Designing for Portable Medical Electronics
Conflicting requirements for portable medical devices constantly challenge design engineers. Systems need to be high-performance but cost-effective, compact yet redundant, robust and efficient. Above all, they need to be reliable. Designing a proper system is challenging, even for experienced power engineers.
Medical device teams often focus on core technology rather than support systems. For example, they may be experts in ultrasound imaging or glucose metering but have minimal experience in power design. This can lead to basic mistakes that impact the performance, lifetime, efficiency and cost of the final product. To help teams avoid this problem, we’ve compiled some of the best design principles as outlined by our top power designers. Learn tips, understand potential pitfalls and know when to seek help.
Designing a power circuit for a single chip is a fairly straightforward process. The challenge is that most medical devices are complex and integrate multiple processors, display backlights, motors, transducers and even modems and transmitters. As a result, media devices often require a range of voltages.
“If you’ve got a 3.3-V supply and you need 2.5 V, but it’s only driving 1 mA or 100 μA, you can just drop a little LDO regulator on there, which is basically not going to burn any appreciable amount of heat,” says Eric Kwok, vice president of engineering at electronic design and manufacturing services provider Axelsys Engineering. “It’s not worthwhile to add a whole separate DC to DC converter stage just for that supply.”
The problem with linear regulators is that they dissipate excess voltage as heat, and wasted heat equals wasted energy. A constant-resistance regulator might support a standard battery life of 15 hours. In contrast, a constant-power regulator can provide nearly 27 hours, resulting in an 80 percent power increase from simply matching the regulator to the battery.
In such cases, a switched-mode power supply may provide a better solution. As its name suggests, a switched-mode power supply turns on and off at a set frequency, adjusting the duty cycle to maintain output voltage at the desired level. This switching process introduces a saw tooth ripple in the output voltage. However, this can present a problem in devices with high-sensitivity sensors. External compensation circuitry typically includes an inductor to smooth out current variations and a capacitor to dampen voltage changes. The larger the capacitor and inductor, the lower the ripple. However, the trade-off is size and cost.
Switching regulators can be classified as step down (buck), step up (boost), or step up/step down (buck-boost). Buck regulators are simple to implement and provide stable performance at a low price. However, to use them, all voltage rails need to be lower than the source voltage. If the voltage drops are small, a buck switching regulator can provide efficient performance. But efficiency drops as voltage step-down increases.
Boost switching regulators store charge in a capacitor to provide stepped-up output voltage. Leveraging a boost regulator requires a source voltage below that of the system’s voltage rails. Although it’s a more expensive and complex approach, boost switching regulators can be an effective solution for a tightly grouped set of voltages that are higher than the output voltage of a standard battery pack but substantially less than the output voltage of multiple packs linked in a series.
Buck-boost switching regulators provide the best overall solution because they step voltage up or down as required. However, buck-boosting switching regulators are more complicated devices, so the trade-off is cost and difficulty of implementation.
Delivering multiple voltages with multiple switches requires implementing compensation circuitry for each, increasing the complexity, footprint, component count and system design and assembly time. Multi-output switching regulators perform well and simplify implementation. These devices use one controller IC with a single quiescent current to deliver multiple voltages in a smaller footprint. “You still have to have input and output capacitance, but it’s easier than using a separate chip for each of those outputs,” says Kwok. Some devices include inductors and capacitors, but the increased integration means more noise and limited performance.
Working With Batteries
Portable medical devices are often powered by three sources – wall plug, battery and a backup battery (i.e. button cell) – and automatically switch among power sources, as needed. For the switching process designers sometimes incorporate diodes as they are economical solutions but they can also introduce losses as high as 0.7 V, which can be a problem for a device running on a button cell that only provides a few volts to begin with. Metal oxide semiconductor field-effect transistors (MOSFETs) post much lower conduction losses, making them a better choice for low-voltage applications.
Input voltage changes over the lifetime of a charge with a battery-powered system. For example, a 9 V battery might drop down to as low as 7 V by the end of its cycle. “Electronics engineers are used to thinking, ‘I have this voltage coming in, I need this voltage coming out,’ but the voltage is not fixed, it’s changing,” says Steve Roberts, director of engineering in Europe for RECOM Power Inc. “You have to take that into account.”
In the case of a AA battery, which decays over time from 1.5 V to perhaps 0.8 V at discharge, a low input boost regulator, such as the SC120 from Semtech Corp., offers an ideal solution. The SC120 provides voltages ranging from 1.8 to 5 V and up to 1.2 A of output current. More important, it requires 0.85 V during startup but only 0.7 V during shutdown, allowing it to function well, even at the end of battery life.
Figure 1: The SC120 buck regulator requires a minimum voltage of only 0.7 V, making it well suited for AA batteries, which drop as low as 0.8 V.
Battery chemistry adds to the challenge. Certain types of batteries perform best with fast, high-current discharge followed by long recovery rather than continuous discharge. When matched to a pulsed application, these batteries can have extraordinarily long lifetimes. For example, if a device is monitoring a slowly changing metric such as temperature, the unit can go into standby mode in between measurements. For example, the R-78AA DC/DC switching regulator from RECOM can be used with pulsed systems to support 20 µA standby, using either a microprocessor or low-power timer. As part of the R-78xx line, the R-78AA delivers conversion efficiencies as high as 97 percent. Standby can dramatically change discharge time and spreading out the load in pulses can increase life for a lead acid battery by 30 or 40 percent.
Figure 2: The R-78AA switching regulator operates at 97-percent efficiency with a 20 μA standby mode.
Integrating power design into the development process is a key to success. All too often, power design is set aside while a design team focuses on the company’s core differentiating technology.
“Often, people who are experts in another area say, ‘Okay, throw a power supply on here,” says Kwok. “They do a rush calculation and overdesign it by 50 percent or 100 percent. They end up with a power supply that’s twice as big as it needs to be and twice as costly. It’s better to invest time up front putting together a power budget that’s as accurate as possible, refining it as the design progresses.”
“Such a budget should not only list components, voltages, and currents but also tolerances,” says Kwok. You have to look very carefully at the data sheets. The standard accuracy spec is ±5 percent, but I’ve worked with designs where it’s actually ±4 or ±3 percent. If there’s a transient load on that supply because you’re using an FPGA or some kind of processor that draws more current, all of a sudden you’re going to see a dip in your voltage and really strange behavior with your digital circuitry. It tends to be very difficult to debug because of the transient condition.”
In addition to considering tolerance, a designer should take a systems-level approach to functionality. Medical devices tend to suffer from design creep – added features and functionalities – because the devices are designed for use by a wide range of operators from surgeons to patients. Such design changes may add value, but a designer should consider whether they add enough value to merit the cost.
“There has to be somebody who’s looking at it and saying, ‘If we make this change, this improvement, how are we going to power it?’” says Matthew Borne, product marketing engineer at Texas Instruments Inc. “If we add this RF amplifier, it’s going to be right in the middle of our battery voltage, so we know we’ll need a buck-boost regulator on it. Adding the additional feature is an efficiency improvement from a performance standpoint, but it’s not necessarily improving efficiency from a power delivery standpoint.”
Perhaps the single biggest challenge in power design is layout. Data sheets list specifications and additional components required. They may also include sample circuits or application notes. Designing circuits and components into the overall layout can be far more difficult than it appears. “[The challenge] is everything beyond the datasheet,” says Borne. “It’s being able to lay the part out on a board and deal with all the parasitic effects and current loops and noise. We have customers with absolutely no power-management experience who come back with problems and we find that they didn’t lay out their board correctly or there’s a trace on the board that’s too close and they’re getting some crosstalk. When customers choose a part, they have to make that part fit onto a board and interact with everything else in the system, and that requires quite a skill set.”
Portable devices lack a true ground. Components need a solid, low-impedance connection to the zero voltage of the battery. The process is as much art as it is science. For example, a via can be well behaved at low frequencies but may act like an inductor at higher frequencies. The trick is not to use one via, but to use five or 10 that all connect between the same layers. This approach provides an efficient connection across all frequency bands.
Designers should also keep current loops small. Make sure the grounding for a high current path on the board doesn’t touch the ground of an analog system or any noise-sensitive components. Dithering and multiphase switching are two methods for reducing noise. Dithering, or slightly varying the switching frequency, replaces a large, single-frequency noise spike with a lower amplitude noise band. Multiphase switching involves adjusting groups of converters so they switch out of phase. For example, TI’s TPS40140 buck controller can deliver two-phase output separated by 180 degrees to reduce ripple. Up to seven of the chips can be stacked together to provide programmable 16-phase output with switching frequencies as high as 1 MHz. The devices are compatible with TI’s NexFET technology, which cuts the gate voltage in half while providing the same resistance, allowing it to deliver 90 percent efficiency with double the frequency.
Figure 3: Stacking seven TPS40140 chips provides 16-channel, interleaved output to minimize ripple.
A current, interesting trend is power scavenging, or designing systems that run on minimal power and use alternative power sources such as, a solar panel, motion-generated power source or fuel cell. Leveraging the power benefits of load pulsing, a micro power device could store charge from a movement generator in a super capacitor and periodically power up, take a measurement, transmit the data and shut down again.
The approach is still in early development, Roberts says. “There are very few really good solutions yet, but I expect to see more of these power scavenging ICs or regulators and things like that in the next year or so.”
Informed power management can significantly enhance portable medical devices, not only to increase battery life but also to ensure stable, low-noise performance. Following best practice design principles helps achieve the most effective power management. For devices requiring many different voltages, a reputable design services team can help deliver a fast solution and can potentially deliver one that is more efficient and cost effective. Whether a designer employs a reputable design services team or designs a power management system alone, it’s important to ‘design in’ power from the very beginning of a project. Thinking about power early and often will help a designer rapidly deliver an effective product to market.
Original article was written by Kristin Lewotsky and may be found on the Digikey website. Reposting of this article was approved by Digkey Corporation.