MOSFETs?
MOSFETs in a push-pull configuration (or in a very similar half-bridge configuration) are designed to provide alternating driving, normally with a PWM signal, in order to deliver a desired amount of power to a load. This is often used in a switching regulator in order to provide high power conversion efficiency with voltage-mode control or current-mode control. It is also used in isolated DC/DC converters as the switching waveform is required to conduct power to the secondary side of the system.
Depending on the driver being used in the system, there may be some conduction losses that arise during the switching instant between the two MOSFETs. As one MOSFET is turned OFF and the other begins to be biased ON, it is possible that the two MOSFETs are ON simultaneously for a brief instant. The resulting conduction losses between the two MOSFETs increase in this condition as they conduct into each other.
Push-pull converter circuit. (Image source: MDPI)
The fact that the two MOSFETs are briefly biased ON during the same instant can be seen in a measurement of the gate-source voltage or the drain-source voltage. Depending on the output capacitances in the MOSFETs (Coss), the waveforms for the two complimentary MOSFETs can be seen to overlap along their edges.
The above condition can occur:
-
When the duty cycle for the push-pull converter nears its maximum value of 50%
-
When the turn-on/turn-off times of the gates are excessively long
-
When the PWM driver’s dead time is too short
Even if the PWM waveform is a perfect step function, the MOSFET’s Coss value will set the lower limit on the turn-on/turn-off time. If the Coss value is too large, it could cause the turn-on or turn-off time to be longer than the dead time and the timing between commutating PWM pulses.
Reduce the Duty Cycle
Because cross conduction results from overlapping waveforms as shown above, the simplest solution is to decrease the duty cycle. Assuming all other parameters are held constant, the circuit will have longer timing between the commutating pulses. When the duty cycle becomes low enough, the gate will have plenty of time to modulate fully on or fully off. Note that the input voltage or transformer turns ratio may need to be adjusted to compensate for the lower duty cycle.
Reduce the PWM Frequency
The PWM frequency is also a determinant of the cross conduction losses because it will determine the time window for each MOSFET to turn off. When cross conduction occurs, the losses will be proportional to (in 1st order approximation):
Loss ∝ (switching frequency)*(switching time)
Switching time is independent of frequency, therefore reducing the frequency increases the time between turn-off transitions for the MOSFETS and thus reduces cross conduction losses.
Of course, the downside is that this will change the DCM inductance limit for the switching converter to a higher value. Compensating for this would require a physically larger inductor on the output side of the push-pull circuit.
Use Different MOSFETs
Another option is to use different MOSFETs with lower Coss value. This will reduce the turn-on time to a lower value. Swapping MOSFETs also allows other parameters to be changed, such as package thermal resistance and voltage/current handling.
A Warning About Changing the Dead Time
Some push-pull PWM drivers intentionally introduce dead time between pulses (10-100 ns) in order to slightly delay pulses with respect to each other. The intention is to allow a gate to more fully and achieve better commutation, ideally limiting cross conduction losses. There is some conduction loss during the dead time, but this can be preferable to cross conduction.
If the gate pulses are overlapping, one might want to swap for a driver with longer dead time, particularly when the PWM duty cycle is near 50%. This is not always the best move because a longer dead time will still carry more conduction losses, albeit less than the cross-conduction loss. Instead, implement a lower duty cycle and adjust the input voltage range to keep the duty cycle within a low range.
Summary
In summary, if the MOSFETs in a push-pull circuit are being damaged as the driving PWM duty cycles reaches 50%, it’s possible that the gate driving pulses in the complementary MOSFETs are overlapping and producing cross conduction losses. Some options for reducing losses include:
-
Decrease the switching frequency of the PWM drive signal
-
Use MOSFETs with smaller Coss
-
Use a PWM driver with some dead time
-
Design the push-pull circuit to use limited duty cycle
Cross conduction is easy to diagnose in a test circuit through direct measurement with an oscilloscope, but it can be tough to diagnose in simulation without accurate MOSFET models. If you plan to simulate, make sure your MOSFET models include the relevant parasitics so that gate switching can be appropriately modeled.
Power circuits can be challenging to analyze from a reliability perspective, but you can fully qualify push-pull circuit behavior and diagnose cross conduction with the complete set of circuit simulation features in PSpice from Cadence. PSpice users can access a powerful SPICE simulator as well as specialty design capabilities like model creation, graphing and analysis tools, and much more.
Subscribe to our newsletter for the latest updates. If you’re looking to learn more about how Cadence has the solution for you, talk to our team of experts.