Ultra-low current isolated switching power supply design challenges

In non-continuous transmission communication systems, switching power supplies often need to work under contradictory conditions, that is, they require high isolation between input and output and extremely low quiescent current in standby mode. Since the power consumed in the working state is much higher than the standby power consumption, this combination of requirements increases the design difficulty. Due to the need to compromise between isolation and low power consumption, currently commercial power modules can hardly meet such requirements.

Wireless communications have grown rapidly over the past five years and will continue to grow. In addition to GSM and 3G mobile communication systems, emerging communication technologies such as Bluetooth, WiFi, Winmax, and ZigBee based on various versions of the IEEE wireless standard 802.xx1 are also developing. Currently, there is an increasing demand for monitoring, including small wireless monitoring and control devices, which must meet demanding size and power requirements. To meet these requirements, integrated suppliers must reduce system size and power consumption through highly integrated chips2,4.

An important indicator of wireless device power supply is to extend the battery life. The main design goal is to reduce power consumption while ensuring the performance of the wireless communication system. Based on the above conditions, the following design features need to be considered:

Non-continuous sending and receiving;

Power supply filtering or voltage stabilization;

Highly efficient circuit topology.

The first characteristic mentioned above depends on the transmission system, the second requirement can be achieved by the switching power supply, and the third requirement is determined by the power consumption of the switching power supply itself. In addition, the standby power consumption needs to be reduced as much as possible. Therefore, the system is optimized based on the above three characteristics.

Discontinuous receiving/transmitting

Because transmitters and receivers consume the most power in wireless systems, many devices use discontinuous transmission/reception schemes to optimize the spatial interface resources and efficiency of the communication link. Since the wireless communication unit does not work continuously from time to time, it helps to reduce the overall power consumption.

On the other hand, discontinuous transmission introduces large voltage ripple and current peak in the power supply5. The stability of the bias voltage directly affects the performance of the transceiver, and the drop of the power supply voltage will greatly reduce the working index of the RF circuit, making it difficult to meet the specification requirements of the communication equipment. When the system is powered by a battery, the battery life and discharge characteristics are also very sensitive to the peak current of the load.

Power supply filtering and voltage regulation

The power supply can be filtered with a large capacitor or other techniques (Reference 6). The supply voltage is regulated by a linear regulator or a switching power supply, which not only reduces ripple but also reduces EMI to maintain the working performance of the wireless device.

High Power Supply Topologies

The efficiency of the power supply is very critical, so the best topology switching power supply needs to be selected. Table 1 lists the common commercial DC-DC conversion modules, but these modules cannot meet our target requirements: maintaining ultra-low power consumption when no-load. Even non-isolated power supplies consume considerable current when no-load. Our goal is to limit the power supply current to less than 12mA under no-load conditions. To achieve this goal, we divide the standby current and quiescent current as follows:

Quiescent current is the supply current required to maintain voltage regulation under no load;

Standby current is the supply current when the power supply is not providing a regulated output to the system.

Finally, we also need to provide isolation to provide the necessary protection for the system to operate reliably in harsh environments.

Current state of the art

The above discussion shows that the following issues need to be considered when designing a power supply for a wireless device:

Very low no-load power consumption;

isolation;

Efficiency and size.

Based on the above three conditions, the following three aspects should be paid attention to when designing an efficient converter:

isolation;

Control methods;

Topology of the feedback loop.

isolation

The input and output isolation of the power supply is achieved by a transformer. For the inverter and flyback topologies, the energy is stored in the transformer inductor. The problem is how to provide feedback from the secondary side of the transformer to the primary side. Most systems implement this by using additional windings or optocouplers. The auxiliary winding increases the complexity and cannot guarantee a sufficiently accurate output voltage at low voltage outputs and when the load changes.

When the power system is working stably, the optocoupler needs a stable current to flow through the primary LED. To optimize the system, this current needs to be reduced as much as possible (Figure 1). This current can be minimized by reducing the transfer coefficient (CTR) of the optocoupler at low currents (63% at 10mA and 22% at 1mA) and reducing the speed of the optocoupler. In addition, the current of the error comparator and precision reference TLV431 needs to be kept to a minimum (Ikmin = 100μA).

Figure 1. The output-voltage-divider circuit generates the error comparator signal used to isolate the switching power supply shown in Figure 3.

For the voltage divider resistors R131 and R137 connected to the reference output, large resistance values ​​must be selected to reduce current. When designing, it is necessary to consider compensating for the input current and the delay caused by the input capacitance (this problem can be solved by capacitor voltage division). Since the output capacitor (C47) is large, low ESR capacitors (tantalum, OsCon, organic aluminum capacitors, etc.) need to be selected. In addition, the capacitor is required to have extremely low leakage current, because leakage current (especially at high temperatures) will produce large losses (for a 16V Kemet T495 100μF capacitor, IL is equal to 16μA at 25°C and reaches 160μA at 85°C).

Control Circuit

The most common power supply topology is current mode pulse width modulation (PWM), which controls the charging current of the inductor by changing the pulse width. When the load is heavy, the pulse width is increased to allow the inductor to store more energy (Figure 2). When the load is light, the pulse width is reduced to reduce the inductor energy storage. For low current loads, the power supply operates in discontinuous mode, and the main current loss comes from the power supply itself.

Figure 2. Pulse-width modulator (PWM) control generates a control voltage (center) and inductor current (bottom line) in response to changes in load current (top line).

The biggest benefit of PWM is fixed frequency, which simplifies the circuit design of EMI control and improves efficiency under heavy load. Its main disadvantage is the relatively large current loss under light load or no-load conditions because the internal oscillator of the regulator operates at a fixed frequency (for example, the current loss of UC3845 under light load is: Icc = 17mA). Figure 3 shows the typical current loss of the voltage and current feedback network of the UC3845 main controller.

Figure 3. The PWM controller (U41) uses optocoupler isolation to generate feedback from the transformer secondary to the primary.

Feedback Network Topology

Voltage feedback is provided by the current flowing through the phototransistor (inside the optocoupler U45) to R135, which needs to be as large as possible to reduce power dissipation, but must also maintain a resistance value that allows the optocoupler to operate normally.

The current feedback is generated by the voltage drop across R134. To reduce power consumption, a voltage divider is used between this voltage and the reference (VREF = 5V, pin 8) using R125 and R133, resulting in a 1V voltage at ISENSE (pin 3). The divider resistors must be large enough to reduce power consumption; however, care must be taken that the RC filter formed by the resistors and C53 does not affect the current signal. The power consumption in the oscillator components R126 and C46 is unavoidable because the output voltage must be maintained at all times.

Update solution to further reduce power consumption

For power supplies based on the UC38C41 or MAX5021 PWM controller and the TLV431C or MAX8515A precision reference, power dissipation can be further reduced through several channels. The power dissipation can be reduced by selecting the appropriate components.

Error Comparator

The TL431 is usually chosen to provide a precision reference, but it is not suitable for this design because the voltage it generates (VA-Kmin = VREF = 2.5V, plus the voltage difference between the U45 LED and R124) is too close to the output voltage of 3.6V. An alternative is to use the MAX8515 shunt reference, which has a reference voltage of only 0.6V and maintains 1% accuracy over the -40oC to +85°C temperature range. This IC is the best choice for low-voltage output applications because it does not have the above-mentioned high reference voltage limitation (the reference voltage reaches 2.5V).

Another choice for this application example is the TLV431C shunt reference, which is available from several suppliers and meets the reference requirements: VREF = 1.24V, 1% accuracy from 0oC to +70°C. The voltage divider network current is fixed at 24μA, ensuring that the reference current (temperature drift 0.5μA) has no significant effect on the output voltage. In addition, it is necessary to ensure that the signal delay caused by the input capacitor does not affect the normal operation of the circuit.

PWM Controller

The traditional UC3845 (Figure 3) controller current consumption is about 17mA (VFB and VSENSE = 0V), which is too high for this application. The MAX5021 can be used instead. The MAX5021 is packaged in SOT23-6, which is the smallest size among similar ICs. It also has the lowest operating current (1.2mA), built-in 260kHz oscillator, 0.6V VISENSE, and can be directly input by optocoupler. Many of its features can meet the requirements of this type of application. Its disadvantage is that the undervoltage protection threshold of 10Voff/24Von is not suitable for 12V input applications. In addition, it has extremely low standby current, which is very suitable for high input voltage occasions.

The last IC to consider is the UCC38C41, which has an undervoltage protection threshold of 6.6Voff/7.0Von and a typical current consumption of ICC = 2.3mA. The current detection circuit consumes 100uA and the optocoupler consumes 530uA. To maintain the phototransistor current, the LED requires at least 1mA. The resulting power supply is approximately 50x30mm in size and includes two optocouplers, one for control loop feedback and the other for detecting the battery voltage at the input. The performance of this power supply is as follows:

Power = 3.6W;

Input voltage range: 10V to 15V;

Nominal input voltage Vin = 12V;

Isolation (galvanic isolation is required);

Buck flyback topology;

Voltage and current control loops;

PWM control mode;

The switching frequency is 250kHz;

The maximum output current is 1A;

The output voltage is 3.6V;

No-load current 5.7mA.

Test Results

Figure 4 is a prototype circuit for several wireless modules, which work in discontinuous mode with a maximum peak current of 3A and a maximum average current of 1A. In order to reduce the maximum peak current and solve the problems caused by it, the design needs to refer to the relevant technologies discussed in References 5 and 6. It is recommended to use large-capacitance, low-ESR capacitors.

Figure 4. This board includes optocoupler isolation for the power supply shown in Figure 3.

The test results (Tables 2 and 3) do not include the common-mode losses of the input filter circuit and the protection circuit. Table 2 gives the no-load current at different input voltages.

The minimum current can reach 5mA, which can be further reduced to 3mA, but it may cause system instability. In order to prevent self-excitation and consider the tolerance of components, the minimum current is set slightly higher than 5mA to leave a certain margin. As shown in Table 3, under nominal operating conditions and typical loads, the circuit is optimized to achieve the highest efficiency. Figure 5 shows the efficiency at different output currents.

Figure 5: The efficiency curve of the power supply shown in Figure 3 shows that the efficiency is very stable (the curve remains flat) under different loads at a nominal 12V input voltage.

According to the data we have, the minimum no-load current of commercially available isolated power supplies with similar characteristics is 20mA. Using common components, the application circuit introduced in this article can reduce the quiescent current to 5mA, which is our design goal of 12mA.