By Roland Saint Pierre
Director of New Product Definition at Power Integrations
Offline flyback power supplies require clamping circuitry (sometimes called buffers) on the primary side of the transformer to limit the drain voltage stress across the power MOSFET switch when it is turned off during normal operation. Different approaches can be used when designing clamp circuits. Low-cost passive networks can effectively implement voltage clamping, but the clamping energy must be dissipated at each switching cycle, which reduces efficiency. An improved approach is to use complementary-driven active clamping techniques for clamping and power switches, which improve efficiency, but they impose limitations on the operating mode of the power supply (e.g., it cannot operate in CCM mode). In order to overcome the design limitations of complementary active clamp circuits, another more advanced control technique can be employed, namely non-complementary active clamping. This technology ensures that clamping energy is used in a more cost-effective manner.
This article will briefly describe the need for primary clamping circuitry in flyback power supplies. Then compare and contrast the use of passive clamping schemes, complementary active clamping schemes, and non-complementary active clamping schemes, and finally introduce a chipset that supports non-complementary clamping schemes and enables ultra-high power density flyback power supply designs.
In a flyback converter, when the primary side switch is off, the voltage (VOR) is reflected from the secondary side to the primary side, and the stored energy is transferred to the load through the transformer (Figure 1). The VOR is amplified by the ratio of transformer turns, and the voltage stress across the switching device is increased by superimposing the VDC input bus voltage. In conventional circuits, passive primary clamping circuitry is used to limit this voltage.
Figure 1: The passive primary clamped RCD solution (highlighted section) dissipates a lot of heat, limiting the efficiency and operating frequency of the flyback power supply
In addition to the voltage stress (VIN + VOR), a large voltage overshoot is generated when the primary switch is turned off, which is caused by the energy stored in the leakage inductance of the primary winding. Clamping circuitry limits the voltage overshoot caused by these three factors to protect the primary switch (Figure 2). In addition, in this circuit configuration, the power switch is turned on at higher drain voltages. Switching losses are proportional to VDS2, so high VDS increases the switching turn-on loss, further reducing efficiency.
Figure 2: Both turn-on and clamping losses are related to switching frequency
The clamping capacitor absorbs the leakage inductance energy, but this energy is then consumed by the clamping resistor. There is an energy loss during each switching cycle, which in reality limits the increase in switching frequency. Lower switching frequencies require the use of larger transformers. Therefore, the use of passive clamps increases losses and necessitates the use of lower switching frequencies, both of which increase the size of the power supply. The use of active clamps can break through these limitations.
Complementary active clamping
The active clamp replaces the resistor in the RCD clamp with a switch, which is usually a power MOSFET (Figure 3). It is not used to dissipate the leakage inductance energy, but to transmit the leakage inductance energy back to the transformer. In the complementary active clamp, the clamp switch is turned on when the main MOSFET is off, with a short dead time between the two. At this point the clamp capacitor is charged. Before the next main MOSFET is turned on, the clamp switch is turned off and the energy in the clamp capacitor is recycled to the output. This active clamping scheme is called the complementary drive scheme because the main MOSFET and the active clamp switch work in a complementary manner.
Figure 3: Simplified schematic of a typical [complementary] active clamping scheme
Zero-voltage switching can be implemented using sophisticated adaptive control techniques to achieve resonance between the leakage inductance and the clamping capacitor. When the clamp switch is turned off, the negative current generated by the resonance of the leakage inductance and the clamp capacitor discharges the voltage across the COSS before the power MOSFET is turned on, thereby achieving zero voltage switching. For designs with high output capacitance, this will result in poor resonance (the output capacitance is reflected through the transformer to the primary, increasing the capacity of the clamp capacitor). Usually there will not be enough leakage-inducted energy storage in the transformer to accommodate this change in clamp capacity. To overcome this problem, two-stage LC filters are often required at the output of the power supply to ensure low primary reflective capacitance while also meeting output ripple requirements. This complementary active clamping scheme is an improvement over passive clamping, but the following limitations remain:
1. Pulse train mode needs to be used under light load, which results in higher output ripple
2. Two-stage output filter
3. Limited to critical conduction mode or discontinuous conduction mode (CrM and DCM); no CCM operating mode makes USB PD designs with a wide output voltage range difficult to achieve
Use non-complementary active clamps to improve performance
A non-complementary control scheme is used, instead of turning on the clamp switch immediately after a short period of time after the main MOSFET is turned off, the clamp switch is turned on briefly before the main MOSFET is turned on. Non-complementary control can operate in continuous conduction mode as well as discontinuous conduction mode (and CrM) and still enables zero-voltage switching. This allows the designed power supply to have a very wide input voltage range and a wide output voltage range, which is required to design an efficient USB PD charger. For conventional control schemes, the synchronous design of the drive signal of the non-complementary clamp switch with the primary switch and the synchronous rectifier switch is challenging. A single controller to manage switching operations for all three devices greatly simplifies the circuit and ensures reliable operation.
Figure 4: For a non-complementary mode switch, the active clamp switch switches switch only once before the main switch is turned on
Non-complementary active clamp control can be implemented using Power Integrations' Innoswitch4-CZ/ClampZero chipset (Figure 5). The InnoSwitch4-CZ device is available in an InSOP-24D package and integrates a reliable and durable PowiGaN 750V switch with a secondary controller to control the main switch, clamp switch, and synchronous MOSFET operation, while including a FluxLink control link that meets safety standards. The InnoSwitch4-CZ IC includes two pins specifically designed for ClampZero active clamping non-complementary control: the upper tube drive (HSD) pin for turning on and off The ClampZero switch, and the V pin for measuring dc bus voltage.
Figure 5: The HSD signal of the InnoSwitch4-CZ is used to control the ClampZero active clamp switch, and the V pin is used to detect high input voltage conditions, which enables discontinuous operation
The secondary-side controller issues instructions to activate the HSD signal, turning on the ClampZero PowiGaN switch so that the leakage inductance and clamp capacitor resonate in the front of the primary PowiGaN switch. There is a very small delay between the off of the ClampZero device and the turn on of the main switch, which can be adjusted externally with a small resistor on the HSD pin to help optimize timing.
In continuous conduction mode, the HSD signal remains open for a quarter of the time of the leakage and clamp capacitor resonance periods. One of the challenges of using this resonant mode over a wide operating range is that the leakage inductance is usually a very small value, with higher voltages across the main switch tube at high voltage inputs, which requires more energy to achieve zero-voltage switching. Therefore, the leakage of energy storage is often not enough. This is why the discontinuous conduction control mode needs to be involved at this time.
For discontinuous conduction mode (high input voltage operation), the HSD signal pulse width becomes excitation inductance (plus leakage inductance, although the leakage inductance is typically very small compared to excitation inductance) and the clamp capacitor produces a quarter of the resonant period in which the resonant is generated. The input voltage information on the V pin is used to control the start-up of discontinuous conduction mode. When high input voltage conditions are detected, the delay between the ClampZero shutdown drive signal and the main switch turn-on drive signal also increases. This provides more time for resonance between the excitation inductor (plus the leakage inductance) and the clamp capacitor to reduce the voltage on the main power switch. This mode of operation does not require the pulse train operation required by the complementary active clamping circuitry, avoiding the risk of higher output ripple and audio noise associated with complementary mode control.
summary
Offline flyback power supplies require the use of primary-side clamp circuitry to protect the power MOSFETs. Using passive RCD clamps is less expensive but less performant. Using active clamps with complementary control schemes can improve performance, but there are still limitations. The InnoSwitch4-CZ IC product family provides a unique control architecture for more complex, non-complementary active clamp control, enabling the design of highly efficient, ultra-compact USB PD chargers with very wide input voltage ranges and large variations in output voltage setpoints. Power Integrations' InnoSwitch4-CZ/ClampZero chipsets can be used to simplify active clamping schemes with non-complementary controls and accelerate time to market.