Building a power supply for discontinuous transmission wireless networks - Embedded.com

Building a power supply for discontinuous transmission wireless networks

Wireless communications have grown rapidly over the last five years, and that growth is expected to continue. Besides GSM (Global System for Mobile) and 3G mobile communication systems, new communication technologies such as Bluetooth, WiFi, WiMAX, and ZigBee have emerged, all based on various versions of the IEEE wireless standard 802.xx.1

The number of control and monitoring applications that include small wireless sensors and control devices has increased substantially, and those devices must contend with the major constraints of size and power. To meet these constraints, the chipset vendors reduce system size2 and power consumption3 by producing ever more highly integrated chips.

However, in wireless communications environments, which are typically characterized by discontinuous transmission, the power-supply circuits operate in a hostile environment that requires high input-to-output isolation and low quiescent current in standby mode. That combination presents design difficulties, because the power consumption in operating mode is much greater than in idle mode. Few if any modules are commercially available for this purpose, due to the necessary trade-offs between isolation and low power consumption.

To address such issues, described here is an isolated, switched-mode power supply intended for wireless devices. It accepts a nominal 12-V input and delivers an isolated 3.6-V output, and its quiescent current draw is among the lowest available. Designed as a power supply for EGSM (Extended GSM), WiFi, and ZigBee communication modules, it also provides remote control for electromechanical actuators and electronic sensors in harsh environments.

Increasing autonomy
Because the main characteristic of a power supply for wireless devices is battery life, the main design goal was to reduce power consumption while maintaining performance in the radio-frequency system. Thus, care was taken that the autonomy of the wireless-communication device increased when these conditions are present:

• Discontinuous transmission and reception.

• Filtering or regulation of the supply voltage.

• A high-efficiency power-supply topology.

The first of these characteristics depends on the transmission system. The second can be obtained with a switched-mode power supply, and the third depends directly on power consumption in the DC-DC converter itself. You must take extra care, therefore, to minimize the current draw under no-load conditions. As a result, we stress this third point in recommending design techniques for optimizing the system.

Discontinuous transmission and reception
Because transmitters and receivers consume the most power in a wireless device, many such devices implement discontinuous transmission/reception to optimize the resources of the air interface and the efficiency of the communications link. Discontinuous operation also helps to reduce power consumption, because the radio's active elements are not continuously on.

On the other hand, discontinuous transmission introduces voltage ripple and current peaks in the power supply.4 The level of bias voltage affects transceiver performance directly, and a drop in supply voltage degrades radio performance, mainly at extremes of the voltage range. This degradation can make it difficult to meet the applicable specifications for certifying a wireless device. Battery life cycle and discharge characteristics are also sensitive to current peaks in a load, if the system is powered by a secondary cell.

Filtering or regulation of the power-supply voltage
The supply voltage can be filtered with a high-valued capacitor or other techniques described in a paper by Jose Ignacio Garate and others.5 The voltage is regulated with a DC-DC converter operating in linear or switched mode. Regulation is not only necessary to reduce voltage ripple, but also to reduce EMC problems and maintain performance of the radio.

Topology of the power supply for high efficiency
Power-supply efficiency is important, so a switching topology offers the best power-supply option. However, the DC-DC converter modules listed in Table 1 , which are typical of those commercially available, don't meet the requirement we're looking for: ultra-low power consumption under no-load conditions.

View the full-size image

Even the nonisolated converters draw a relatively high current with no load. Taking this data into account, we therefore set a goal of 12 mA maximum for the DC-DC converter's no-load current. To achieve that goal, we distinguish between standby current and quiescent current as follows:

Quiescent current is the supply current necessary to maintain a regulated supply voltage under no-load conditions.

Standby current is the supply current drawn when the system is not producing a regulated output voltage.

Finally, you may need to add galvanic isolation to ensure efficient protection for devices designed to operate in hostile environments.

Design goals and issues

In the design of a power supply for portable or wireless devices, three key technical requirements should be taken into account: very low no-load power consumption, isolation, efficiency, and size. Issues that arise in meeting these requirements involve the correct isolation configuration, the control method, and the topology of the feedback loop

Isolation
Isolation between input and output is achieved with a transformer, and for inverting or flyback configurations, the energy is stored in the transformer's inductance. The problem, then, is how to provide feedback from the secondary to the primary. The majority of systems use an extra auxiliary winding or optocoupler for that purpose. An auxiliary winding, however, increases the complexity while not providing enough output-voltage precision for low outputs and variable loads.

When the system is in regulation, an optocoupler needs constant current through the primary-side LED. To optimize the system, this current has been minimized as much as possible, as Figure 1 shows.

View the full-size image

The minimum limit is set by a reduction of the optocoupler's current transfer ratio (CTR) at low current (63% at 10 mA and 22% at 1 mA), and a reduction of speed (2µs at 20 mA and 6.6 µs at 5 mA). We must add yet another limitation, which is the minimum current (Ikmin = 100 µA) that must flow from the error comparator through the precision shunt regulator TLV431.

For the resistive divider connected to the shunt regulator's output (R131 and R137), high-value resistors have been selected to minimize the current draw. You must take care to compensate for possible delays caused by the input current (Iref = 0.5 µA) and the input capacitance (this problem can be solved using capacitive dividers). The output filtering capacitor (C47) is large, so you may need to choose a low-ESR electrolytic type (such as tantalum, Os-Con, and organic aluminum). If so it must also be a low-leakage type, because current leakage can be significant, especially at higher temperatures. (For a 16-V Kemet T495 100 µF capacitor, IL equals 16 µA at 25°C and 160 µA at 85°C.)

Control circuit
The most common power-supply control scheme is current-mode pulse-width modulation (PWM), in which variable-width pulses control the inductor's charging current. When the load is heavy, the width of the applied pulse increases to store more energy in the inductance, as Figure 2 shows.

View the full-size image

Under no-load or light-load conditions, the width of the control pulse narrows to store less energy in the inductor. For low-current loads, the power supply operates in discontinuous mode and the main current draw is the power supply itself.

The main advantage of PWM control is its fixed switching frequency, which simplifies the circuit design with regard to controlling EMI and optimizing efficiency for heavy loads. Its main drawback is the current draw under no-load and light-load conditions, because the oscillator in the regulator chip has a fixed frequency (a lightly loaded UC3845, for example, draws Icc = 17 mA). Figure 3 shows a typical UC3845 configuration in which the total current drawn is that of the main controller plus that of the voltage- and current-feedback networks.

View the full-size image

Topology of the feedback loop
Voltage feedback is produced by routing current from the phototransistor (of optocoupler U45) through R135. The value of R135 must be large to minimize power, but must also be small enough to supply the minimum current needed for operation of the phototransistor.

Current feedback is obtained though the voltage drop in R134. To minimize loss, we use R125 and R133 to divide the drop between this voltage and the reference (VREF = 5 V, pin 8), thereby allowing a 1-V equilibrium at ISENSE (pin 3). These connections improve efficiency by reducing the voltage drop in R135. The divider resistors must have high values to minimize power consumption, but you must also take care that the RC filter formed with C53 doesn't affect the current signal. Power consumption in the oscillator components (R126 and C46) is unavoidable, because the voltage output must be maintained at all times.

New approaches to further reduce current

Several alternatives, based on Texas Instruments' UCC38C41 or Maxim's MAX5021 PWM controller and TI's TLV431C or Maxim's MAX8515A precision reference, can further reduce current drawn by the converter. The associated components are selected for minimum possible power consumption.

Error comparator
The classic TL431 can sometimes implement a precision reference. That option is not available in this case, because the resulting voltage (VA-Kmin = VREF = 2.5 V, plus drops in the U45 LED and in R124) is too close to the desired 3.6-V output. One alternative is the MAX8515A shunt regulator from Maxim. It includes a voltage reference of only 0.6 V, with 1% tolerance in the range -40ºC to +85°C. This IC is the best option for applications in which the circuit must provide lower output voltages, because it doesn't have the limitation mentioned above (a “large” reference voltage of 2.5 V).

Another option for this example is the TLV431C shunt regulator. Available from several manufacturers, it also meets the requirement of VREF = 1.24 V, with 1% tolerance from 0°C to +70°C. Current through the output divider is fixed at 24 µA, to ensure that the reference current (0.5 µA, with thermal drifts) doesn't have a significant effect on the output voltage. Also, the signal delay due to input capacitance is not enough to warrant the substitution of a capacitive divider.

PWM controller

The classic UC3845 shown in Figure 3 draws about 17 mA (VFB and VSENSE = 0V), which is excessive for this application. One possible alternative is the MAX5021 current-mode PWM controller. Available in a SOT23-6 package, it is among the smallest in its class. It also has the lowest typical current draw (1.2 mA), plus a 260kHz internal oscillator, VISENSE of 0.6 V, direct input from the optocoupler, and other features suitable for this application. One drawback, however, is an undervoltage-lockout threshold of 10 Voff /24 Von , which makes it also unsuitable for this particular 12-V input application. On the other hand, its ultra-low standby current makes it the preferred choice for other, higher-input-voltage applications.

The final IC to be considered is the UCC38C41, which specifies an undervoltage lockout of 6.6 Voff /7.0 Von and a typical current draw of ICC = 2.3 mA. In the voltage adder, the current detector draws 100 µA (ICS = 2 µA), and the phototransistor of the optocoupler, 530µA. To allow that level of phototransistor current, the LED must draw somewhat more than 1 mA. The resulting power supply, shown in Figure 4 , measures less than 50 x 30 mm. It includes one optocoupler for the control-loop feedback and one for measuring battery voltage at the input. The power-supply characteristics are:

View the full-size image

• Power = 3.6 W.

• Input voltage range: 10 V to 15 V.

• Nominal Vin = 12 V.

• Isolated (needs galvanic isolation).

• Step-down flyback topology.

• Voltage and current control loops.

• PWM control scheme.

• Switching frequency: fC = 250 kHz.

• Maximum output current = 1 A.

• Regulated output voltage = 3.6 V.

• No-load current draw = 5.7 mA.

Measurements and results
The Figure 3 prototype circuit included several wireless modules featuring discontinuous transmission, with currents reaching maximum peaks of 3A and a maximum average of approximately 1A. To reduce the current peaks and the associated radio problems, you should make use of the techniques described in two papers by Jose Ignacio Garate and others.4, 5 A capacitor of high value and low ESR is highly recommended.

Measurement results (Tables 2 and 3 ) don't include losses in the common-mode input filter or the protection circuitry. Table 2 gives values for the power supply's input and output variables for different input voltages under no-load conditions.

View the full-size image

View the full-size image

The minimum current draw achieved is 5 mA. You can reduce that to 3 mA, but the low-value resistor necessary to achieve 3 mA causes the control loop to become unstable. To prevent self oscillation and to accommodate component tolerances, a security margin is introduced by setting the current draw slightly higher than 5mA. As shown in Table 3 , the optimum efficiency is reached at normal conditions with a nominal load. Figure 5 shows the efficiency for different output currents.

View the full-size image

For commercially available, isolated power supplies with similar characteristics, we found that the lowest no-load current draw is around 20 mA. By designing a circuit with off-the-shelf components that obtains a quiescent current of 5 mA, we have surpassed our goal of 12 mA.

Jose Miguel de Diego has a BS in electrical engineering from the E.T.S.I.I. of Navarra University, San Sebastin, and a doctorate in industrial engineering from the High School of Engineering of Bilbao, Spain. His research areas are power-supply systems and renewable energies, and he has worked in R&D at industrial electronics companies, including Ericsson Radio S.A., Spain. You may reach him at jm.dediego@ehu.es.

Jose Ignacio Garate has a BS and an MS degrees from the High School of Engineering of Bilbao, Spain, both in telecommunication engineering. He is working on his doctorate on discontinuous current consumption in battery-powered wireless terminals and output power control for radio-frequency power amplifiers. He worked in the R&D departments of Ericsson Radio S.A., Spain, and Ericsson ABB, Lund and Gable, Sweden. You may reach him at joseignacio.garate@ehu.es.

Javier Monsalve Kagi graduated with a BSEE and MSEE from Worcester Polytechnic Institute, Massachusetts. He is a field applications engineer at Maxim Integrated Products, Inc., and previously worked for Spanish companies doing analog and digital electronic design in avionics and designing an IC for temperature measurement and voice synthesis (thermometer for the visual impaired). You may reach him at javier.monsalve@maxim-ic.com.

Endnotes:
1. Carney, William. “IEEE 902.11g: New Draft Standard Clarifies Future of WLAN,” Texas Instruments Inc., 2002. PDF available on http://focus.ti.com/docs/solution/folders/print/47.html.

2. Haroun, Ibrahim, Ioannis Lambadaris, and Roshdy Hafez. “RF System Issues in Wireless Sensor Networks,” Microwave Engineering Europe , Nov. 2005, pp. 31–35. PDF available on www.mwee.com/mag_archive/mwee1105.html.

3. Application Note 664, “Feedback Isolation Augments Power-Supply Safety and Performance,” Maxim Integrated Products, Jan 22, 2001 (see www.maxim-ic.com/an664). Also see June 19, 1997 issue of EDN (www.edn.com/archives/1997/061997/13df_04.htm).

4. Garate, Jose Ignacio, Jose Miguel de Diego, and Maite Sierra. “Consequences of Discontinuous Current Consumption on Battery Powered Wireless Terminals,” [ISIE06, Paris, France, Oct. 2006]. IEEE Explore: http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=4153462

5. Garate, Jose Ignacio and Jose Miguel de Diego. “Improvements of Power Supply Systems in Machine to Machine Modules and Fixed Cellular Terminals with Discontinuous Current Consumption,” (Digests 9th ICIT06, Mumbai, India, Dec. 2006.)

Further reading:
“MAX1649/MAX1651, 5V/3.3V or Adjustable, High-Efficiency, Low-Dropout, Stepdown DC-DC Controllers,” Maxim Integrated Products, Datasheet 10-0305, Rev 2; 9/95. www.maxim-ic.com/quick_view2.cfm/qv_pk/1028

“MAX1771, 12V or Adjustable, High-Efficiency, Low IQ , Step-Up DC-DC Controller,” 2002, Maxim Integrated Products, Datasheet 19-0263; Rev 2; 3/02. www.maxim-ic.com/quick_view2.cfm/qv_pk/1030

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.