Improving Reliability and Performance of Medical Designs with Smart MOSFETs

All products for medical applications require high reliability while still providing the new technologies and features that end users want. As competition between medical device companies and their end applications increases, features are increasing dramatically, but another factor that may cause product failure is not considered. All of these factors are related to power, and it is important that we use the latest technology to minimize the risk.

Smart MOSFETs are one of these enablers and are increasing in popularity. Standard P-channel FETs are often used to switch power distribution nodes, connect charging paths, hot-swap connectors, DC current, and more due to their simple drive requirements. Because these components are in the critical path, their failure can render downstream sensors or processors inoperable, so investing in reliable power switches is a wise move. Compared to the equivalent P-channel/N-channel combination approach, Intellimax FETs integrate P-channel FETs and logic-level drivers to simply control this reduced Rdson FET. For increased reliability, these components integrate ESD protection, thermal protection, overcurrent protection, overvoltage protection, and reverse current blocking. All of this brings higher value and higher reliability to medical applications.

The following article will introduce the technology of load switches and the reasons why they exist in current power architectures. Its application cases will be presented in the laboratory. We will discuss applications less than 6V, which rechargeable portable medical applications should benefit from. This article will also discuss the new 40V smart FET application enabled by the latest technological advances of Quick Semiconductor, and will provide valuable analysis results to show how smart FETs have become a trend in the intelligent development of the medical industry.

The Evolution of Load Switches in Battery Applications

Since batteries were introduced into electronic products, the need for power isolation has always existed. Introducing batteries as a mobile power source means that the battery will be constantly charged and discharged during use. Obviously, the energy-saving characteristics of the design will directly affect the time between normal use and charging. In recent years, battery technology has not seen any significant improvements, and there are no major breakthroughs in the future. Therefore, it is necessary to rely on integrated circuit (IC) technology to comply with strict power consumption specifications to extend the working time of the device.

Before we discuss load switches, we need to review battery technology, the load on the battery, and the requirements for load switches. Under fixed charging conditions, estimating battery life can be relatively simple if all current consumption paths are known. The general situation is that it is not the controlled duty cycle sensor with a current of 100mA that alone contributes to the power consumption, but rather many always-connected sinks of less than 1mA that slowly consume energy. These sinks must be roughly added to the power equation, however, which is more difficult because transient spikes occur when a given function or sensor is enabled. The amplitude and period of these spikes are monitored and used for the energy calculation, which is usually a single peak result multiplied by the number of spikes.

Once all the typical loads are known, it is straightforward to calculate the operating time. Batteries are now measured in mAh rather than coulombs, meaning that a 1000mAh battery can deliver 1A for one hour or 100mA for 10 hours at its nominal battery voltage.

Battery operating time (h) = battery rating (mAh) / total current consumption (mA)

When the operating current is divided into 100ms of operation at the surge current (eg 1500mA) and the remaining time at the continuous current (eg 20mA LED indicator), the average current during this time can be calculated linearly.

Average current per hour = (1.5A×0.100s∕3600s) + (0.020A×3599.9s∕3600s) = 20.04mA

With this concept of power consumption in this time domain, it can be quickly understood that load switches can be used to isolate continuous, but smaller current draws. Short-duration spikes are not the culprit, and if not isolated, hundreds of uA current draws can add up to mA levels. This transition brings up the importance of soft power ramps, especially when power is used by downstream ICs to reduce unwanted large voltage spikes on fragile mAh battery ratings.

The effects of surge and steady power consumption can be discussed separately. The effects on the battery vary greatly depending on the battery chemistry and the time between surges and power consumption. A general idea is that a reasonably proportioned surge can result in longer battery life than a light, continuous load. For specifics on this, please consult the battery supplier. The voltage drop of the battery pack as the power is consumed is also not discussed. In the above formula based on pure current, we assume that the voltage Vbatt is constant. Again, this depends on the technology used in the battery. For alkaline primary batteries (non-rechargeable), Vmax is 1.5V, and in most cases Vmin is assumed to be 0.9V here. Rechargeable single-cell Li-ion batteries have a nominal state voltage of 3.7V, but can be charged to a maximum of 4.2V and still drop to a minimum voltage Vmin of 2.5 to 3V, which has a greater impact on actual charging.

Understanding how actual current consumption depletes the battery level, we can now look at different ways to isolate downstream consumption. Terms such as high side and low side switches will be used. High side means that the switch will be in the operating level (rail) circuit and the current actually flows from the source to the load and returns through the ground circuit. Low side switches are on the opposite side of the load and allow current to flow to the ground circuit.

Applying this simple switch schematic to common FET types, Figure 1 shows how basic N-channel and P-channel MOSFETs perform for load isolation, each with its own advantages and disadvantages. Starting with the PN junction cross-section image, we can quickly illustrate that cross-section b is like a P-channel on the high side. The N-channel is used to drive the gate to simplify the logic input control. The disadvantage of schematic b is that if the load voltage is higher than the battery voltage, it can forward bias the body diode. Schematic c solves this disadvantage by using dual P-channel FETs on the high side, which is a very common battery isolation method for the main level.

Why can't N-channel FETs be used for high-side switches? The textbook characteristic of N-channel FETs is the ability to start the switch and keep it in the linear region. According to the threshold voltage in the datasheet, the gate voltage must exceed the drain voltage. Because the main level in battery applications is usually the highest level available, a bootstrap or isolated drive method must be adopted. This will bring additional costs, however, this N-channel high-side switch method is necessary for higher current applications. Depending on the voltage range, the Rdson of the N-channel can be reduced by 20-50%. In addition to the losses caused by Rdson, higher voltages, that is, above 200V, make P-channel FETs either expensive or completely unavailable due to technological limitations.

picture

Smart MOSFET Technology

For most applications, conventional load switches are effective, but this article will focus on medical applications only. These devices require extremely high reliability and are not rechargeable in most cases, so power consumption and isolation must be carefully studied.

Fairchild Semiconductor’s Intellimax portfolio meets the functional requirements of smart MOSFETs. Figure 2 shows its standard internal block diagram, although it will vary from device to device based on the required features. This diagram is based on a P-channel, with the high-side circuitry located between Vin and Vout. The number of pins has been minimized to keep the package size as small as possible. When it comes to packaging, these devices can be packaged in chip scale packaging (CSP) as small as 1mm×1mm, or in the widely used leadless uPak package, also known as MLP. For prototype needs and designs with less space constraints, SC70, SOT23 and SO8 can also be used.

picture

The operating voltage Vin of smart MOSFETs varies depending on their manufacturing process. For the Intellimax product line from Fairchild Semiconductor, the recommended operating voltage range is from 0.8V to 5.5V. High voltage smart FETs will be discussed later in this article. It is important to note the difference between input voltage and control voltage. The input voltage Vin is the actual rating for the high-side load switch. The control voltage level marked as ON in Figure 2 is the voltage value required to turn on the load switch. Figure 3, taken from the Intellimax FPF1039 datasheet, shows the actual Von voltage required to turn on the integrated P-channel FET as it relates to the Vin supply voltage.

The datasheet specification adds a buffer for process, voltage, and temperature variations, indicating that Von must exceed 1.0V to turn the switch on and must be below 0.4V to turn the switch off. This results in a very simple drive circuit that can be directly connected to a microprocessor. This Von specification varies from part to part and may not necessarily be as flat as in Figure 3. Don't stop at the line in the datasheet that shows the static threshold level; refer to the curve for full details.

picture

As mentioned above, this logic level Von allows for easy functional interfacing to a microprocessor, but thermal shutdown and over current protection (OCP) also interface well via the Flag pin. This feature is not integrated in the smallest Intellimax solution such as the FPF1039, so we turned to the FPF2303. This dual output load switch is capable of driving 1.3A loads and has all the features mentioned previously, but also includes a Flag feature and reverse current blocking. Flag is an open drain logic level that can be directly connected to the status pin on the processor. Reverse current blocking is as shown in the traditional load switch diagram, but requires a dual MOSFET approach. Fairchild's proprietary approach integrates this into the P-channel and as an additional feature within the IC without external components. If a situation occurs where the load side of the switch is at a higher potential than the battery side, then a reverse current blocking feature is necessary. This can occur in a system with multiple batteries of the same initial voltage, or during voltage spikes. Bulk capacitors also tend to provide delta values.

For load switches, an often overlooked specification is the ESD rating, as most MOSFETs in the past did not have ESD protection integrated into them. More recently, ESD protection has been added to discrete P-channel MOSFETs, where they are simply used as cost-effective load switches. This comes in the form of back-to-back zener diode clamps on the FET gate. This increases the capacitance of the gate, making it a less likely candidate for switching applications (motor drives, power supplies, etc.), but makes the gate more robust with the addition of a 2K HBM (Human Body Model) Zener diode. Intellimax has even gone a step further and integrated ESD structures into the smart FETs to double the ESD rating to 4KV HBM. ESD can be further improved in the future. For medical applications, ESD is an important characteristic because boards are often shipped unpackaged between assembly rooms to complete placement in plastic cases and sealed enclosures. Each shipping point is a potential risk for ESD-related failures, especially when pins and connectors are connected from the board to the battery or interposer.

A feature of the next generation of smart FETs that we should delve further into is what happens when the switch is turned off? Conventional load switches using discrete P-channels can be completely off and connect the input to the output, regardless of heavy loads or large capacitance on the output pin. If this happens, the primary input level will typically show a voltage sag, which can affect precision analog-to-digital converters (ADCs) or sensors associated with bias levels. In the past, resistor/capacitor (R/C) networks were added to the gate to slow down the turn-on speed, but this increases the design time and size of the project. Intellimax supports a slew rate control feature that minimizes level disruption by limiting the inrush current at the input. Figure 4 shows an example of this approach in a lab test for an empirical study. Note the effect on Vin level using the traditional P-channel approach on the left and the effect on the Intellimax device on the right.

picture

Smart MOSFET increases reliability

The requirement to disconnect the load from the input to prevent further damage when an adverse event occurs is a major concern in solving reliability issues. Traditional load switches in the past have been very simple and did not provide current protection or thermal protection. Current protection can be added, but this will add some external components and require more precise selection tolerances for passive components. In summary, can the passive approach react in a short enough time to prevent downstream damage? Thermal sensing is applied on a similar comparison basis.

The details of overcurrent and overtemperature shutdown events vary from device to device. While some shutdowns are immediate and require a power cycle to reconnect to the load, others go through a retry mode that continually attempts to reconnect once it is confident that the temperature and current levels are safe. A careful review of the datasheet can eliminate any confusion in device selection. For thermal shutdown on Intellimax devices, most ICs in general do not rely on this feature as a general practice. That said, in normal use, if thermal events are expected, the general practice of separate temperature sensing should be used. Relying on continuous overtemperature shutdown may degrade the performance of the IC.

If an overcurrent is detected, the threshold level can be preset in the IC factory. It is also possible to set the level externally using a resistor to ground in some smart load switches. While most have short-circuit protection, the latest addition is a significantly improved tolerance in specific current disconnects, ranging from 100mA to 2A. In just a few years, current sensing tolerances have been reduced from 30% to 10% accuracy. When selecting threshold levels, note that the minimum and maximum specifications can vary based on process, voltage, and temperature. The dynamic range of current is relatively large, making it difficult to provide accurate and consistent transition points. It is also difficult to react to very slow current climbs when approaching the detection point. If accurate current sensing and load disconnection are important, it is possible to add a small amount of inductance to the output. This will "buffer" the change in current di/dt, allowing the smart FET to sense delta more accurately. The size of the inductor will directly reflect the sensitivity of the current transition. After an overcurrent event occurs, each series of smart MOSFETs reacts differently. Some shut off completely, others ramp down the current in preset steps, and some even provide a fixed voltage output at the safest tolerable current limit. Pay close attention to this specification when selecting components.

Smart MOSFET Specification Comparison

Having discussed the advantages, what are the possible disadvantages or sensitive specifications that must be closely evaluated when selecting a smart MOSFET? The key lies in the smart function within the smart FET. Of course, power must be used to sense the current and drive the high-side switch. This will be written in the quiescent current specification of the data sheet, which is the effective current used within the IC to calibrate and drive the load switch. For the Intellimax product line from Fairchild Semiconductor, this specification is less than 1μA minimum. For applications seeking maximum battery life, the listed leakage current must also be carefully compared.

When comparing Intellimax FETs, perhaps the most common data in the datasheet being evaluated is the same data that is of interest on a regular discrete MOSFET datasheet. The on-resistance of the high-side FET, called Rdson, is the key number used to calculate the losses across the load switch. This Rdson will vary based on the input voltage, since the same Vin is used to drive the high-side FET, so it is practical to use Ron as a target figure for a particular application. When the application will actually operate at 50%, Vin is often used to calculate the lowest Ron, so do not compare the absolute lowest Rdson in two datasheets. Based on this Ron value, if the current required by the load is known, the losses across the FET can be calculated. For Intellimax, Rdson can range from 20 ohms to 200 ohms, depending on the characteristics and package size.

Another data sheet detail that is sometimes overlooked is the maximum voltage of the high-side FET. To keep Rdson as low as possible, the Intellimax line limits the input voltage to 6V. This is perfect for battery-powered applications, whether it is a 3.7V rechargeable battery or a AA battery pack. 3.7V single-cell Li-Ion battery packs are becoming very common in portable medical applications due to the widespread use of cell phones. However, medical applications may also require hydraulic pumps or fans to operate at voltages that are independent of the core battery pack. The most common batteries here are double or triple stacked rechargeable batteries, bringing the voltage to 8V to 12V. In the past, discrete MOSFETs were used at these voltage levels. New developments have enabled smart FETs to reach higher voltages.

Fairchild Semiconductor's AccuPower family of integrated load switches is based on a 40V absolute maximum, 36V recommended process, a big technology leap for medium voltage applications. The first IC will use 100 ohm technology and have the same features supported by the Intellimax family, but will also include adjustable current limit and a power good (Pgood) pin. Because of the longer voltage ramp, the load should be at 36V, and the Pgood feature will indicate to the microprocessor that the output voltage is acceptable. Adjustable current limit opens up medical applications. AccuPower devices can be used to drive DC solenoids, fans, pumps, etc. Even if the battery voltage is at 12V, the L di/dt voltage spike across the dynamic winding load will easily exceed the 12V breakdown voltage or even the 20V breakdown voltage of discrete FETs. The 36V breakdown voltage supports these types of loads with 12V and possibly 24V battery voltages. The FPF2700 device supporting these voltage levels is available now.

Smart MOSFET for Medical Applications

After reviewing the latest in battery technology and the transition from traditional load switches to smart FET load switches, we can see how medical applications benefit, but the perceived value may vary. Portable medical devices value the disconnection of the power supply and load in order to extend battery life. However, as we discussed, what happens after the switch is disconnected is just as important, if not more important. Power conditioning adds immediate reliability to higher current applications when inrush current or overcurrent occurs.

Regardless of the application, trends in load isolation points continue to evolve, and smart MOSFETs can help achieve higher performance and greater reliability. A range of features require rapid implementation if medical applications are to maintain their edge over competitors. Traditional P-channel FETs will continue to be used for simple switches, but when reliability and time to market become key indicators of product design, the latest advances in smart MOSFET technology cannot be ignored.