Low power consumption is imperative for modern meter design. Whether you’re measuring gas, water, electricity or heat, you must limit the current draw to achieve long battery life and/or prevent inaccurate readings. Fortunately, TI provides a variety of solutions to meet these requirements. In this post, I’ll review some common configurations to see where TI products can aid your design.

Electricity meters

Electricity meters are everywhere but their power designs are hardly constant. Driven by cost, size and efficiency, designers implement power schemes of different complexities. The differences spur from how they decide to convert the AC line voltage to a DC rail for the microcontroller (MCU)orsystem on chip (SoC). Figure 1 is a high-level example of a single-phase electric meter.

Figure 1: Single-phase electricity-meter block diagram

There are several common techniques employed to generate a regulated DC rail for the MCU. The most cost-conscious implementation is the capacitor drop, which involves traditional half-wave rectification via a diode and capacitor, combined with a capacitor and resistor in series and a Zener diode in parallel (see Figure 2).

Figure 2: A simplified capacitor-drop power supply

The resistor and capacitor in series limit the amount of current being sourced. The Zener diode regulates the voltage at the input of the low-dropout regulator (LDO). (The Zener diode should not be used as the point-of-load [PoL] regulator to the MCU. Because the Zener voltage can vary a considerable amount, using a Zener diode as the PoL regulator risks regulating a voltage outside the MCU’s specified supply-voltage range. A Zener diode also does not incorporate other important features common in an LDO, including current limit and load and line regulation.) For more information on these components, see the additional resources section below.

The LDO provides a stable supply-voltage rail for the MCU. However, you can’t use just any LDO in this scheme. Since the voltage at the input of the LDO is prone to transient voltage spikes, the LDO must have a wide-input voltage range to accommodate such events. The system’s leakage current must also be controlled to comply with industry standards. Therefore, you must have an LDO with small quiescent current to limit current draw. The TPS709 is an example of a regulator that has both a wide-input voltage range (up to 30V) and a low quiescent current rating of 1.3µA.

In the event that the line voltage goes down, measures should be taken to ensure continued operation. The optional diode shown in Figure 2 prevents the current from taking a reverse path should the input voltage of the LDO be lower than the output voltage. It is marked optional because some LDOs (like the TPS709) have integrated reverse-current protection, which makes this diode redundant.

Figure 3: A capacitor-drop supply with a battery connected after the LDO

Figures 3 and 4 feature a battery placed in parallel with the line voltage. This is the secondary power source when the line voltage is not present. The battery kicks in once its nominal voltage less the diode voltage drop is greater than the connected-node voltage. The battery can be connected after the LDO (Figure 3) or before (Figure 4). However, I recommend connecting the battery before the LDO when possible to ensure a stable regulated voltage within the range of the MCU.

Figure 4: A capacitor-drop supply with a battery connected before the LDO

Water meters

A water meter’s power supply is much simpler given that it does not operate off of an AC line; it uses a battery for operation. In some cases, this battery must provide operation for 10 years or more. To achieve this, current draw must be kept to a minimum. An LDO like the TPS782 aids this type of design since it both regulates an output voltage and draws minimal quiescent current, as shown in Figure 5.

Figure 5: The power supply for a water meter

Although it may not be intuitive, using an LDO in conjunction with a battery can reduce the amount of power consumption. This is due to the relationship between supply voltage and supply current in low power microcontrollers like TI’s MSP430. Although the supply voltage range may be wide (the supply voltage range of MSP430 may range from 1.8V to 3.6V), supplying a higher voltage rail will correspond with a higher supply current. Therefore, it is advantageous to use a lower voltage supply rail to curb unnecessary current draw. This is discussed at length in another blog linked in the additional resources section below.

Whether your meter runs off an AC line, a battery or both, using the right LDO ensures proper operation and reduced power consumption.

Additional resources:

• Learn more about capacitor-drop components and their design by reading the Capacitor Power Supplies section of this application note.
• Read the blog post, “Drive MSP430 low-power even lower – part 1.”

In 2013, the largest sporting event in the world experienced a 34-minute play delay due to a power outage.  The delay time was caused  by the use of older lighting technologies that required a cool-down period before being turned back on and a warm-up period to reach full output levels. This long delay could have been shortened with the use of low-power LED-based lighting solutions, like that from Ephesus Lighting Inc. These solutions offer wireless real-time control in the form of on/off, dimming, sequencing and other effects.

This type of implementation is relatively new to the market and enables a number of advantages in terms of energy efficiency and functionality over more traditional systems, but these systems do have some additional requirements to consider. The perception of hundreds of cameras flashing all over a stadium could be quite useful in highlighting an event, but this type of effect would require low latency for hundreds of wirelessly controlled fixtures to enable instant (to the human eye at least) on and off capabilities. Additionally, these systems can consume significantly less power than other solutions when on, but ultra-low power consumption when the lights are off can help contribute to meeting energy requirements. Security is another important requirement as well since these lights should only be controlled by a stadium’s operations team.

In the Ephesus system, Anaren Air modules (based on TI’s Sub-1 GHz CC1101 RF transceiver) are paired with an ultra-low-power MSP430FRxx FRAM microcontroller to enable low-power wireless lighting control. Mark Bowyer, Director of Wireless Business Development at Anaren, explained the use of an MSP in the system:

“It was the FRAM in the device that was a deciding factor in the selection. As power interruptions and inconsistent power cleanliness is a constant battle, we needed to be able to retain fundamental command and control code in more of an ‘e-ink’ style repository. This allows us to retain system reboot and scene command sets in a ROM-style storage medium with flash accessibility and speed characteristics while maintaining our demand for the lowest power consumption while in dormant mode.”

In a system like the one described above, there are several benefits enabled by FRAM. Combining the non-volatility, write speeds, and low power of our FRAM MCUs with the integration of AES modules and memory protection units are what make devices like the MSP430FR5969 MCU stand apart from competitive solutions.

Are you interested in creating a wireless system like this one?

Consider getting started with the new MSP-EXP430FR6989 LaunchPad and the 430BOOST-CC110L RF BoosterPack. Then why not try adding state restoration after power fail with the new Compute Through Power Loss (CTPL) FRAM utility.

At TI, we celebrate the makers and hobbyists who enjoy creating and innovating on their own time. In our ongoing DIY with TI series, we share their incredible Do It Yourself inventions using TI technology.

Trey German – enthusiastic geek and ambitious cook – knew he’d found the right mix of gadgetry and elusive Texas barbecue magic when he sliced the first brisket from the Internet-enabled smoker he created.

“I saw how the fat had perfectly rendered and the meat was all juicy, and I knew it was going to be something special,” he said. “That first bite was magical. It rivaled anything you could get at a barbecue joint.”

Like any good Texan, Trey loves barbecue. But he doesn’t like tending a smoker. Instead, like any good engineer, he created a do-it-yourself digital gadget that lets him simply tap his smartphone to control the temperature in the smoker and monitor the doneness of the meat from air-conditioned comfort.

The idea of a smoker connected to the Internet of Things may take a while to reach deep into the heart of traditional Texas barbecue culture, but Trey is a barbecue futurist. His smoker can send him a text or tweet at him when his meat is ready to eat.

“I’m a busy guy and I’ve got lots of stuff to do,” said Trey, LaunchPad applications manager based in Houston. “I can’t spend the time sitting by a smoker, but I really like smoked short ribs, brisket and things like that. This gives me a way to cook great barbecue without a huge time investment. We’re in Texas and we love our barbecue.”

The digital heart of his smoker was created with TI components, including an MSP430™ LaunchPad, a SimpleLink™ Wi-Fi^® CC3100 BoosterPack for wireless Internet connectivity, and an ADS1118 BoosterPack to communicate with thermo-couplers that monitor the temperature of the smoker and meat.

Trey’s smoker – at least its electronics – was one of three DIY innovations he entered in the DIY with TI event in Dallas in late May. Other projects were:

• A smart thermostat he developed that, from the convenience of a smartphone, can control an air conditioner to conserve electricity when a home isn’t occupied. That device – still a work in progress – uses an MSP432™ LaunchPad, a SimpleLink Wi-Fi CC3100 BoosterPack, and custom-designed boards to control the system and connect to the air conditioner.

• A souped-up electric vehicle he built for $500 to participate in the Power Racing Series, part of the Maker Faire held in San Mateo, Calif., in mid-May. The vehicle – a cart cobbled together with two children’s bicycles and battery packs from a Toyota Prius – incorporates TI microcontrollers and a power amplifier. It was designed to show off TI’s motor control technology. Moving forward, Trey plans to build a new car from scratch that includes an enhanced motor driver.

“DIY Day is an amazing event,” Trey said. “This brings a lot more people into the TI family and TI ecosystem. If we can show people all the cool things we can do with the technology, they’re going to want to do it, too. That’s the end goal. We want people to make cool stuff that’s going to make the world a better place.”

*The following contest deadline is Friday, July 17. Enter TODAY for your chance to win!

Do you dream of creating a solution to make monitoring health and fitness activities easier? Using near field communication (NFC) in your designs can help! We want to hear your ideas for the next wearable design based on our RF430FRL15xH NFC sensor transponder in our ultimate NFC wearable design contest. There’s no need to actually create your design just yet — all you need to do is share your concept with us. If your design concept wins, you’ll receive the tools you need to get started and potentially be eligible to win a Sony® SW3 SmartWatch 3 SWR50. Additionally, finalists will have the opportunity to be featured in a blog post on the Launch Your Design blog.

Here’s how you can enter to win (full rules and eligibility requirements can be found here):

1. Submit your IoT design idea that incorporates TI’s RF430FRL15xH NFC sensor transponder at the TI E2E™ Launch Your Design Portal by July 17, 2015.
2. A panel of three judges, including representatives from TI as well as Holly Lawrence, marketing and social media manager for Digilent Inc., will judge the entries based on criteria such as product/solution uniqueness, contemporary (”coolness”) factor and overall TI semiconductor usage.
3. Ten winners will receive the RF430FRL15xH  NFC sensor transponder and any of the following TI products from the list below:
   1. Sensor Hub BoosterPack: Evaluate the RF430FRL15xH  NFC sensor transponder with seven on-board sensors, including motion tracking, a pressure sensor, humidity and ambient temperature sensors, ambient and infrared sensors and a non-contact infrared temperature sensor.
   2. TRF7970A evaluation module: Test the performance of RFID/ NFC transceivers, custom firmware, customer designed antennas and potential transponders for a customer-defined RFID/NFC application.
   3. MSP430 flash emulation tool: Features a USB debugging interface to connect any MSP430™ microcontroller (MCU) to a PC for real-time, in-system programming and debugging.
   4. NFC sensor transponder patch: Features a flexible PCB, small form factor, battery less design with an integrated temperature that can be used as the development platform.
   5. AFE4403 analog front-end EVM: A fully-integrated analog front-end (AFE) EVM suited for heart rate monitors and pulse oximeter applications.
   6. After the ten winners have been determined and contacted on July 24, 2015, each will have until Sept. 18, 2015 to complete their design. One completed design submission will be selected as the winner for the grand prize: a Sony SW3 SmartWatch 3 SWR50.

If you have any questions while creating your design concept, be sure to head over to our TI E2E community, where a TI expert can help you out.

By entering this promotion, you represent that you are a product or software developer, and should you win this promotion, you will use the EVM only in a research and development setting to facilitate feasibility evaluation, experimentation, or scientific analysis and not for any consumer or household use or as a part or subassembly in any finished product.

Get the full contest details here. 

The evolution of Ethernet is fascinating. More than 40 years ago, Robert Metcalfe was asked to create a local area network (LAN) for the storied Palo Alto Research Center. The result of Metcalfe’s innovative work would be standardized as Ethernet – an adaptive technology that would forever revolutionize the world of communications. Fast-forward to modern society, and Ethernet is everywhere.

 What is Ethernet?

Today, many people think that Ethernet means the Internet. While these two concepts are indeed related, Ethernet is simply an interface specification (IEEE 802.3) comprising many subsections and specifications defining the physical and data-link layers of the Open Systems Interconnection (OSI) model. One of the most important pieces that came out of IEEE 802.3 is the Ethernet physical layer transistor (PHY).

 Figure 1 shows an example block diagram of how data is transferred to and from a standard RJ45 Ethernet cable to a processor.

Figure 1: Ethernet PHY system block diagram

 These are the three things you should know about Ethernet PHY:

1. It is a transceiver that is a bridge between the digital world – including processors, field-programmable gate arrays (FPGAs) and application-specific integrated circuits (ASICs) – and the analog world. An Ethernet PHY is designed to provide error-free transmission over a variety of media to reach distances that exceed 100m.
2. The Ethernet PHY is connected to a media access controller (MAC). The MAC is usually integrated into a processor, FPGA or ASIC and controls the data-link-layer portion of the OSI model. The media-independent interface (MII) defines the interface between the MAC and the PHY. Variations of the MII are available that provide minimal pin count and varied data rates depending on system requirements.
3. Speed matters. There are many different variants of the Ethernet standard, with corresponding PHYs, that range from 1Mbps to 100Gbps. The majority of Ethernet applications use a 10/100Mbps (see TI’s DP83848) or 10/100/1000Mbps PHY. The physical mediums that carry the data to the Ethernet PHY include twisted pairs, CAT5, coaxial cables, backplanes and fiber optics.

 Without Ethernet, you would not be able to quickly send data from one point to another. Forty years after its creation, Ethernet has become truly indispensable. What else would you like to know about Ethernet PHY? Leave a comment below. Watch next month for a blog post on how Ethernet technology is shifting new modern markets.

 Additional resources

• Find out more about TI’sDP83848 single-port 10/100Mbps Ethernet PHY.
• Learn about TI’s Ethernet portfolio.
• Check out this electromagnetic interference (EMI)/electromagnetic compatibility (EMC)-compliant industrial temp dual-port Gigabit Ethernet TI Design reference design.

In part 1 of this blog, I discussed the necessity of interleaving the four phases of a synchronous buck to minimize input/output voltage ripple and improve thermal performance. You can further enhance thermal performance by following some key layout guidelines to ensure that the power is dissipated evenly across all four phases.

All four phases must have the same exact L-C output filter; you should select these filters to minimize the DC resistance (DCR) (inductor) and equivalent series resistance (ESR) (capacitor) losses. The LM5119 will share current between the four phases evenly, provided that both the shared traces between the controllers and the feedback traces are noise-free. The four switch nodes are the noisiest on the board, and to avoid noise coupling must be avoided. You can accomplish this by routing noise-sensitive traces along the edge of the board. Figure 1 highlights how the compensation pins of both ICs are tied together using an internal layer (shown in yellow) along the edge of the board.

Figure 1: COMP pins of U1 and U2 tied together on an internal layer

Figure 2 shows the V_OUT trace being carefully routed along the edge of the board to avoid the switch nodes. If possible, you should route this trace on the same internal layer as the COMP pins and surround the trace with GND copper pours to ensure noise immunity.

Figure 2: Feedback trace routed along the edge of the board to avoid noisy switch nodes

The power stage must be laid out to ensure that the noise is contained while having enough copper to withstand the power requirement. Make sure that the input capacitors are as close as possible to the drain of the high-side MOSFET and the source of the low-side FET. This will minimize switch-node ringing. Next, minimize the switch node so that it is only as big as necessary to carry the high currents. Keep all of the power-stage components on the top side of the board to contain noise to one side, so that it is not communicated to other layers.

Figure 3 shows the top-layer power stage for one of the phases of the automotive multi-phase synchronous buck power module reference design (PMP10979). The power flow starts from the bottom with input capacitors (C6, C7), then to the high-side and low-side FETs (Q1, Q3), followed by the inductor (L1) and ending with the output capacitors(C17, C18).

Figure 3: Top-layer placement of power-stage components

I used four layers for this design, with 2oz of copper to ensure good heat transfer. The customer requirements entailed that I only mount components on the top side of the board. To reduce board size further, you can place the ICs on the bottom layer, directly under the two phases they control. This will also help to reduce the length of key traces such as feedback and COMP. The top layer should have most of your IC signals, as well as copper pours for switch nodes, input and output voltages, and GND. Reserve the second layer for GND only and run multiple vias to dissipate the low-side FET heat quickly and efficiently. Reserve the third layer for V_IN, V_OUT and any other IC signals. The fourth layer is all GND.

The overall benefit of interleaving the four phases and having an optimized layout is that it will increase the efficiency of the power supply. Figure 4 shows the high efficiency measured from PMP10979, with various input voltages.

The automotive and industrial industries are booming, and demands on the non-isolated synchronous bucks are becoming more strenuous. Using a multiphase design in order to reduce the losses will also reduce part counts, decrease the board size and provide better thermal performance.

Figure 4: Efficiency of PMP10979 with different input voltages

Additional resources: 

Read more Power Tips blogs

Unlike traditional pulse-width modulation (PWM) power converters, resonant converter output voltages are regulated by frequency modulation. Therefore, the design methodology of a resonant converter will be different from a PWM converter.

Among various types of resonant converters, the LLC series resonant converter (LLC-SRC) in Figure 1 attracts interest because of its better output regulation, lower circulating current and lower circuit cost.

Figure 1: LLC-SRC with AC input/output voltage

The series resonant characteristics allow the switching network in a DC/DC LLC-SRC, such as that shown in Figure 2, to have very wide region of zero-voltage switching (ZVS); hence, the LLC-SRC can easily achieve over 94% efficiency in front-end power-supply applications and operate at a high switching frequency.

Figure 2: LLC resonant half-bridge converter

Similar to the design process for PWM converters, the first step when designing an LLC-SRC is to select the desired operation frequency at full load. The remaining steps are different, because there is no duty-cycle factor in a resonant converter. The duty cycle remains unchanged in an LLC-SRC and is ideally 50%. Figure 3 shows a design flow chart for an LLC-SRC from TI Power Supply Design Seminar topic “Designing an LLC Resonant Half-Bridge Power Converter.”

Figure 3: LLC resonant half-bridge converter design flow chart

Notice that M_g is the DC voltage gain, L_n is the ratio of L_m and L_r, and the quality factor is defined as Equation 1:

Also, f_n is the normalized frequency defined as f_n = f_sw/f_o, where

The gain curves in both the M_g/Q_e and M_g/f_n charts are derived from the LLC resonant tank shown in Figure 1, which is also a linearized circuit of a LLC resonant half-bridge converter.

Figure 3 provides a simple circuit parameter selection process of an LLC resonant half-bridge converter. By checking the f_{n_min}, f_{n_max} locations on the gain curves, you will be able to design a high-efficiency LLC resonant half-bridge converter with ZVS on the switching network under all input conditions.

When designing a LLC resonant half-bridge converter, keep in mind that:

• f_{n_min} needs to be above the ridges of the gain curves in the M_g/f_n chart at all times. This is to make sure that the MOSFETs maintain ZVS.
• LLC-SRC efficiency can only be optimized at one operation point. When f_sw = f_o, the series L_r and C_r become zero impedance (Figure 4); the converter has best efficiency at that point. You will need to decide the line/load condition that you want to optimize and make sure that your switching frequency is at resonant frequency at that condition.

Figure 4: LLC-SRC with AC input/output voltage when f_sw = f_o

Additional resources

• Find more information about LLC-SRC design and optimization in the TI Power Supply Design Seminars papers.
• Read my previous blogs on LLC converters
  • • Get to know LLC series resonant converter design
    • How much can a LLC series resonant converter do?

Lack of wind at the beach may not mean much to many of us, but to a kite boarder, it means their drive to the beach was all for nothing. But with the help of a GSM weather station logging and transmitting real-time data and a few kite board lovers at THEWINDOP.com that trip to the beach will not be a waste of time anymore. The founders of THEWINDOP wanted a source for real-time high-resolution wind speeds directly from the beach to assess if kiteboarding was possible.

The weather station they created is a cost-optimized solution for collecting wind, temperature and humidity data down to a time period of a few seconds. This is done without external power (leverages a solar panel and battery) and can be used worldwide by leveraging a cellular modem (GPRS).

So how much data has been collected? Over a two-year period, it has logged more than 11 million data points. This is possible due to the high write endurance of the on-board MSP430FR5969 FRAM microcontroller. The on-chip AES could also be used to encrypt the data on-the-fly for wireless transmission.

“The FRAM series allows us to achieve ultra-low-power as well as simplify data buffering in our firmware… We have had a unit up and running in St. Andrews [Scotland] since 2013” - Andy Maginnis, co-founder of THEWINDOP.

To learn about other unique FRAM use-cases, or how it could improve your designs, head over to ti.com/fram. 

Smart meters in a smart grid have the ability to measure, control and communicate in one single device.  For those battery powered smart devices like flow meters, electrical energy is precious to keeping them “smart”. Having a smarter design to lower power consumption can keep them alive for a longer time.

The extended scan interface (ESI) is a module for the ultra-low-power MSP430FR69X microcontroller (MCU) which consists of multiple sub-modules, such as analog front ends and different state machines. Those modules automate sensor measurements without CPU intervention to minimize power consumption of the system.

Sensor circuit design also takes a big role to reduce system power consumption. Many sensors have an enable pin for low power operation. The enable pin is typically controlled by the MCU. If the measurement requires high sample rate, the MCU needs to wake up often, which increases power consumption.

The excitation circuit and the sampling and hold circuit of the ESI have the characteristic that the ESICHx pins or the ESICOM pin can be configured to connect to the ground during measurements. The connection is controlled by the timing state machine of the ESI. The ESICHx pins or the ESICOM pin can be used as enable pins to activate the sensors without MCU intervention. Therefore, the MCU hands-off the measurement process completely to the ESI and goes into sleep mode for saving power.

Figure 1 - Sensor signal path and the pins that can be act as enable pins

The TI Water Meter Reference Design for GMR Sensors, using extended scan interface (ESI) (TIDM-GMR-WATERMTR) shows an example to reduce the power consumption of the GMR sensor circuit. GMR sensors are typically connected to the power supply and the ground of the system. This connection causes a constant current flow that wastes energy as the measurement only takes a very short period of time. By connecting the GMR sensor to the ESICOM instead of system ground, the ESICOM acts as an enable pin. The ESICOM connects to the ground to turn on the GMR sensor only when the measurement is in process. When the system is idle, the ESICOM disconnects the ground so that the GMR sensor consumes zero power.

The TI Water Meter Reference Design for Optical Sensors, Using Extended Scan Interface (ESI) (TIDM-OPTICALWATERMTR) shows an example to use ESICHx pins as enable pins. This configuration enables multiple sensors to turn on individually during the measurement. If higher current is required to drive the sensor, a simple transistor circuit can be used to support the sensor.

Figure 2 - EN signal path

More information can be found here: 

• TIDM-GMR-WATERMTR
• TIDM-OPTICALWATERMTR
• TI’s smart metering solutions

TIers put our company’s commitment to citizenship into action every day in their jobs and in their communities. In our ongoing series “CiTIzenship," we feature TIers around the world making a difference in their communities and creating technology for good.

After watching a friend’s deep grief over the death of her infant son, Si Diep felt he needed to do something. So he completed a triathlon to raise money for cancer research.

“The baby was only a few months old when he was diagnosed with Leukemia,” Si said. “He spent most of his time in the hospital. The cancer was not only killing him, but it also affected his family. I wanted to raise money for cancer research, and putting my body through endurance sports seemed like a good idea.”

Si, an electrical engineer in our Santa Clara, Calif. office, had never competed in endurance sports, so he signed up with The Leukemia & Lymphoma Society’s Team in Training, a sports-training program that raises money for blood-cancer research. His initial goal was an Olympic-length triathlon, which includes a 1.5-kilometer swim, 40-kilometer bike ride and 10-kilometer run.

He finished that first triathlon in 2013 and raised about $6,000 for Leukemia research.

But he was just getting started. Since his first race, Si has completed a half-length triathlon, run a marathon and participated in a 100-mile bike race to raise money for charities. He served as a Team-in-Training captain to help people just starting to prepare for a race. Now he’s preparing for a full-length Ironman triathlon – a 2.4-mile swim, 112-mile bike ride and 26.2-mile marathon.

“People donate to these causes year after year, and even though I don’t have to, I want to push myself harder and harder every year to justify that,” he said. “I have to push myself because I made a commitment. I’m doing this for a cause. I cannot let my friends and family down. I have to finish what I started.”

Si’s involvement in worthwhile causes extends beyond endurance sports. With his wife’s encouragement, he has returned to his native Vietnam twice to build homes with Habitat for Humanity’s Global Village program. In addition, he serves as a coordinator for the Family Giving Tree that TI sponsors every Christmas in Santa Clara and he promotes TI’s annual United Way campaign with his fellow employees in the Bay Area.

For these efforts, Si was among 12 individual employees and three employee teams recently honored with TI Founders Community Service Awards for outstanding volunteerism and contributions toward building strong communities.

“Training for these races requires a lot of time and energy,” Si said. “However, it’s nothing compared to what a cancer patient goes through during treatment. When I hit a rough stretch during my training, I think of these guys … But community service doesn’t need to involve anything big. Just an hour here or there is enough to help out others.”

In previous installments of this series, I discussed implementation alternatives and how those decisions affect and are affected by device parameters. In this post, I’ll explain how device parameters as well as system factors can affect the achievable accuracy.

 To understand how accurate a measurement we can achieve, we first need to understand the sources of potential errors. Here is a list of two types of error sources – note that this is not an exhaustive list, but rather highlights some of the main sources:

 Amplifier-related error (noise) sources:

• Input offset voltage (V_OS).
• V_OS drift.
• Common-mode rejection ratio (CMRR).
• Power-supply rejection ratio (PSRR).
• Gain error.
• Gain-error drift.

 Nonamplifier error sources:

• Printed circuit board (PCB) layout.
• Shunt-resistor tolerance and drift.
• Gain-setting passives tolerance, matching and drift.

 The worst-case accuracy is a simple linear summation of all error sources, e, shown in Equation 1:

                       (1)

 It is statistically improbable that all errors are maxed at the same time, so a more probable accuracy equation would be a root-sum-square of the error sources, shown in Equation 2:

   (2)

 First, let’s consider the external error sources. The V_SENSE that the amplifier sees is what is across the input pins, which will be different than what a designer may experience if measuring across the resistor. Ideally V_SHUNT = V_SENSE, but parasitics and trace etch will cause this not to be true. A Kelvin connection will minimize the PCB parasitic-error contribution.

 When it comes to shunt-resistor tolerance and drift, there will be a trade-off between the cost you are willing to pay and the required performance for the application. The shunt-resistor temperature drift is one of the main error sources over temperature, unless you choose a costly resistor. High-precision (0.1%), low-drift (50ppm/°C) resistors on catalog distribution sites cost somewhere between US$4.00 and US$8.00 in 1,000-unit quantities.

The error contribution of an operational amplifier’s external circuitry is eliminated using a current-shunt monitor due to its integrated, matched, low-drift gain-resistor networks. The effects of these on-chip networks are factored into the error contributions of the current-sense amplifier itself.

 When it comes to amplifier error sources, you will need to make similar trade-offs on precision versus cost. It is important to review the operating conditions of each of the parametric specifications listed in the datasheet and compare those to your actual operating conditions. For this discussion, I will focus on two of the main error contributors: input offset voltage and gain error.

 The input offset voltage will be the dominant error source at low V_SENSE levels. If you simplify and assume that V_SENSE = V_SHUNT = I_LOAD x R_SHUNT (no parasitic error contribution), then the error contribution of V_OS is calculable using Equation 3:

                                    (3)

 Let’s look at two different current-shunt monitors, the INA199 and INA210, to see how V_OS will affect the error. The INA199 datasheet lists the V_OS(MAX) as 150µV, while the INA210 is 35µV. Table 1 shows the respective error contributions using a current value of 1mA and a 1Ω shunt resistor.

Table 1: INA199 and INA210 current-shunt monitor error contributions

If you are measuring low current values that will result in low V_SENSE values, it is critical to minimize V_OS to minimize the error. As V_SENSE grows relative to V_OS(MAX), this error contribution is minimized. If V_SENSE is 1,000 times the value of V_OS, then the error contribution is 0.1%. On the other end of the spectrum is when V_SENSE is much greater than V_O. The main error contribution will be the gain error. In most datasheets, this is specified as a flat percentage, and the contribution is a straightforward percentage adder.

 Let’s see how these two errors contribute to a total error, again assuming that they are the only two error sources. Figure 1 shows how the linear sum as well as the root-sum-square method of error calculation work for the INA199 and INA210.

Figure 1: INA210 and INA199 total error

 Here, the offset voltage is the major error source when V_SENSE is low, and gain error dominates when V_SENSE is high relative to V_OS.

 My analysis has simplified the error calculations to two primary sources. However, current measurement accuracy is a very complex subject with many moving parts that need to be traded off against each other to maximize performance under specific operating conditions. One of the key elements for a more thorough analysis is temperature. The temperature drift will affect multiple specifications, including shunt-resistor value, offset voltage and gain error. Using a current-shunt monitor that has zero drift, including the INA210 or INA282, will help minimize the offset drift contribution.

 Additional resources

• Read the datasheets for the INA210, INA199 and INA282 current-shunt monitors.
• Read this “Current sensing tutorial” article series on EE Times.
• Watch our training series on current-sense amplifiers.
• Check out related TI Designs reference designs:
  • High-Voltage 12V-400V DC Current-Sense Reference Design.
  • Current-Shunt Monitor with Transient Robustness Reference Design.
  • EMC-Compliant High-Side Current-Sensing Reference Design with Overvoltage Protection.

Finally, we have arrived at the concluding entry in this "Understanding MOSFET Data Sheets" blog series attempting to demystify the power MOSFET data sheet. In this blog, we will take a look at some of the other miscellaneous switching parameters that appear in the MOSFET data sheet, and examine their relevance (or lack thereof) to overall device performance.

On the one hand, switching parameters like output charge (Q_OSS) and the reverse recovery charge of the FET’s intrinsic body diode (Q_rr) are crucial elements that are responsible for a great portion of the FET’s switching losses in many high frequency power supply applications. Sorry if it may be starting to sound like a broken record at this point, but designers need to be careful when comparing FETs based on these parameters, because as is so often the case, test conditions matter!

Figure 1 below shows the output charge and reverse recovery charge as two sides to the same coin, measured at two different rates of di/dt on TI’s CSD18531Q5A 60V MOSFET.  On the left,  Q_rr was measured at 360A/µs to be 85nC, and on the right, it was measured at 2000A/µs to be 146nC. While there is no industry standard for di/dt to measure the part at, we have seen competitors rate all the way down to 100A/us, in order to give the appearance of extra low Q_rr.

Figure 1: Q_rr and Q_OSS measured on the CSD18531Q5A at 360A/µs (left) and 2000A/µs (right).

Q_rr can have an even stronger dependence on the diode forward current (I_f) the test was conducted at. And even further complicating matters is that some vendors do not include Q_OSS as a separate parameter, but rather just absorb this into the specification of Q_rr. Besides the test conditions listed on the data sheet, other considerations like parasitic board inductances and subjective measurement methodologies make it virtually impossible to compare these parameters from separate vendors’ data sheets. That is not to say they are not important parameters to consider and design around, but for reliable comparative data, the only effective solution is to collect it independently using a common methodology and board.

The last parameters I will mention in this series are switching times. These four parameters are defined generally by the waveform below in Figure 2 and appear on virtually every vendor’s data sheet. They are so dependent on board and test conditions that one veteran in the FET industry (and personal mentor) often cites these as “the most useless parameters on the FET data sheet.”  Meant to give an indication of switching speed, the reality is these can be just as much a reflection of driver strength and drain current as they are the FET characteristics. TI includes these parameters as tested at the device’s rated current, while others will test these at only 1A I_D, to give the appearance of a faster switching device. Much more indicative of the device’s actual switching speed are the gate charge parameters and the internal gate resistance of the device, R_g, both of which are significantly less susceptible to these specmanship games.  

Figure 2: Waveform defining the MOSFET data sheet switching times.

Thank you for taking the time to read this series on MOSFET data sheets. I hope you found this an enlightening read, and walk away with a more clear understanding of the value and ambiguities of the parameters that appear on the power MOSFET data sheet. Feel free to share the entire series with anyone who might benefit from this perspective on how to read a MOSFET data sheet. And don't forget to watch the video "NexFET™:Lowest Rdson 80 and 100V TO-220 MOSFETs in the World" and consider one of TI’s NexFET power MOSFET products for your next design.

The proliferation of applications needing momentary backup power has contributed to an increase in demand for supercapacitors. Supercapacitors, also known as ultracapacitors, are electrochemical capacitors with the ability to store more energy than regular capacitors. They can be charged and provide energy much faster than batteries. Figure 1 compares conventional capacitors, supercapacitors, conventional batteries and fuel cells in terms of power and energy density.

Figure 1: Energy versus power density for different energy-storage devices

A significant advantage of supercapacitors is their ability to withstand thousands of cycles before degrading, compared to hundreds of cycles for batteries.  Also, Supercapacitors have the ability to be deeply discharged when compared to batteries as shown in Figure 2. However, because of the decomposition voltage of the electrolyte, most supercapacitors have a maximum rating of 2.7V-3V. Figure 2 compares the charge/discharge profiles of supercapacitors and batteries.

Figure 2: Charge/discharge cycles of supercapacitors and batteries

Recent developments in supercapacitors have led to the introduction of lithium-ion hybrid capacitors chargeable to a higher voltage (up to 4V) with less self-discharge and thus greater energy density. A disadvantage with these kinds of supercapacitors is their inability to be discharged to lower than about 2.2V without becoming damaged.

The composition of supercapacitors makes their self-discharge rate significantly higher than batteries. The greater the operating temperature under which a supercapacitor operates and the higher the voltage to which they’re charged, the faster the aging process; the capacitance of the supercapacitor decreases and the equivalent series resistance (ESR) increases. This means that the amount of energy that the supercapacitor can provide for an application diminishes. Equation 1 expresses the energy of a supercapacitor as:

                         (1)

W is the energy provided by the supercapacitor, C is the capacitance of the supercapacitor and V is the voltage of the supercapacitor. The ESR of the capacitor contributes to the power losses of the system.

Figure 3 shows the effect of temperature and voltage on the aging of a supercapacitor. Just a 10° temperature increase can cut the life expectancy of a supercapacitor by half. Also, it’s common practice to charge the supercapacitor to a voltage lower than the nominal to increase its longevity.

Figure 3: Life span versus cell voltage of supercapacitors at different temperatures

Since the maximum voltage to which a supercapacitor can be charged is between 2.7V and 3V, it is necessary to connect several supercapacitors in series for most applications. Thus, you must balance the supercapacitors; otherwise one unit could be charged more than the other and lead to unequal capacitor aging, subsequently reducing the pack’s ability to provide the required energy for the application.

Among the several methods employed in balancing supercapacitors are passive balancing using resistor strings, using switched resistors, using Zener diodes and active balancing. The first three methods result in power losses in the resistors, while the fourth method is the most efficient but also the costliest.

When supercapacitors are used in backup power applications, you must monitor their capacitance and ESR to ensure that they can provide the minimum required energy the application demands. TI’s bq33100 is a supercapacitor manager that can keep track of the capacitance and ESR of the supercapacitor bank; it also monitors other vital statistics such as temperature, voltage, current and safety conditions. The bq33100 can notify the host if the capacitor bank is weak and also carry out balancing. The bq24640, a high-efficiency switch-mode supercapacitor charger, has the ability to charge capacitor banks from 2.1V to 26V.

TI’s supercapacitor manager reference design (TIDA-00258) is a system solution for evaluating and monitoring supercapacitors. It incorporates a linear charger reference design and the bq33100 super capacitor manager. Next time you are evaluating a back-up power supply for your design, consider the power of supercapacitors!

Wouldn’t it be much easier if all devices in an electronic design operated from the same supply voltage? Unfortunately, not all functions have the same supply-voltage requirements, requiring the generation of multiple supply rails within a given device design. You may need multiple supply voltages, even for an IC like a high-performance data converter.

If a voltage rail does not need to be particularly efficient, requires a low load current or just needs to be the very lowest cost, then a simple low-dropout regulator (LDO) is usually the best approach.

If a voltage rail requires a very high load current (say, greater than a few hundred milliamps), or needs low output ripple, consider an inductive solution, with its high efficiency.

But for solutions requiring relatively low load currents and efficiencies higher than LDOs, plus cost/size constraints, a charge-pump DC/DC device is worth considering.

Figure 1: Example of a charge pump (12mm²) versus an inductive solution (29mm²): V_IN = 3.0V to 4.5V and V_OUT = 5V at 150mA

Charge-pump DC/DC devices enable you to generate different output voltages (boost/buck/inverted) from a given input voltage. These devices only require a flying capacitor to support the charge pump itself and decoupling capacitors for input- and output-voltage pins. Figure 1 above highlights an example where an application needing a Vout of 5V with a load of 150mA compares between a charge pump boost and an inductive boost – a greater than 30% solution size saving.

Table 1 summarizes the different voltage-conversion methods.

Table 1: Design criteria comparisons between DCDC converter approaches 

A charge-pump solution can be useful if you need a single special-voltage rail for a specific low-load device within a system. In such a system, an inductive DC/DC usually supports the overall system rail given the higher load requirement. A switch-capacitor DC/DC generates the specific light-load rail – be it a boost, a buck or an inverted voltage from the main system rail – and does so with only the use of capacitors, with reasonable efficiency, in a small solution size and very cost-effectively.

Additional resources

• Learn more about “The Forgotten Converter” in this white paper.
• Read this application note about how the TPS3619 charge pump with MSP430 microcontrollers can reduce power consumption.
• Find help deciding between a charge pump, inductor-based converter or LDO.

The publication Planet Analog picked up one of my blog posts for publication to their readers. I invite you to go to their site for the full post. 

To complement the 2-wire and 3-wire sensor transmitter blogs that Kevin Duke and I have already published on Precision Hub, my next few blog topics will be on 4-wire sensor transmitters. The blogs will explain 4-wire transmitter basics, the circuit structure of 4-wire output stages, and provide examples of each isolation scheme available with 4-wire transmitters.

To continue reading this post, check it out on Planet Analog! 

As a long-term member of the WEBENCH® design development team and previous product line applications engineer, I have been working for many years to bring power supply design tools into the hands of the novice and time-pressed engineer.  Our goal has always been to give designers the tools to automate time consuming tasks and help get you to market faster.  While I think we have definitely created a very valuable tool for the budding engineer, we have recognized that there are a whole lot of folks that are well-versed in power supply design who don't need entry-level tools, but do need more control over the design and can benefit from some of the same automated tools.  The new WEBENCH Advanced Tools we have just added are some of the most challenging to develop so far, but now WEBENCH Power Designer offers more advanced design control for the expert designer and a simulation export function that gets even the most complex designs into your CAD tool.

Advanced Options lets you target and control the solutions presented for your requirements.  With over 1,500 possible switcher and LDO IC + topology combinations supported by WEBENCH Power Designer, there are often dozens of candidate solutions that match your required input voltage range and load voltage/current.  To get to a short list of solutions that are an even better fit, use the Advanced Options to specify regulator design targets and component selection.

For example:

• Limit the height or size of external components.

• Specify ceramic-only capacitors.

• Customize the sorting of the solutions.

• Feature filters include IC features, control mode, supported WEBENCH tools.

• And more!

Figure 1: Advanced Options are used to specify regulator design targets. (Selections shown for example.)

Figure 2: Feature Filters narrow the solutions further. (Selections shown for example.)

Compensation Designer helps you adjust the power supply frequency response for optimum stability.  You can:

• Target a specific phase margin or circuit bandwidth.

• Fix a low phase margin that may result from changes in critical components.

• Adjust compensation to speed up load transient response.

WEBENCH power designs are set up to have a phase margin between 35 and 90 degrees, ideally 45 to 50 degrees, as this gives a stable transient response. In some designs, if you change a critical component such as the inductor or output capacitor, the phase margin may be reduced to a point where the response is no longer well-damped. Or, when the phase margin is more than 50 degrees, the transient response can be too slow. In either case, Compensation Designer can be used to tune the compensation for fast, stable response.

Compensation Designer offers four types of control:  target of phase margin and crossover frequency, manual control of poles and zeroes, ability to specify compensation components and automatic recalculation of compensation (“fix it for me”).

Figure 3: Using Compensation Designer, a large phase margin is reduced and the circuit bandwidth is increased, improving the regulator’s transient response.

Simulation Export provides a way to bring your power supply design into your local electrical simulation environment. This allows you to merge your power supply design with other circuitry in your application. This feature is also useful for simulating complex circuit models that would take too long to simulate in WEBENCH exceeding the WEBENCH simulation time limits.

Export your design into your local Altium, OrCAD, or TINA-TI environment where available.  The schematic and simulation parameters are pre-set to the startup simulation of the online WEBENCH design, so you can simply run the simulation as-is.  You can also modify the netlist, to add components related to connecting the power supply into your larger design, or for diagnostics such as transient response.

Figure 4: Altium Designer supports the WEBENCH Connector, which includes the ability to open an exported WEBENCH Design and simulate it using TI’s WEBENCH Simulation Engine.

Figure 5: The WEBENCH Power Design schematic opens in Altium Designer, pre-set to run a startup simulation.

These three new advanced tools in WEBENCH now make the process of power design simple for even the most complex systems. Give them a try for your next power supply design!