Saturated inductance characteristics ● Thermal characteristics
A saturated inductor is a power device that absorbs current spike energy through hysteresis losses (rather than eddy current losses or copper losses) during the saturation process. The main heat power comes from the magnetic core.
This requires, on the one hand, that the magnetic core be made of high-frequency material, and on the other hand, that the core temperature never exceed the Curie temperature under any circumstances. This means that the magnetic core of a saturated inductor should have the most favorable heat dissipation characteristics and structure, namely: higher Curie temperature, higher thermal conductivity, larger heat dissipation area, and shorter heat conduction path.
Basic methods of buffering:
● Inserting an inductor of a certain type along the path of the surge current spike, such as one with buffering characteristics:
●Since the insertion of a buffer inductor significantly increases the workload of absorption, the buffer circuit generally needs to be used in conjunction with an absorption circuit.
● The buffer circuit delays the surge of the conduction current, enabling a certain degree of soft turn-on (ZIS).
● Transformer leakage inductance can also act as a buffer inductor. LD buffer characteristics:
● No absorption circuit is required.
● The current stress of the buffer diode is comparable to or even greater than that of the topological freewheeling diode.
● The loss of the buffer diode can be simply understood as the loss reduced by the switching transistor.
● Properly designed buffer inductor (L3) parameters can significantly reduce switching losses and achieve high efficiency. LR buffer characteristics:
●A snubber circuit is needed to transfer the remaining energy from the inductor.
●The loss of the buffer energy release resistor R is relatively large, which can be simply understood as the loss transferred from the switching transistor.
● The R and L parameters must be optimally matched, and parameter design and debugging are relatively difficult to master.
●High efficiency can still be achieved as long as the parameters are appropriate.
Saturated Inductor Buffer
●The electrical properties of a saturated inductor are sensitive to di/dt.
● At the rising edge of an inrush current, it initially exhibits a large impedance, which gradually enters saturation as the current increases, thereby delaying and weakening the inrush current spike, thus achieving soft turn-on.
●When the current reaches a certain level, the saturated inductor exhibits very low impedance due to saturation, which is beneficial for efficient power transmission.
● When the current is turned off, the inductor gradually exits the saturation state. On the one hand, the saturation inductance in the previous saturation state is very small, meaning that the stored energy and the required energy release are small. On the other hand, the recovery of the inductance during exiting the saturation state can slow down the rate of voltage rise, which is beneficial for achieving soft turn-off.
● Taking Ls2 as an example, 5u represents a magnetic circuit cross-sectional area of 5mm², which is roughly equivalent to a small 442mm² magnetic core made of PC40 material.
●Saturation characteristics
Obviously, saturated inductors generally do not need to use air gaps or low permeability materials that are not easily saturated.
● Initial inductance equivalent characteristics
Under otherwise identical conditions, a magnetic core with lower permeability and more turns has a similar initial inductance to a saturated inductor with higher permeability and fewer turns, and the buffering effect is roughly the same.
This means that it is always possible to directly use a 1-turn feedthrough inductor, because any multi-turn inductor can always be matched with a core of higher permeability to achieve the same 1-turn equivalent. It also means that the maximum permeability of the core is limited; if a suitable core is used with a 1-turn saturated inductor, there is no possibility of using a core of higher permeability with fewer turns.
●Equivalent characteristics of magnetic core volume
All other things being equal, the saturation inductance buffering effect of magnetic cores of the same volume is roughly equivalent. Therefore, magnetic cores can be designed according to the magnetic circuit that best facilitates heat dissipation. For example, a slender tubular magnetic core obviously has a larger heat dissipation surface area than a toroidal core, multiple small magnetic cores are better than a single large magnetic core, and a feedthrough inductor is obviously better than a multi-turn inductor.
●Combination characteristics
Sometimes, a single-material magnetic core is insufficient to achieve the required buffering effect in engineering. Using a combination of magnetic cores made of different materials may be necessary to meet the engineering requirements. Passive and Lossless Buffer Absorption: ● If the buffer inductor itself is lossless (non-saturated inductor), and its inductive energy storage is processed through lossless absorption, it constitutes a passive and lossless buffer absorption circuit, which is essentially a passive soft-switching circuit.
●The presence of a buffer inductor delays and weakens the turn-on inrush current, achieving a certain degree of soft turn-on.
●The presence of the lossless absorption circuit delays and reduces the dv/dt of the turn-off voltage, achieving a certain degree of soft turn-off.
● The conditions for achieving passive soft switching are roughly the same as those for lossless absorption. Not all topologies can be used to build a passive soft switching circuit. Therefore, in addition to classic circuits, many passive soft switching circuits are highly sought-after patents.
● Passive, lossless soft-switching circuits are significantly more efficient than other buffering and absorption methods, and their efficiency is comparable to that of active soft-switching circuits. Therefore, if a circuit can achieve passive soft switching, there is no need to use active soft switching. ● Electrolytic capacitors in the circuit generally have a large ESR (typically on the order of hundreds of milliohms), which causes two problems: first, the filtering effect is greatly reduced; second, the ripple current generates significant losses in the ESR, which not only reduces efficiency but also directly leads to reliability and lifespan issues due to the heating of the electrolytic capacitor.
●The common method is to connect a high-frequency lossless capacitor in parallel with the electrolytic capacitor. However, this method cannot fundamentally change the above problem because the high-frequency lossless capacitor still has a large impedance in the commonly used frequency range of the switching power supply.
The proposed solution is to separate the electrolytic capacitor and the CBB using an inductor. The CBB is located on the high-frequency ripple current side, while the electrolytic capacitor is located on the DC (power frequency) side, each undertaking its corresponding filtering task.
●Design principle: The resonant frequency Fn of the Π-shaped filter network should be offset from the PWM frequency Fp. Fp can be taken as (1.5～2)Fn.
●This design concept can be extended to bidirectional buffers for DC bus filtering, or other circuit structures with significant filtering stress. Ringing hazards:
●MEI test is prone to exceeding the standard in ringing frequency.
● Ringing will cause losses in the ringing circuit, resulting in device heating and reduced efficiency.
●If the ringing voltage amplitude exceeds the critical value, it will cause ringing current, disrupting the normal operating condition of the circuit and significantly reducing efficiency.
Causes of ringing:
● Ringing is mostly caused by the resonance of junction capacitance and an equivalent inductance. For a specific frequency of ringing, a cause can always be found. Capacitance and inductance can determine a frequency, and the frequency can be observed. Capacitors are mostly the junction capacitance of a device, while inductance may be leakage inductance.
● Ringing is most likely to occur in lossless (resistance-free) circuits. For example: resonance between the secondary diode junction capacitance and secondary leakage inductance, resonance between stray inductance and device junction capacitance, resonance between the absorption circuit inductance and device junction capacitance, etc.
Suppression of ringing:
●Magnetic bead absorption: As long as the magnetic bead exhibits resistance at the ringing frequency, it can absorb a large amount of ringing energy. However, inappropriate magnetic beads may also increase ringing.
●RC absorption, where C can be roughly equivalent to the ringing (junction) capacitance, and R is selected according to the RC absorption principle.
● Change the resonant frequency. For example, simply reducing the ringing frequency to be close to the PWM frequency can eliminate ringing on the PWM.
●In particular, improper design of the input and output filter circuits can also cause resonance, requiring adjustment of the resonant frequency or other measures to avoid it. Energy absorption and reuse: RCD energy recovery circuit ●By separating the forward and reverse loops of the absorption circuit to form positive and negative current paths with relative zero potential, positive and negative voltage outputs can be obtained. The key design points are:
● The parameters of the RCD snubber circuit should primarily meet the snubber requirements of the main circuit. Increasing the snubber power to increase the DC output power is not recommended. The output current is controlled by L1 and R1. The impedance of the flyback loop should also meet the flyback time requirements of the snubber loop. Adjusting the values of L1 and R1 controls the output power. When R1 is reduced to 0, the circuit reaches its maximum possible output current and maximum output power.
● The output voltage can be basically set arbitrarily by the Zener threshold voltage; however, attention must be paid to the power matching of the Zener diode. RCD clamping energy recovery circuit.
● A circuit for a 12V 1KW secondary-side full-wave rectifier, originally a 3.5W RC circuit, uses an RCD clamp to absorb and recover energy for a 3W 24V fan. The output voltage of the RCD clamp absorption circuit is related to the clamping voltage, and its controllable range is limited. If the load of the recovered power supply is uncertain, it is necessary to ensure that the absorption state remains unchanged under any load condition and does not affect the main circuit. Pay attention to the grounding of the recovery circuit to avoid it becoming a source of common-mode interference. Adjust R1 to strictly control the absorption degree and ensure clamping operation.
When measuring ripple, it is important to know the upper limit of the ripple bandwidth. Since ripple is low-frequency noise, an oscilloscope with a bandwidth not exceeding the upper limit of the ripple is generally used.
During measurement, the oscilloscope's bandwidth limiting function must be enabled first, limiting the bandwidth to 20MHz.
Connect the probe's shield ground and output ground directly to reduce loop interference caused by excessively long ground wires.
A small ceramic capacitor and a small electrolytic capacitor are connected in parallel at the probe access point to filter out external interference signals and prevent them from entering the oscilloscope.
IV. Methods for suppressing ripples
Power supply output ripple mainly comes from five sources: low-frequency input ripple, high-frequency ripple, common-mode ripple noise caused by parasitic parameters, and ripple noise caused by closed-loop regulation control.
Common methods to suppress these ripples include increasing the capacitance in the filter circuit, using an LC filter circuit, employing a multi-stage filter circuit, replacing the switching power supply with a linear power supply, and proper wiring. However, taking targeted measures based on their classification often yields better results.
1. Suppression of high-frequency ripple
High-frequency ripple noise often originates from high-frequency power conversion circuits. In high-frequency power conversion circuits, the input DC voltage is converted by high-frequency power devices and then rectified and filtered to achieve a regulated output. This output typically contains high-frequency ripple with the same frequency as the switching operating frequency. The magnitude of its impact on external circuits depends mainly on the switching frequency of the power supply, the structure and parameters of the output filter. In the design, maximizing the operating frequency of the power converter can reduce the filtering requirements for high-frequency switching ripple.
2. Suppression of low-frequency ripple
The magnitude of low-frequency ripple is related to the size of the filter capacitor in the output circuit. The capacitance cannot be increased indefinitely, inevitably resulting in residual low-frequency ripple at the output. The AC ripple, after attenuation by the DC/DC converter circuit, falls within the low-frequency noise range, and its magnitude is determined by the gain of the control system and the DC/DC converter circuit. Since the ripple suppression capability of current-source and voltage-source controlled DC/DC converter circuits is relatively low, and their output low-frequency AC ripple is relatively large, filtering measures must be taken to achieve low-ripple output from the power supply.
For some power supplies, increasing the closed-loop gain circuit of the DC/DC converter and using a pre-regulator circuit can enhance the ripple suppression effect. Low-frequency ripple can also be suppressed by changing the capacitance of the rectifier filter and adjusting the parameters of the feedback loop.
3. Suppression of common-mode ripple
Common-mode ripple noise typically occurs in switching power supplies. When a rectangular wave voltage from a switching power supply is applied to power devices, it interacts with the parasitic capacitance between the power devices and the heatsink base and the primary and secondary windings of the transformer, as well as the parasitic inductance in the wires, generating common-mode ripple noise. Methods for suppressing common-mode ripple noise include:
1) Reduce the parasitic capacitance between the control power devices, transformer, and chassis ground, and add a common-mode rejection inductor and capacitor at the output terminal;
2) EMI filters can effectively suppress common-mode ripple interference;
3) Reduce the amplitude of switching glitches.
4. Suppression of closed-loop control circuit ripple
The cause of closed-loop control loop ripple is generally due to inappropriate parameter settings in the loop. When there is a certain fluctuation at the output, the feedback network feeds the fluctuating voltage at the output back to the regulator loop, causing the regulator to generate a self-excited response, thereby generating additional ripple.
The main suppression methods include: suppressing the regulator's self-oscillation response, appropriately selecting the loop amplification factor, improving regulator stability, and connecting an LDO filter to the power supply output. These are the most effective methods for reducing ripple and noise.
