EDN Network

The first 50G passive optical network (PON) gateway system-on-chip (SoC) incorporates a neural processing unit (NPU) for edge AI inference and offers native compatibility with the Wi-Fi 8 standard. Broadcom’s BCM68850 CPE gateway claims to offer NPU-accelerated solutions across cable, PON, Wi-Fi, and set-top box platforms, ensuring resilient infrastructure for AI offloading and high-efficiency multi-gigabit workloads.

BCM68850 aims to reshape the broadband edge as the home’s central intelligence hub. Source: Broadcom

The gateway SoC delivers full 50G throughput to meet multi-gigabit bandwidth. Besides a dedicated NPU, which reduces cloud latency and enhances data privacy by keeping sensitive information on premises, it also features a dedicated CPU for third-party and operator applications that leverage industry-standard middleware.

That, in turn, optimizes CPU and memory resources to ensure the home gateway can handle the massive data throughput required by edge AI-centric applications. “Home gateway solutions such as Broadcom’s BCM68850 SoC are critical to future-proofing the network edge,” said Jaimie Lenderman, practice leader for optical, IP, and broadband infrastructure market research at Omdia.

The standalone 50G PON Gateway SoC claims to provide an industry-standard ITU-T path for operators to future-proof their networks by processing and transmitting high-density payloads in a fraction of a millisecond. Moreover, its “burst and release” capability ensures near-zero-jitter essential for latency-critical applications.

Broadcom is currently sampling the BCM68850 home gateway SoC to its early access customers and partners.

Related Content

• Guaranteeing QoS in PON Designs
• Broadcom delivers Wi-Fi 8 chips for AI
• Gateways bridge gap between home, Net
• CES 2026: Wi-Fi 8 silicon on the horizon with an AI touch
• Gateway lets consumer control home appliances over mobile wireless nets

The post The 50G PON gateway SoC supports edge AI, Wi-Fi 8 appeared first on EDN.

How to deal with the reality that a standard buck controller power stage won’t work for negative-input conversion.

Telecommunications equipment, industrial test and other applications require a negative input to negative output voltage conversion. Because dedicated controllers for this topology are rare, you need a workaround to generate a stable output.

One solution I’ve found is to connect a boost controller’s GND pin to the negative input rail, which repurposes the device as a negative-input, negative-output buck controller and eliminates additional gate-drive circuitry. It then becomes possible to have a level-shifted feedback network regulate the output. So in this power tip, I’ll discuss two approaches using a traditional switch-mode power-supply: one using a buck controller with external field-effect transistors (FETs) and one using a buck converter with integrated FETs.

Comparing standard and negative-input, negative-output buck controllers

The power stage of a standard buck controller (Figure 1) closely resembles a negative-input, negative-output topology.

Figure 1 This simplified, standard buck controller schematic resembles that of a negative-input, negative-output topology.

A buck controller operates by applying a pulse-width modulation waveform to an inductor-capacitor filter. Each switching cycle starts when the main switch turns on, increasing the inductor current. Current flows from the input capacitor through the inductance to the output capacitor and back to the input capacitor. During the off time, the current commutates to the low-side diode (or switch) and the inductor current decreases.

Why a standard buck controller power stage won’t work for negative-input conversion

A negative-input, negative-output buck controller behaves very similarly to a standard buck controller. The main difference is that all currents flow in the opposite direction.

You cannot use a standard buck controller power stage, though, because of the orientation of the diode and metal-oxide semiconductor field-effect transistor (MOSFET) (with its internal body diode). Rotate these components as shown in Figure 2.

Figure 2 This simplified schematic details a negative-input negative-output buck controller.

Note that the output voltage cannot become more negative than the input voltage.

As an example, with a –48V input and a 50% duty cycle, the controller generates a –24V output. The controller’s control law decreases the negative output voltage toward the level of the negative input by increasing the “on” time of the main FET. So at a theoretical 100% duty cycle, the output voltage nearly equals the input voltage of –48V.

A standard buck controller will also not work here because both the input and output voltages are negative. For a negative-input, negative-output buck controller, the main FET connects to –Vin, and the cathode of the diode connects to GND. However, a boost controller works if you connect the GND pin to the negative input – a necessary step because otherwise all internal signals would be negative, creating a problem. Another reason is that a boost controller uses a ground-referred gate driver. Connecting the GND pin to the input voltage allows you to drive the FET without additional circuitry.

Using a nonsynchronous boost controller as a negative-input, negative-output buck controller

Figure 3 shows an example schematic using the Texas Instruments (TI) nonsynchronous boost controller to drive the main FET of a negative-input, negative-output buck controller.

Figure 3 This simplified schematic showcases Texas Instruments’ LM5155 boost controller.

Because –Vin is connected to the GND pin, all internal signals reference –Vin. Since –Vin typically varies across an input voltage range, this behavior can cause difficulties when enabling or disabling the device, regulating the output, or other protection features. Typically, you will need a level shifter (for example, an isolated type or one with bipolar FETs) to regulate the negative output voltage.

Configuring a boost converter for negative-input operation

A boost converter with internal switches can also work in theory, because the source of the main switch connects to the GND pin and the drain of the rectifier FET connects to the Vout pin.

Figure 4 shows a block diagram of a boost converter and the connection of the switches to the integrated circuit (IC) pins.

Figure 4 This boost converter block diagram includes the switches’ connections to the integrated circuit pins.

The challenge in using a converter is that many signals are internal. Some ICs integrate the output voltage divider, which makes regulating a negative output voltage difficult. Because all internal voltages reference the negative input, the output voltage would follow the input voltage. In that case, you can use the COMP output instead of the internal feedback. Connecting an optocoupler as a level shifter between COMP and GND provides one method to regulate the negative output.

Figure 5 shows how to connect a boost converter to a negative-input, negative-output buck power stage. The GND pin connects to the negative input, and the Vout pin (or FB pin) connects to power-stage ground. You can short the FB pin and use the COMP pin with an optocoupler to regulate the output. Keep all voltages, including current-sense signals, below the maximum limits of the boost converter.

Figure 5 This simplified schematic employs the boost converter shown earlier.

Design considerations

You can use a nonsynchronous boost controller such as TI’s LM5155 or TPS40210 as a simple, cost-effective solution for generating a negative output from a negative input. To increase the efficiency, replace the diode with a MOSFET, though doing so requires a synchronous boost controller that drives two switches. Negative voltages can easily cause confusion. In particular, you must check all internal signals and verify that no voltages are exceeding the controller’s maximum rating.

Florian Mueller is a systems engineer and Member Group Technical Staff in TI’s Power Supply Design Services group. He has a master’s degree in electrical engineering from the Technical University of Haag, Germany. Florian’s main focus lies on industrial high-voltage designs for different end equipment.

 

Related Content

• Power Tips #151: Improving efficiency in 48V-input multiphase buck converters with GaN
• Power Tips #146: Design functional safety power supplies with reduced complexity
• Power Tips # 141: Tips and tricks for achieving wide operating ranges with LLC resonant converters
• Power Tips #123: Using a double-boost converter to extends the power range of high-conversion-ratio designs

The post Power Tips #153: How to generate a regulated negative output from a negative input using a boost controller appeared first on EDN.

In embedded systems, where failure is not an option, NOR flash devices storing boot code, firmware images, and critical application data are subject to gradual wear over their operational lifetime. That wear is not invisible; it’s reflected in internal device registers accessible at runtime without the need for external test equipment. Per-sector erase cycle counts, single-bit and double-bit error correcting code (ECC) event counters, and hardware-accelerated cyclic redundancy check (CRC) integrity results collectively form a health profile.

This profile covers user-defined address ranges and is updated continuously as the system operates. On certain device variants, an on-board temperature sensor provides confirmation that the device is running within its rated thermal envelope. These are no fault flags that fire after something has gone wrong. They are observable quantities whose value lies in being monitored over time.

The central premise of flash diagnostics is shift from reactive fault handling to proactive health monitoring. Fault handlers consult device status when an operation fails.

In contrast, diagnostics applications read the same registers on a schedule, build a time series, and watch for early indicators of wear. Early warning paves the way for preventative maintenance, fixing impending problems before they trigger failure.

Reading wear, ECC, CRC, and thermal trends over time

Program/erase cycle counts per sector are the most direct measure of wear. Flash arrays in real-world applications are not erased uniformly. Sectors holding frequently updated data, such as fault codes logged by an automotive ECU or a partition used for over-the-air updates, accumulate cycles at a much higher rate than sectors holding static firmware.

Some NOR flash memories—such as Infineon’s SEMPER NOR Flash—offer built-in wear leveling defenses that distribute P/E cycles across the full address range. A diagnostics application periodically tracking per-sector counts can identify this imbalance early and provide the system with the information needed to act, whether by redistributing write activity or by flagging sectors approaching their service limit.

ECC event counts add a sensitivity that cycle counts alone cannot provide. Single-bit events are corrected transparently by on-chip logic and produce no visible effect on system operation, but their rate carries information about how individual cells are aging. A sector whose single-bit event rate begins to rise is showing early signs of cell wear, something the cycle count alone may not yet reflect.

When this trend is observed, rewriting the sector contents to restore cell charge state is one response the diagnostics system can initiate. To ameliorate system performance, the process can be scheduled during low-activity periods. Whether and at what threshold to trigger a refresh is a configurable decision. Double-bit events represent a harder boundary: the device detects them but cannot correct them, and their occurrence is recorded with sector address and timestamp for subsequent analysis.

CRC integrity checks over defined address ranges complement the bit-level view ECC provides, catching consistency issues that fall outside the scope of individual ECC words. For example, CRC is often used to validate a full firmware image region after an OTA update completes. Thermal reading, where available, confirms whether the device has been operating within its rated temperature range. This data assists in evaluating whether observed ECC trends reflect normal aging or accelerated cell wear from sustained thermal stress.

Diagnostics across AUTOSAR, Linux, and bare metal

The same NOR flash device frequently appears in multiple ECU variants within a single vehicle platform, each running a different software environment. A diagnostics software module such as SEMPER Diagnostics Library can be configured to span this portfolio, covering AUTOSAR Classic and Adaptive, Linux, QNX, RTOS, and bare-metal environments without changing the underlying health monitoring logic. What differs between environments is only the integration surface.

In AUTOSAR, the diagnostics module fits as a complex device driver. Positioned above the memory hardware abstraction layer, it accesses device-specific commands and register reads that the standard flash driver interface does not expose, while making its outputs available to upper-layer software components through defined RTE ports.

Figure 1 Here is how SEMPER Diagnostic Library software architecture operates in AUTOSAR environment. Source: Infineon

In a POSIX environment such as Linux or QNX, the same logic runs in user space and issues health queries through the IOCTL mechanism on an extended driver. Where the system is a heterogeneous SoC, a diagnostics agent in the real-time domain writes health query results to a shared memory region. A counterpart Linux user-space process then reads through a character device, packages with device identification and timestamps, and routes to a storage destination.

Within Linux, the Memory Technology Device (MTD) subsystem is the integration point for the flash driver, and IOCTL commands on an extended MTD driver are the mechanism by which device-specific health metrics cross the user-space boundary without touching standard read/write paths. On bare-metal or RTOS systems, the library links directly with the memory driver and is scheduled by the task manager.

In the case of SEMPER NOR Flash, SEMPER Diagnostics Library provides the diagnostic data, and the user is free to log it to local flash, route it to the cloud, store it in an external database, or any other destination that fits their system architecture. Similarly, fleet-connected deployments can route the same data off-device for population-level analysis. The underlying algorithms are identical across all environments; only the integration scaffolding differs.

Diagnostics library: Architecture and demo

Figure 2 Integration examples are shown for SEMPER Diagnostics Library module across different software environments. Source: Infineon

Figure 3 The demo setup is running SEMPER Diagnostics Library on Linux (RaspberryPi) while showing Erase Count and ECC Errors per sector. Source: Infineon

The SEMPER NOR Flash diagnostics software dashboard, shown below, visualizes per-sector erase counts and ECC counts in real time, along with device metadata—Device ID, Chip Size, Protocol, ECC State, Address Mode, Page Size—giving engineers a turnkey view of the flash health profile without requiring custom tooling.

Figure 4 The diagnostics software dashboard visualizes per-sector erase counts and ECC counts in real time. Source: Infineon

Fleet telemetry and predictive maintenance

Health metrics tagged with a unique device identifier and correlated with vehicle operating history become qualitatively more useful at scale. Patterns invisible at the level of a single device become apparent when data from a large population is examined holistically.

For example, a correlation between a specific duty cycle profile and accelerated sector wear may appear random as a single event, but causal when considered in aggregate. This is the difference between diagnosing a device that has already failed and identifying a population that may fail while every unit in it is still functioning normally.

Estimating useful lifetime also benefits from the same accumulated data. A static model applying a single worst-case endurance figure will produce overly conservative estimates. SEMPER Diagnostics Library’s adaptive lifetime estimation concept goes further: observed erase count progression, ECC event rates, and thermal history enable a per-device estimate that reflects how that specific unit has been used with the potential to refine it further through fleet-level pattern recognition, identifying trajectories that have historically preceded reliability events.

Act before wear

NOR flash devices save a continuous stream of health data in their internal registers, yet most systems discard it. Per-sector erase counts, ECC event trends, CRC integrity results, and thermal confirmation collectively describe how a device is aging under its actual operating conditions. The information is available at runtime, and no additional hardware is required to harvest it. The longer it is collected, the more valuable it becomes.

A diagnostics framework such as SEMPER Diagnostics Library captures this data, made possible via hardware such as SEMPER NOR memory, consistently across AUTOSAR, Linux, and bare-metal environments, routes it across processing domain boundaries, and makes it available for both on-device response and population-level analysis.

This gives engineers advanced notice to act before wear affects system reliability. In applications where that lead time separates a scheduled maintenance event from an unplanned failure, the case for building it in from the start is clear.

Saurabh Tripathi is senior applications engineer at Infineon Technologies.

Related Content

• The Shift from 2D to 3D NOR Flash
• Flash 101: NAND Flash vs NOR Flash
• NOR Flash Next in AI-Driven Memory Crunch
• NOR flash memory certified for ASIL-D functional safety
• How NOR flash helps overcome design challenges in wearables

The post Flash diagnostics and health monitoring for NOR memory appeared first on EDN.

Engineers must carefully manage power, protection, and sensing interactions at every design level to achieve reliable, always-on operation in a body-worn form factor. (Source: Getty Images)

Continuous glucose monitors (CGMs) have reshaped diabetes management by delivering real-time glucose readings, freeing patients from frequent finger-stick testing. These compact, wearable devices not only enhance quality of life but also allow clinicians to adjust therapy based on accurate, continuous data streams.

Behind this innovation lies a complex engineering challenge: Designers must develop a device that operates safely and reliably on a micro-scale power budget, fits within a compact, body-worn form factor, and maintains precise sensing accuracy in all conditions. Every component, whether analog, digital, power management, or protective, must contribute to long-term reliability and patient comfort. In many designs, even microamp-level leakage or a single mechanical failure point can limit device lifetime or compromise reliability.

Magnetic sensing, particularly tunnel magnetoresistance (TMR) technology, offers a practical approach for implementing sealed, contactless activation and other event-based state-detection functions without materially impacting battery life. This article examines the role of magnetic sensing in CGM architectures, explains the operating principles of TMR switches, and discusses their applications for activation, alignment confirmation, and auxiliary-state detection. Design tradeoffs, implementation considerations, and package-level constraints are also explored to help engineers evaluate when TMR sensing is appropriate in CGM designs.

The role of CGMs in connected healthcare

CGMs are central to modern diabetes care. They measure glucose concentration in interstitial fluid using a sensor inserted beneath the skin, which transmits readings wirelessly to a smartphone, insulin pump, or cloud-based management system.

Connected drug delivery system example (Source: Littelfuse Inc.)

The benefits of CGMs are well-established: reduced glycemic variability, better HbA1c levels, and fewer hypoglycemic episodes. As the technology matures, CGMs are now prescribed for a wider population, including patients with Type 2 diabetes, gestational diabetes, and even pre-diabetic conditions, expanding their relevance across preventive medicine and chronic care.

From an engineering perspective, these devices embody the broader trend toward connected, always-on healthcare systems, in which safety, data integrity, and energy efficiency are equally critical.

System architecture and design constraints

A typical CGM system includes five key components:

• The glucose sensor and analog front end amplify and condition microvolt-level signals from the biosensor.
• The microcontroller processes data, handles algorithms, and manages wireless communication via Bluetooth Low Energy or proprietary protocols.
• The power-management circuitry regulates energy from a small rechargeable or disposable battery.
• The wireless interface communicates readings to companion devices or cloud platforms.
• Temperature sensing, protection, and activation circuits safeguard operation and enable user interaction.

Simplified CGM system block diagram (Source: Littelfuse Inc.)

These modules must function continuously for seven to 14 days on a single charge, all while exposed to motion, sweat, temperature fluctuations, and electrostatic discharge (ESD). Component size, thermal behavior, and power efficiency dictate patient comfort and product usability.

Engineering challenges unique to CGM design

Engineering challenges in CGM design include achieving ultra-low power consumption and extreme miniaturization in limited PCB space while maintaining electrical safety/isolation and environmental resilience. Designs must also meet stringent regulatory compliance requirements:

• Ultra-low power consumption: Every microamp of leakage current reduces battery life. Components must have negligible quiescent draw.
• Miniaturization: Patch-style and implantable CGMs allow only millimeters of PCB space, demanding small-package, high-performance devices.
• Electrical safety and isolation: Circuit faults must be contained quickly to protect the patient and device integrity.
• Environmental resilience: Resistance to sweat, vibration, and humidity ensures consistent operation throughout the wear cycle.
• Regulatory compliance: Designs must comply with IEC 60601, ISO 13485, and 21 CFR 820 requirements for safety, quality, and EMC performance.

Meeting these demands requires careful component selection and system-level integration.

Magnetic activation for sealed, contactless operation

Power-on and reset functions are fundamental in wearable devices. Traditional mechanical pushbuttons introduce contamination paths, wear over time, and complicate waterproofing. The activation circuit keeps energy consumption during the shelf life to a minimum, ensuring the device remains safe to operate after 24 months. Magnetic activation provides a contactless alternative that enhances durability and hygiene.

Three magnetic-switching technologies are available: reed relays, Hall-effect sensors, and TMR switches. Each presents tradeoffs in power consumption, sensitivity, and footprint (see Table 1 for a comparison). In practice, the key differentiator is standby current, whereby TMR operates in the nanoamp range, versus milliamps for typical Hall-effect devices.

Table 1: Sensing technologies comparison (Source: Littelfuse Inc.)

TMR sensors offer a highly effective combination of performance characteristics for CGM applications: nanoamp-to-microamp power levels, compact LGA packages, and omnipolar detection for flexible magnet placement.

For example, Littelfuse TMR magnetic switches detect flux changes as low as 9 Gauss and draw only 160 nA in low-speed mode. Their contactless operation enables features such as automatic power-on when the device is applied to the skin or activation during packaging removal. Because they have no moving parts, TMR switches are immune to vibration and moisture, providing a lifetime of tens of billions of switching cycles.

TMR magnetic switches such as the TMR LGA4 Switch LF21173TMR enable contactless activation through a sealed enclosure. (Source: Littelfuse Inc.)

By eliminating mechanical interfaces, engineers reduce mechanical failure risk, improve sealing, and extend battery life—all critical for patient-worn electronics. This approach makes TMR switches particularly attractive for designs in which activation must remain available throughout storage and use without impacting overall system power budgets.

Thermal monitoring and patient safety

Temperature sensing plays multiple roles in CGM design:

• Electronic safety monitoring detects abnormal heat buildup from circuit faults or battery degradation.
• Patient protection prevents surface temperatures that could irritate or burn the skin.
• Sensor compensation adjusts for temperature-dependent enzymatic reactions that influence glucose readings.

Compact NTC thermistors, such as Littelfuse’s 0803-KR, 0603-RB, and 1206-LR series, offer ±5% accuracy in packages as small as 1.6 × 0.8 × 1.0 mm. Engineers often use multiple thermistors: one near the biosensor for reaction compensation and another near the battery or power-management circuitry for thermal safety monitoring.

Precise thermal feedback not only protects users but also enhances measurement accuracy, contributing directly to clinical reliability.

The number, location, and role of temperature sensors vary by CGM architecture, but designers generally distinguish between temperature sensing for safety monitoring and temperature sensing used for measurement compensation.

Integrating protection and sensing for reliable operation

Effective CGM design blends protection, sensing, and activation elements into a cohesive system. Integration offers several key benefits:

• Extended battery life through ultra-low leakage protection and sensing components
• Improved mechanical reliability by eliminating moving parts and exposed contacts
• Simplified certification when using pre-qualified components compliant with medical standards
• Enhanced user confidence through consistent, failure-free performance

When these design principles are applied, engineers can focus on refining algorithms, connectivity, and patient-experience features rather than troubleshooting hardware faults.

Regulatory and compliance considerations

Every CGM must meet stringent international standards to ensure safety and performance. Table 2 outlines the most relevant to electronic subsystems.

Table 2: Applicable international standards for CGM compliance (Source: Littelfuse Inc.)

Choosing electronic components with existing documentation for these standards can streamline risk management files and accelerate regulatory review.

Future trends in CGM and wearable design

As wearable healthcare expands, designers are targeting a reduction in device size, longer lifetimes, multi-sensor integration, and cloud-connected analytics. Each evolution places an even greater emphasis on power efficiency and electrical safety.

Emerging technology trends include:

• The integration of multi-parameter sensors (glucose, lactate, temperature, and hydration)
• The use of energy-harvesting or inductive-charging technologies to extend operating life
• The implementation of advanced protection monitoring, such as built-in diagnostics for ESD or fuse status
• The development of biocompatible, flexible electronics to further improve patient comfort

Component suppliers that offer medically focused design support and validated protection portfolios will play a crucial role in accelerating these innovations.

CGMs exemplify the convergence of biomedical science and advanced electronics. To achieve reliable, always-on operation in a body-worn form factor, engineers must carefully manage power, protection, and sensing interactions at every design level.

By integrating TMR magnetic switches for contactless activation, NTC thermistors for safety and compensation, low-leakage ESD/TVS diodes for transient protection, and miniature medical-grade fuses for fault isolation, developers can meet the strict performance and safety requirements of modern medical devices.

The result is a new generation of CGMs that are smaller, more power-efficient, and more reliable, meeting the practical constraints of wearable system design while enabling accurate, continuous monitoring.

About the author

Marco Doms is a senior manager of business development new platforms at Littelfuse Inc. Doms studied electrical engineering and holds a Ph.D. in MEMS. He was the head of R&D at two other sensor companies before joining Littelfuse in 2022. Doms has a long history in position sensors (especially xMR) and managing R&D and Innovation teams—from chip to system level. At Littelfuse, he started as an innovation manager, led the EBU Advanced Development team, and introduced an Innovation/Idea Management process. In his current role, Doms is responsible for several platforms with entirely new products or product features that require additional internal and customer coordination.

Marco Doms is senior manager of business development for new platforms at Littelfuse Inc.

The post Designing low-power CGMs with TMR-based magnetic sensing appeared first on EDN.

While its vehicle battery resurrection skills are uncertain at best, this device also offers other useful abilities.

Two-plus years back, within my teardown of my PowerStation PSX3:

which I described at the time as being:

…(among other things) a portable recharger and jump-starter of vehicles’ cells. It’s also a portable tire inflater. And it’s an emergency light and USB power source, too…

I took advantage of the opportunity to also editorially “rip” into three newer solid-state and Li-ion battery-based versions of the same concept:

I tried three of these widgets, one claiming to deliver 1200 A of “peak” cranking juice:

Another spec’ing 1500 A:

And a third that promised to deliver 2000 A:

They all promptly went back to Amazon as full-refund returns. Now granted, if someone had left their interior dome light on too long and the battery was drained too low to successfully turn over the engine but still had some “life” one of these might suffice…And I’ll grant them one other thing: they’re certainly small and light.

But 2000 A of cranking current? Or even 1500 A? Mebbe for a fraction of a second, the time necessary to drain an intermediary capacitor, but not long enough to resurrect a significantly drained battery. Therefore, the quotes I put around the word “peak” earlier. Such products exemplify the well-worn saying, “mileage may vary”. Give me an old-school lead acid battery instead, any day!

Regarding my “They all promptly went back to Amazon as full-refund returns” comment, while that was my original intent, I didn’t end up fully actualizing it. The “1200 A” and “1500 A” variants indeed did get shipped back to the retailer. But, curiosity-motivated, I decided to keep the “2000 A” model, Spanarci’s ZETA2000, around if for no other reason than as a future teardown candidate.

Calling Cupertino…

That future is today. As usual, I’ll start with some outer box shots (sparing you the blank sides), as usual accompanied by a 0.75″ (19.1 mm) diameter U.S. penny for size comparison purposes:

Although my skepticism about the device’s jump-starting potential is already obvious at this early point in the writeup, I was admittedly impressed by the aesthetics and overall packaging of the product. Dare I even say it was Apple-reminiscent?

The cleverly labeled “Never Say Never” envelope, reminiscent (at least to me) of SpaceX’s three autonomous spaceport drone ships (i.e., floating rocket booster landing pads), “Of Course I Still Love You”, “Just Read the Instructions” and “A Shortfall of Gravitas”, contains literature bits:

Gee, I wonder what’s inside this translucent plastic sleeve?

To stretch the suspense, I’ll temporarily set it aside and investigate the lower box level instead:

Within is the to-vehicle battery cable harness, conveniently accompanied by USB-A-to-USB-C and USB-C-to-USB-C cables useful both for recharging the device’s internal battery pack and for powering other connected devices. Hold that thought:

Here’s the male connector at the end of the cable harness…

Burlesque finale

And here’s what it plugs into…

at out dissection patient, finally unswathed for its reveal. Top first:

Here’s the front:

Underneath the rubberized flap labeled “INPUT OUTPUT” at the right end are, likely unsurprisingly, first a bidirectional USB-C PD 30-W connector used for both device charging and for charging/powering another tethered device, such as a smartphone. The other, USB-A in form factor, is unidirectional (output-only) for similar tethered device “juicing” purposes.

Onward. Left side:

Rear; under this flap, cryptically (ha!) labeled “JUMPER CABLE” is the battery-cable harness connector you saw earlier:

And what the heck is that on the right side? A multi-LED strip, creating a 300-lumen four-mode (50% and 100% brightness stable, and both SOS and strobe pattern) flashlight, that’s what it is!

Last but not least, here’s the bottom view:

accompanied by a zoom-in of the specs:

Before opening ‘er up, I’ll note a few other feature set nuances. Like the conventional (i.e., AC-powered) solid-state charger that I tore down earlier this month, it supports various safety features such as short-circuit and “reverse” protection:

That said, there’s also “FORCE” support for dead cells and daring users:

The teardown’s the thing…

And now, let’s dive inside. Zoom back out on that earlier bottom overview shot and you’ll discern eight round rubber pieces, one in each corner and two more both at top and bottom:

I bet you can guess what comes next:

Eureka! Screw heads (trust me, they’re there, deep inside the recessed dimness)!

And what comes after that, dear readers? You got it right:

Dare I draw another analogy to Apple craftsmanship? Seriously, I’m impressed with the neatness and overall robustness of the insides, too!

Here’s the inside of the case topside:

…wherein I’ll detail the insides of the thing

And the overview that’s likely of greater interest to all of you!

Dominating the landscape, aside from the display, that is:

is the largest IC on this side, at center (horizontally) and toward the bottom (vertically). It’s Holtek’s HT67F489 8-bit RISC microcontroller, unsurprisingly with an integrated LCD controller and also containing (among other things) 8 Kwords of flash memory (4 Kwords on the more modest HT67F488 sibling, which the datasheet informs me (PDF) has been discontinued, anyway), 256 bytes of RAM and 64 bytes of EEPROM (none on the HT67F488). Also note two mode-select switches at far right, which mate to rubberized front panel buttons.

Let’s get that PCB out, shall we? Three screws hold it in place:

Guess what comes next?

In addition to noticing the now-absent screws (and their previously visible heads) in the next photo, I’d also like to draw your attention to the smaller but still-square IC to the right of the aforementioned HT67F489. It’s Southchip Semiconductor Technology’s SC2001 USB-PD controller. Given what you already know about the capability of the USB-C connector on the front of the device, this chip’s presence and functions shouldn’t be a surprise.

Here goes nothing:

My, what a big power source you have…

At left is the 44.4 Wh lithium polymer battery pack:

To its right is a beefy Sanyi Seiko SEV8-P-112DM 4-pin high-power relay:

soldered to a mini-PCB:

And the remainder of the compartment mostly consists of a bunch of now-disconnected wire harnesses:

The destination of one of them was, I admit with no shortage of chagrin, initially identity-baffling to me, until I pulled it out. See that gold-colored half-oval to the far right?

Oh yeah. The LEDs. Duh on me:

Underneath that large green region on the PCB underside is, as far as I can feel, nothing notable save for mounting-bracket sites and solder points related to the LCD on the other side:

The PCB-mounted speaker in one corner delivers a loud “beep” tone if, for example, you’ve got your to-battery connections reversed:

The one next to it is “just” an inductor (L1 is peeking out from the PCB under the white glue):

It, along with the rest of the components surrounding it (and some of those on the other side), implements a largely unmemorable power management subsystem.

In closing, I’ll share a side view of the USB-C and USB-A connectors; since the PCB is upside-down from its normal operating orientation, so are they:

…the better to incinerate you with, my dear

With that, I’ll close for today. Speaking of closing, I’ll keep the device disassembled for a while post-publication of this teardown. Then I’ll carefully reassemble it in the hopes of resurrecting it. If you smell smoke, see flame, or hear a loud “boom”, you’ll know my efforts didn’t succeed.

—Brian Dipert is the associate editor, as well as a contributing editor, at EDN.

Related Content

• The PowerStation PSX3: A portable multifunction vehicular powerhouse with a beefy battery
• NOCO’s Genius 1: A trickle charger that tries harder
• A battery charger that loudly hums: Dump it or just make it dumb?
• Dead lead-acid batteries: Desulfation-resurrection opportunities?

The post Portable jump starters: A dubious primary use case, but not a total waste appeared first on EDN.

Gas discharge tubes (GDTs) harness the physics of controlled sparks to provide reliable surge protection, making them a fundamental safeguard for modern electronic circuits.

They are deceptively simple devices that rely on ionized gas to tame the chaos of voltage surges. When a transient spike threatens sensitive circuitry, a GDT responds with a controlled spark, safely channeling excess energy away from the system.

Compact, rugged, and reliable, these components have become indispensable in applications ranging from telecom lines to industrial equipment. In essence, GDTs turn sparks into protection, making them a cornerstone in the engineer’s surge-defense toolkit.

It’s worth noting that GDTs are sometimes referred to as plasma arrestors. The two names describe the same device; a sealed tube filled with inert gas that forms a plasma arc when voltage exceeds its breakdown threshold. “GDT” is the term most often used in engineering literature and standards, while “gas plasma arrestor” tends to appear in catalogs or marketing to highlight the plasma discharge mechanism.

Inside the spark: How GDTs work

From the first spark to the final safeguard, gas discharge tubes show how even the simplest devices can deliver powerful protection where it matters most. To understand why, let us take a closer look at how they work.

At the heart of a GDT is a sealed chamber filled with inert gas such as neon or argon. Two electrodes face each other across a small gap inside this chamber. Under normal operating conditions, the gas is non-conductive, and the tube behaves like an open circuit. But when a voltage surge pushes the potential across the electrodes beyond the breakdown threshold, the gas ionizes. This ionization triggers a plasma discharge—controlled spark—that suddenly makes the tube conductive.

The plasma arc provides a low-resistance path, diverting the surge current safely away from sensitive components. Once the surge subsides and the voltage drops below the sustaining level, the plasma extinguishes, and the tube returns to its insulating state. This simple cycle—breakdown, conduction, recovery—is what makes GDTs both rugged and reliable in protecting circuits against transient overvoltages.

Figure 1 A medium-duty 2-electrode gas discharge tube safeguards telecommunications, industrial, and consumer electronics from voltage surges. Source: Bourns

Shared sparks, shared protection

Building on the fundamentals, the next nuance lies in how protection is applied across conductors. A two-lead GDT serves as a straightforward single-path protector, perfect for shunting individual DC rails or coaxial cables to ground. But when you place two separate two-lead tubes across a data pair, they will never fire at precisely the same instant, leaving a harmful “transverse voltage” between the lines.

A three-lead GDT solves this by enclosing both conductors in a common gas chamber. The moment one side ionizes, the entire tube triggers, discharging both lines to ground simultaneously. This synchronized action delivers the balanced protection that sensitive telecommunications and differential data circuits demand.

Figure 2 A 3-lead GDT ensures simultaneous crowbar action across differential lines, preventing unbalanced residual voltages during a surge event. Source: Littelfuse

It’s important to note at this point that standard GDTs are commonly available in both 2- and 3-electrode configurations, whereas high-voltage variants are primarily limited to 2-electrode designs with select 3-electrode exceptions. While 2-electrode devices are typically deployed for either line-to-ground or line-to-line protection, a 3-electrode GDT provides the advantage of addressing both protection paths within a single component.

Practical implementation of GDTs

When selecting a GDT for a specific application, the primary objective is to ensure the device remains inactive during normal operation while reacting instantaneously to overvoltage transients. This requires careful evaluation of key electrical parameters, starting with the DC spark-over voltage. To prevent “nuisance” triggering, the GDT’s minimum breakdown rating should typically be 1.2 to 1.5 times the peak operating voltage of the system.

Furthermore, because GDTs are “crowbar” devices, engineers must account for follow-on current, the current that continues to flow through the ionized gas after the surge has passed. If the system’s power source can sustain this arc, additional current-limiting components or a coordinated circuit design may be necessary to ensure the GDT successfully resets to its high-impedance state once the transient is cleared.

However, follow-on current is often absent from GDT datasheets because it’s not a fixed constant of the device, but rather a system-dependent behavior. A GDT is essentially a triggered short circuit; once ionized, its resistance drops so low that the resulting current is determined almost entirely by your power supply’s voltage and internal impedance.

While manufacturers provide the arc voltage and the glow-to-arc transition current, they cannot predict your specific source’s capacity to sustain that arc. Consequently, engineers must use those parameters to calculate the “holdover” risk themselves, often necessitating components like metal oxide varistor (MOV) to effectively “starve” the arc and allow the GDT to reset.

To round out the technical profile, several other parameters define a GDT’s performance and longevity. Maximum impulse spark-over voltage is critical, as it indicates the highest voltage level the device allows during a fast-rising surge before it triggers. To gauge durability, engineers look at nominal impulse discharge current, which is the peak surge current the GDT can survive for a set number of pulses, and alternating discharge current, which measures its ability to handle sustained AC faults.

Additionally, maximum capacitance must be minimal to ensure signal integrity in high-frequency lines, while minimum insulation resistance ensures the GDT remains electrically “invisible” until a surge occurs.

Figure 3 Plot illustrates the GDT voltage breakdown characteristic. Source: Author

As a worthy take on paper, the GDT’s protective behavior is defined by its transition through distinct electrical phases, captured sequentially in Figure 3. The process initiates with the sparkover voltage, the exact point where the internal gas ionizes and becomes conductive. Immediately following this breakdown, the voltage falls to a relatively stable plateau known as the glow region, where current flows but remains limited.

As the surge energy intensifies, the device undergoes the rapid glow to arc transition, the critical threshold where the discharge collapses into a highly conductive plasma. This leads immediately to the arc voltage, the final “crowbar” state where the voltage drop plummets to its absolute lowest point. Identifying this transition sequence is vital, as the low arc voltage is precisely what triggers the risk of sustained follow-on current from the system’s power source.

GDTs are often evaluated against IEC 61000‑4‑5, the international surge immunity standard, because their protective behavior directly addresses the transient overvoltages defined by this test. The standard specifies surge waveforms—most notably the 1.2/50 µs voltage impulse and the 8/20 µs current impulse—to replicate lightning‑induced or switching transients. In these scenarios, GDTs act as frontline protectors, clamping and diverting surge energy away from sensitive equipment to ensure compliance and resilience.

Bonus insight: How to test a GDT surge arrestor

Have you ever wondered how to verify whether a GDT surge arrestor is still healthy and ready to protect against lightning, static, or electromagnetic pulse (EMP) events? An EMP is a sudden burst of electromagnetic energy—often from lightning strikes, solar storms, or even man-made sources—that can damage sensitive electronics. The only definitive way to confirm a GDT’s readiness is to make the device “fire”.

The most reliable approach is a DC high-voltage ramp test, performed with a power supply or a megohmmeter. Because a GDT behaves like an open circuit under normal conditions, you gradually increase the DC voltage across its terminals until it reaches the rated breakdown point. To ensure safety and prevent excessive current once the tube fires, a series resistor should always be included in the test circuit. This resistor limits the surge current, protects the power supply, and prevents overstressing the GDT during repeated tests.

Sparking applications, igniting ideas

Gas discharge tubes prove their worth across a wide spectrum of systems. In telecommunications, they safeguard MDF modules, xDSL equipment, RF systems, antennas, and base stations. In industrial and consumer electronics, they protect power supplies, surge protectors, alarm systems, and even irrigation systems.

Positioned in front of and in parallel with sensitive lines—power, communication, signal, and data transmission—GDTs shield equipment from transient surges caused by lightning strikes or switching operations. Under normal conditions they remain invisible to the signal, but when an overvoltage surge arrives, they switch to a low-impedance state and divert the energy safely away from the circuitry.

These sparks of protection are more than circuit defense; they are design opportunities. For makers and engineers, the challenge is to take this proven sequence from sparkover to arc and reimagine it in your own projects. Every surge control is a chance to build systems that are not only safer but smarter. So let the sparks inspire you: experiment boldly, refine relentlessly, and turn protective theory into resilient innovation.

T. K. Hareendran is a self-taught electronics enthusiast with a strong passion for innovative circuit design and hands-on technology. He develops both experimental and practical electronic projects, documenting and sharing his work to support fellow tinkerers and learners. Beyond the workbench, he dedicates time to technical writing and hardware evaluations to contribute meaningfully to the maker community.

Related Content

• Power-Back Surge Protection Circuit
• Overvoltage-protection circuit saves the day
• Robust overvoltage protection in high-density electronics
• How to prevent overvoltage conditions during prototyping
• Gas Discharge Tubes: Old Protection in a New Bottle–Literally

The post Gas discharge tubes (GDTs): From sparks to circuit protection appeared first on EDN.

Medical imaging has become one of the most critical pillars of modern healthcare to provide insights into diagnosis, treatment planning, and disease management. However, the very success of imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI) has created a growing challenge of data and decision-making. As imaging generates more information to interpret, artificial intelligence helps to improve these systems by supporting faster, smarter workflows for higher-accuracy diagnoses.

The volume of imaging studies has increased substantially over the past decade, putting additional pressure on the shortage of radiologists. At the same time, technological advances in scanner hardware have enabled the acquisition of thinner slices and higher-resolution images, with a single CT or MRI exam consisting of hundreds or thousands of images.

In clinical settings, the challenge is not whether scans have sufficient data but whether the health system can reconstruct, review, quantify, and interpret the data fast enough to support timely clinical decisions. We look at the use of AI and some of the popular deep-learning models in medical imaging and diagnostics while also examining how AI is being integrated across the imaging workflow.

AI enhances medical imaging systems such as CT or MRI scanners by supporting faster, smarter workflows for higher-accuracy diagnoses. (Source: Adobe Stock) AI across the medical imaging pipeline

AI is moving medical imaging and diagnostics from early-generation concepts and narrow automation toward broader integration across the imaging pipeline. The integration of AI is augmenting a wide range of tasks, from the moment an exam is ordered to the final clinical interpretation, to improve speed, accuracy, consistency, and efficiency. This approach addresses the critical bottleneck in the modern imaging workflow, turning a linear and often manual process into a more optimized, data-driven, and intelligent system.

The influence of AI begins even before a single image is acquired. This includes administrative and logistical steps that are important for optimization. For example, natural-language-processing models can analyze a patient’s clinical history and the reason for an exam within the electronic health record to help automate the selection of the most appropriate imaging protocol.

During the acquisition stage, AI contributes to image quality and efficiency. In CT, AI can automate and optimize scan ranges and radiation dose parameters based on the patient’s specific anatomy to ensure diagnostic-quality images are obtained at the lowest possible radiation exposure.

Image reconstruction is another impactful application of AI. Deep-learning reconstruction has changed this process. These models are trained on a large dataset of high-quality images to produce images with significantly lower noise and higher signal-to-noise ratio from under-sampled or low-dose raw data. For MRI, this means scan times can be reduced by up to 75% in some cases, without sacrificing image quality.

Once the images are created, AI is used for analysis and interpretation. In this phase, it helps radiologists in extracting clinically relevant information. Automated segmentation is the key task in which AI algorithms delineate anatomical structures, organs, or pathologies with high precision. This is an important prerequisite for quantitative analysis and is used to accelerate standardized assessment workflows, such as for the prostate imaging reporting and data system.

After the segmentation, AI tools for detection and triage can screen images for critical findings, such as intracranial hemorrhage, pulmonary embolism, or large vessel occlusions in stroke patients. However, AI in this setting is changing the order, speed, and consistency of review. A triage algorithm can bring a suspected emergency case to the top of the queue, while the radiologist remains responsible for confirming the findings, considering clinical context, and issuing the final report.

AI models in modern imaging diagnostics

The growth in powerful deep-learning architectures today serves as the engine for modern medical AI to perform complex tasks such as detecting minute pathological changes, precisely segmenting anatomical structures, and fusing information from different clinical sources.

Convolutional neural networks (CNNs) have become the go-to architecture for most AI medical imaging applications, especially in radiology. Their design is inspired by the human visual cortex and is well-suited for processing grid-patterned data such as images.

While CNNs are useful for classification tasks, medical imaging requires a more granular understanding of spatial information, such as tracing the boundaries of an organ or tumor. This task involves assigning a class label to every pixel in an image. For this purpose, encoder-decoder architectures, most popularly the U-Net, have become the de facto standard.

The U-Net design addresses this challenge by combining semantic context with low-level, high-resolution spatial information. The architecture has two main components: the encoder and the decoder. As the image data goes deeper into the encoder, the spatial resolution decreases, but the number of feature channels increases. This allows the architecture to capture context-rich information from the image.

The decoder’s role is to take the compressed, high-level feature representation from the encoder and progressively up-sample it back to the original image resolution to generate a pixel-wise segmentation map. It achieves this by using a learned transposed convolution to increase spatial dimensions.

The U-Net architecture uses skip connections that create a pathway for information to flow from the encoder to the decoder at corresponding levels of resolution. This fusion provides the decoder with the fine-grained spatial details that were lost during the down-sampling.

This is necessary in many diagnostic cases that are not a simple classification problem. The model not only needs to identify that a tumor, lesion, or abnormality is present, but it also outlines the boundary, calculates volume, compares change over time, or separates healthy tissue from pathology. This pixel-level requirement is why encoder-decoder architectures have become key to segmentation workflows.

The success of this concept has led to variants designed to further improve performance. U-Net++, for example, introduces nested and dense skip pathways to reduce the semantic gap between the encoder and the decoder feature maps, while Attention U-Net integrates attention mechanisms that allow the model to focus on the most relevant image regions. Other advanced versions, such as nnU-Net, provide a self-configuring framework that automatically adapts the network architecture and preprocessing steps for any given segmentation task.

However, CNNs have limitations in modeling long-range dependencies and global context within an image. This led to the exploration of Vision Transformers in medical imaging. Transformers can model relationships across wider image regions, which is useful for tasks in which pathology, anatomy, and clinical context are distributed across a larger field of view.

But at the same time, they face a domain gap between the natural images on which many of these models are pretrained and the unique characteristics of medical images. The black-box nature of these models raises concerns about interpretability, which is important for clinical trust and high-stakes decision-making.

How AI improves medical insights

Integrating AI into medical imaging brings enormous improvements in speed and operational efficiency. By automating time-consuming tasks at multiple stages, AI targets the workload pressures and delays that modern radiology faces. This results in faster diagnoses and more timely patient intervention.

AI is also increasing the accuracy of the diagnostic quality of medical imaging by reducing variability. Human interpretation is always subject to limitations because of fatigue, perceptual errors, and inter-reader variability, whereby different radiologists may interpret the same image differently. AI provides a more powerful set of tools to augment human perception.

AI systems are particularly strong in pattern-recognition tasks and have demonstrated the ability to detect subtle abnormalities that may be missed by the human eye. In lung cancer screening with CT, for example, studies have shown that AI algorithms can achieve a nodule-detection sensitivity exceeding 95% for nodules of 4 mm or larger.

In lung cancer screening with CT scans, AI algorithms can improve nodule-detection sensitivity and reduce the risk of a missed diagnosis. (Source: Adobe Stock)

A research study shows that AI detected 8.4% more lung nodules in patients with complex lung diseases. Similarly, in mammography, AI models have performed comparably to human experts in detecting breast cancer in certain validation studies. These systems function as a highly effective second reader that can help radiologists focus on potential concerns and reduce the risk of a missed diagnosis.

In addition, radiomics is built upon the foundation of AI-driven quantification. For example, radiomic features extracted from pre-treatment CT images have been used to predict survival in lung cancer patients, while signatures from MRI scans have shown a correlation with recurrence risk in glioblastoma patients.

What’s next

The current advancements are setting the stage for a future in which AI will be deeply integrated into diagnostics and patient care. One of the most important future directions is the maturation of multimodal AI and foundation models for a wider range of data types, including imaging, genomics, proteomics, digital pathology, clinical notes, and even real-time physiological data from wearable sensors.

The future of AI is likely to be one of human-AI collaboration. AI will handle the data-intensive tasks of detection, measurement, and quantification, while radiologists focus on higher-order tasks of complex synthesis and clinical correlation.

The post AI-powered medical imaging: Turning data into faster diagnoses appeared first on EDN.

Surface-mount polymer positive temperature coefficient (PPTC) devices are widely used resettable protection components in modern electronic systems. These devices are commonly selected when designers require automatic recovery after a fault, which makes them useful in consumer electronics, telecommunications equipment, industrial systems, and medical devices.

Despite their widespread use, detailed SPICE models for these components remain uncommon. Datasheets typically provide static electrical characteristics, but dynamic models that allow designers to simulate real operating conditions are rarely available.

Yet many practical operating scenarios can be reproduced effectively through simulation. Examples include overcurrent or short to ground events on a USB 5-V supply line, short circuit faults on a lithium-ion battery, or motor stall protection under varying ambient temperatures.

The value of such simulations is clear: they illustrate the consequences of complex overcurrent conditions or potential failure scenarios before hardware is built. In particular, modern simulation tools such as QSPICE make it possible to evaluate large parameter sets quickly, allowing designers to explore how protection devices behave under a wide range of electrical and thermal conditions.

However, modeling these devices accurately raises important questions. Designers must consider whether the model includes the large tolerances inherent to these components, whether ambient temperature influences the results, whether device aging after a trip event is represented, and how trip-time curves versus fault current are incorporated.

All these concerns are valid. If models are to be used effectively, these factors must be addressed so that simulations produce results that reflect real-world behavior.

PPTC SPICE model genesis and thermal fundamentals

From a functional perspective, a PPTC device can be approximated as a thermal mass with a baseline electrical resistance at ambient temperature. When a fault current flows through the device, resistive heating raises the internal temperature. Once the temperature reaches the trip temperature (Ttrip), the device undergoes a rapid increase in resistance. This sharply reduces the current and stabilizes the device at an elevated temperature.

The heat balance of the system can be described by the following equation:

In this equation:

• Cth represents thermal capacitance
• dT/dt represents rate of temperature change
• D represents thermal conductivity constant
• Tamb and T represent ambient and device temperature
• R(T) represents temperature-dependent resistance
• I represents fault current

Before tripping, the resistance can be approximated as a linear function of temperature using a temperature coefficient:

From equations (1) and (2), we derive and simplify:

In fact, here 𝑇𝑎𝑚𝑏 = T25 = 25°C

By integrating equation (3) from t = o to ttrip and from T = 25°C to T= Ttrip, we get:

When the theoretical curve derived from this model is plotted and compared with measured device data, differences become apparent, as shown in Figure 1.

Figure 1 The theoretical trip-time versus fault-current curve is compared with representative measured data. Source: Vishay

The simplified single thermal block model reveals clear limitations at both low and high fault currents. At high currents, the ideal slope of ln (trip time) versus ln (fault current) should be −2, because trip time is inversely proportional to dissipated power, which scales with the square of current.

In practice, the observed slope is significantly flatter. This indicates that additional physical mechanisms influence the thermal response at high current levels. At low fault currents, the theoretical curve rises more steeply than measured results. This discrepancy arises because a PPTC cannot be modeled as a single thermal mass dissipating heat through a single thermal resistance.

Additional factors include:

• Heat dissipation through solder joints and PCB copper
• Delayed thermal propagation in the polymer structure
• Resistance changes caused by device aging

Practical SPICE modeling of PPTC devices

Under lower current conditions, the thermal behavior of a PPTC device can be represented more accurately using a multi-stage thermal network. An example of this approach is shown in Figure 2, where a three-stage thermal Cauer network represents heat flow inside the device.

Figure 2 Here is an example of a three-stage thermal Cauer network representing heat flow within a PPTC device installed in an application circuit. Source: Vishay

In this model:

• Tpptc represents temperature of internal polymer material
• Telectrode represents temperature of device electrodes
• Tpcb represents temperature at PCB solder joint
• Tambient represents far-field ambient temperature

Each stage includes thermal resistance and thermal capacitance elements that model heat transfer between nodes.

The simulations described in this work were implemented in QSPICE, which provides powerful parameter-sweep capabilities. This allows multiple device and environmental parameters to be varied simultaneously, enabling thousands of simulations to be performed in a short time.

Once the thermal parameters are adjusted to match measured trip-time behavior, the simulated results closely reproduce measured data. The comparison between simulation and measurement is shown in Figure 3.

Figure 3 Simulated trip-time versus fault-current characteristics is compared with measured data. Source: Vishay

Remaining deviations are largely attributed to measurement conditions and inherent device tolerances. Nevertheless, the multi-stage thermal model provides a substantial improvement over the simplified single-block model.

Additional model features can also be incorporated, including:

• Resistance aging following a trip event
• Failure behavior when applied voltage exceeds rated limits
• Ambient temperature dependence

Influence of ambient temperature

Ambient temperature has a strong influence on PPTC device behavior. The simulation circuit used to investigate this behavior is shown in Figure 4:

Figure 4 An example application circuit is used to evaluate PPTC behavior under surge conditions at different ambient temperatures. Source: Vishay

Two successive voltage pulses are applied under varying temperature conditions. The resulting waveforms are shown in Figure 5.

Figure 5 Simulation results display device response to successive voltage pulses at different ambient temperatures. Source: Vishay

Here, several effects can be observed:

• Trip time varies with ambient temperature
• Device resistance increases following the first trip event
• Excessive voltage may force the device into a short circuit state

Because of this temperature dependence, datasheets typically provide thermal derating curves describing how allowable current varies with ambient temperature. An example of such a curve is shown in Figure 6.

Figure 6 A typical thermal derating curve illustrates how allowable hold current decreases as ambient temperature increases. Source: Vishay

The behavior behind this curve can also be reproduced through simulation. The circuit used for this analysis is shown in Figure 7, and the corresponding simulation results appear in Figure 8.

Figure 7 A simulation circuit is used to evaluate device behavior under varying load resistance and ambient temperature conditions. Source: Vishay

Figure 8 Simulation results display trip-time behavior as temperature and load conditions change. Source: Vishay

From these results, the variation of fault current as a function of ambient temperature can be extracted. The resulting simulated derating curve is shown in Figure 9.

Figure 9 The simulated thermal derating curve is derived from the SPICE model. Source: Vishay

Application simulations

Once realistic device models are available, they can be used to simulate complete application circuits.

USB power supply protection

A typical USB protection circuit is shown in Figure 10.

Figure 10 The application circuit illustrates a USB power supply protected by a PPTC device. Source: Vishay

Simulation results for plug-in disturbances and short circuit events are shown in Figure 11.

Figure 11 Simulation results show device response during plug-in events and short circuit disturbances. Source: Vishay

At 100 ms, a load short circuit is introduced. The PPTC device transitions from a low resistance state to a high resistance state, limiting the current.

Motor stall protection

An example motor protection circuit is shown in Figure 12.

Figure 12 The circuit example of a small electric motor is protected against stall conditions using a PPTC device. Source: Vishay

Simulation results demonstrating the stall event appear in Figure 13.

Figure 13 Simulation results show motor current increase and protection device trip during a stall event. Source: Vishay

When the motor stalls, I(Vspeed) drops to 0, and the current rises sharply. The protection device heats and transitions to a high resistance state, limiting current and preventing thermal damage.

Lithium-ion battery short circuit protection

The simulated battery protection circuit is shown in Figure 14, with results presented in Figure 15.

Figure 14 The simulation circuit represents a lithium-ion battery system with short circuit protection. Source: Vishay

Figure 15 Simulation results display surge current and resulting resistance increase after a load short circuit. Source: Vishay

At lower ambient temperatures, the device responds more slowly, allowing higher surge currents before tripping.

QSPICE modeling upside

The SPICE modeling approach described here demonstrates that realistic PPTC device behavior can be reproduced by fitting model parameters to datasheet characteristics and incorporating accurate thermal representations.

Using a simulation environment such as QSPICE enables designers to explore these behaviors efficiently through extensive parameter sweeps and transient analysis. The models can account for:

• Resistance tolerance
• Aging effects
• Ambient temperature dependence
• Overvoltage behavior

By incorporating these models into full application simulations, designers gain insights that cannot be obtained from datasheets alone. Nevertheless, simulation should always be complemented by laboratory validation before releasing a design into production.

Alain Stas is head of product marketing for non-linear resistors at Vishay.

Related Content

• QSPICE: Transient Domain Analysis
• QSPICE Simulator for Electronic Circuits
• QSPICE: A Simulation Tool for Power Circuits
• QSPICE: Introduction to the Simulation Environment
• QSPICE Revolutionizes Power, Analog Device Circuit Simulation

The post From specification to simulation: Modeling PPTC devices in QSPICE appeared first on EDN.

A clever millimeter-wave lens enables a high-speed, backscatter-powered GHz-band link.

Wireless system designers are often asked to deliver on seemingly incompatible and contradictory goals such as supporting ultrahigh wireless data rates, and do so at ultralow power. Accomplishing this, even if possible, is a challenge which may require lots of technical “tricks”  including advanced techniques, custom components, and more.

Now, a team at Georgia Institute of Technology (better known as Georgia Tech) has demonstrated a what they call a first-of-its-kind lens-enabled backscatter system capable of multi-gigabit data rates. At the same time, this backscatter-powered system operates using only a fraction of the power required by conventional wireless devices, therefore bringing high-speed connectivity to disbursed systems.

In general, due to power constraints, backscatter has typically been used only to send small amounts of data, most often in simple identification and sensing systems. However, the researchers say that backscatter doesn’t have to be slow and can operate at gigabit‑per‑second speeds while remaining ultra‑low power—with the right architecture. They foresee applications such as battery-free sensors embedded throughout smart cities and with digital infrastructure for a localized IoT arrangement.

Their lens-enabled backscatter system is capable of multi-gigabit data rates, reaching up to 4 gigabits per second (Gbps) (Figure 1). This dielectric lens focuses incoming millimeter-wave energy (such as from 5G systems) onto an array of tiny antenna elements, allowing both wireless energy capture and high‑speed backscatter communication within the same system.

Figure 1 A close‑up view of the device displays an array of tiny antenna elements positioned behind the lens, each modulating reflected wireless signals to enable high‑speed communication with minimal energy use. (Image source: Georgia Tech School of Electrical and Computer Engineering)

Signals at these frequencies are highly directional and sensitive to alignment; even a small misalignment can break the link. Their lens overcomes that constraint by enabling high gain and wide angular coverage simultaneously, without the need for active beam steering.

The system that can communicate over a ±55-degree field. In their tests, the researchers achieved data rates of up to 4 Gbps with sustained gigabit communication at distances of up to 20 meters, using high-order modulation schemes like those used in modern cellular networks. The system is extremely efficient and requires just 0.08 picojoules per bit. The link-budget analysis projects 1 Gbps back-scatter ranges up to 2.6 km under the 75 dBm effective Isotropic radiated power (EIRP) that is permitted in 5G millimeter-wave systems.

At the core of the millimeter-wave identification (mmID) is a broadband, cross-polarized antenna designed to operate across the full 26–30 GHz band. A broadband element is essential to sustain gigabit-level backscatter, since narrow- band operation would constrain throughput and increase distortion under high-order modulation. Cross-polarization is critical at mmWave, as a co-polarized backscatter would be masked by strong transmitter-receiving coupling from the reader.

To meet these requirements, they implemented a single-layer, capacitively coupled patch antenna designed in CST Microwave Studio and fabricated on Rogers 3003 (εr = 3:00, tan δ = 0:0013), with thickness of 0.254 mm (Figure 2).

Figure 2 a) Layout of the cross-polarized capacitive-coupled patch antenna with dimensions W = 2.85 mm, LS = 1.1 mm, and GC = 0.12 mm. b) Measured vs. Simulated S11 results of the broadband antenna. c) Layout of the FET-based mmWave modulator with dimensions R1 = 1.11 mm and R2 = 1.24 mm. d) Measured vs. Simulated S21 results of the mmWave modulator network. e) Layout of the pixel backscatter element, comprised of the broadband antenna and FET-based wireless mixer. (Image source: Nature Communications)

Gigabit backscatter at mmWave frequencies requires an antenna module that delivers both high gain and wide angular coverage. A dielectric lens provides an efficient solution, acting as a passive focusing element that concentrates incident energy onto the pixel. A key contributor to this long-range performance is the PTFE dielectric lens, which passively concentrates incident mm-wave energy onto the pixel element in a manner analogous to an optical lens. To extend the single pixel design into a practical mmID with wide angular coverage, a 25-element broadband cross-polarized pixel array was implemented, arranged in three concentric rings with a central element (Figure 3).

Figure 3 a) Proposed broadband cross-polarized mmID featuring 25 antenna elements with dimensions L1 = 26 mm, L2 = 52 mm, L3 = 78 mm, W = 90 mm, S = 13 mm, and R3 = 1.35 mm. b) Proposed PTFE lens with dimensions labeled D1 = 74 mm, D2 = 120 mm, and h = 25 mm. (Image source: Nature Communications)

The team performed extensive tests spanning a range of frequency bands, data formats, modulation types, and more, with detailed quantitative results summarized in various tables (Figure 4). They have shown that it is possible to extract GHz-range ambient-RF energy effectively using a printed lens-like antenna.

Figure 4 a) Experimental setup of the proposed lens-based mmID at incidence angles of 0∘ and 55∘ from the PoC reader. b) Block diagram of the PoC reader transmitting and receiving chain to interrogate the lens-based mmID and demodulate the gigabit per second data-rate backscatter. (Image source: Nature Communications)

The detailed project is a fascinating investigation and exploration into RF-based energy harvesting and ultralow-power systems design, without speed compromise. It is described in detail in their readable paper “Broadband multi-beam lens-assisted mmID enabling multi-gigabit backscatter data rates for next-generation wireless networks” published in Nature Communications.

What’s your view on the innovation and cleverness of this project? Is it as impressive as they maintain, or just a well-crafted and analyzed implementation of existing ideas? Is it yet another attention-getting energy-harvesting scheme with added gigahertz connectivity, or does it represent a genuine advance?

—Bill Schweber is a degreed senior EE who has written three textbooks, hundreds of technical articles, opinion columns, and product features. Prior to becoming an author and editor, he spent his entire hands-on career on the analog side by working on power supplies, sensors and signal conditioning, and wired and wireless communication links. His work experience includes many years at Analog Devices in applications and marketing, and he also developed significant mechanical-engineering insight while designing control electronics for large materials-testing systems.

Related Content

1. Nothing new about energy harvesting
2. An energy-harvesting scheme that is nearly useless?
3. Is triboelectricity the next harvesting source?
4. Tilting at MEMS Windmills for Energy Harvesting?

The post GHz-harvested power for Gbps wireless link is a double win appeared first on EDN.

Artificial intelligence and machine learning (AI/ML) are increasingly being integrated into medical systems. This delivers smarter and faster care by bringing intelligence closer to where the data is created and used.

The most recent advances in computer vision, large language models, edge computing, and real-time signal processing are improving medical diagnosis and reducing the latency between data acquisition and actionable clinical results. This is enhancing medical imaging use cases and enabling advances in robotic surgery and remote monitoring while delivering more integrated systems to improve patient care.

We introduce some real-world products and solutions that demonstrate how the combination of AI/ML and healthcare has become a reality.

Medical imaging

In medical imaging, AI/ML techniques, including convolutional neural networks, are being deployed directly within magnetic resonance imaging (MRI) and computed tomography (CT) systems. Platforms such as Royal Philips’s AI-enhanced MRI and CT suites and Aidoc’s radiology triage are examples of how models trained using large datasets, helping to detect critical conditions with very low latency. Butterfly Network has further expanded this concept by embedding AI inference into portable, handheld ultrasound hardware.

Philips’s AI-enhanced MRI and CT

At the 2025 European Congress of Radiology, Royal Philips announced a generation of AI-integrated MRI solutions. The main innovation is SmartSpeed Precise, a system powered by dual AI engines built around the company’s existing Compressed Sense and SmartSpeed platforms.

SmartSpeed Precise improves image sharpness by 80%, enabling a better visualization of anatomical structures. Applied to standard Sense imaging, the technology allows scans to be completed in one-third of the time without affecting the quality of the diagnostic images.

A faster scan time means a more comfortable experience for the patient, with less time spent without moving inside the scanner. Moreover, patients can access a diagnosis more quickly, as wait times are reduced. One clinical site reduced exam times to under 60 minutes per slot for whole-body multiparametric exams, enabling it to scan two additional patients per day.

Recently, Royal Philips also received FDA 510(k) clearance for its Verida system, a next-generation spectral CT scanner that integrates AI-driven reconstruction with a specialized dual-layer detector architecture. At the core of the Verida system is a third-generation Nano-panel Precise dual-layer detector.

Unlike photon-counting detectors that utilize direct conversion, this scintillator-based stack employs two layers (the Nano-panel Precise detector) to capture low- and high-energy photon spectra from a single X-ray source, providing spectral results 100% of the time. This “always-on” technology enables spectral imaging capabilities to be active for every patient, on every scan, without requiring special procedures or separate, time-consuming scans.

Inside the signal-processing chain, the system uses a deep-learning reconstruction engine, a properly trained neural network that provides an estimated 80% reduction of the image noise, maintaining the spatial resolution. The computational back end can handle high-throughput processing, up to 145 images per second, enabling full-volume spectral data processing in under 30 seconds, 2× faster than previously available technology.

Figure 1: Royal Philips’s spectral CT Verida system (Source: Royal Philips) Aidoc’s AI-powered clinical platform

Aidoc’s AI-powered platform, adopted by over 1,600 hospitals worldwide, is built around the idea that connected teams deliver better outcomes. At the heart of the solution is aiOS, the company’s proprietary enterprise platform, which operates as an always-on intelligence layer across a health system. The platform covers 75% of the patient population, according to the company, enabling physicians to make accurate decisions using real-time data and allowing care teams in multiple departments to collaborate on a unified patient journey.

The clinical solutions have five core specialties. Radiology solutions include image-based triage and quantification, powered by 18 FDA-cleared algorithms and eight FDA-cleared partner algorithms. Beyond imaging, Aidoc extends into cardiology, neurovascular care (including stroke and brain aneurysm detection), vascular conditions such as pulmonary embolism and aortic disease, and spine solutions.

A mobile care coordination app delivers real-time notifications for time-sensitive cases, built-in risk stratification, and a mobile imaging viewer, with electronic health-record data automatically fed in to facilitate communication between divisions. Adopted by several leading health systems, Aidoc has delivered significant improvements, including turnaround-time reductions of up to 55% for intracranial hemorrhage cases and length-of-stay reductions of up to 26% for pulmonary embolism cases.

Aidoc’s AI-powered triage platform for large vessel occlusions and medium vessel occlusions proved effective in a study presented at the International Stroke Conference 2026.

In a comparative study of 1,557 CT angiography exams by the University of Texas Medical Branch, Aidoc also showed 92.6% sensitivity for large vessel occlusions, a rate significantly higher than the 70.4% offered by traditional solutions.

Butterfly Network’s ultrasound technology

Butterfly Network, a company specializing in ultrasound technology, has received FDA 510(k) clearance for its Gestational Age (GA) Tool, the first “blind-sweep” AI-powered ultrasound application for pregnancy dating. The technical innovation consists of replacing traditional piezoelectric transducer arrays with Ultrasound-on-Chip technology, which integrates a complete ultrasound system onto a single CMOS chip.

The GA Tool employs a deep-learning inference engine trained with a dataset of >21 million ultrasound images. In contrast to conventional fetal biometry that requires precise manual alignment and measurement of the biparietal diameter or femur length by an experienced sonographer, the “blind-sweep” method employs a simplified acquisition protocol. The operator performs guided probe sweeps over the maternal abdomen without the need to interpret images in real time or optimize for targets.

The AI algorithm then examines the resulting volumetric data to estimate the gestational age between 16 and 37 weeks. The system learns the mapping of anatomical landmarks to gestational maturity to ensure high fidelity of diagnosis and provides results similar to traditional biometric assessments in less than two minutes.

Robotic surgery and remote monitoring

In the field of surgery, robotic platforms such as Intuitive Surgical Operations Inc.’s da Vinci system include real-time haptic feedback loops and computer-assisted motion scaling to minimize the distance between the surgeon’s input and the end-effector’s output. Edge ML models on devices such as DexCom Inc.’s G7 and Eko Health Inc.’s cardiac sensors analyze continuous streams of biosignals locally, sending only clinically relevant anomalies, as in remote monitoring.

Intuitive’s da Vinci 5

It is one of the most advanced robotic systems for minimally invasive surgery. The name is no coincidence. Leonardo da Vinci was the first to study the movements of the human body systematically and, in 1495, to design a prototype of a humanoid robot, the mechanical knight (called automa cavaliere in Italian).

With a processing capacity 10,000× greater than the previous Xi generation, the da Vinci 5 supports a suite of digital and tactile features intended to improve surgical precision through real-time data analysis and haptic integration.

A major technical upgrade is the introduction of Force Feedback technology. It uses force-sensing instruments and new algorithms to measure the physical resistance that the robotic arms encounter. This information is then fed back to the surgeon’s console so that the surgeon can “feel” tissue tension and resistance. To complement this haptic information, the system also incorporates a Force Gauge, a real-time visual display of pressure information.

The Intuitive Hub and its ML models provide the AI capabilities of the system. These models produce automated case insights by algorithmically dividing surgical video into discrete phases such as dissection, retraction, and suturing. The system provides objective performance metrics based on the instrument kinematics and phase durations. These metrics can be used to compare individual surgical techniques against standardized clinical benchmarks, with a focus on motion efficiency and instrument economy.

During active procedures, the da Vinci 5 utilizes predictive analytics and vision-based AI to enhance the surgical field. The in-console video replay feature allows surgeons to access recorded segments of the ongoing procedure directly through the 3D viewer. This function is supported by AI overlays that can highlight specific anatomical landmarks or track instrument paths.

Figure 2: Intuitive’s complete da Vinci 5 system, consisting of the tower, generator, console, and patient cart (Source: Intuitive Surgical Operations Inc.) Dexcom G7

The Dexcom G7, a continuous glucose monitor (CGM), has evolved into an AI-driven health device. According to recent updates from DexCom, the G7 (Figure 3) now integrates sophisticated AI to simplify daily diabetes management and deliver deeper metabolic insights.

A relevant AI feature is Smart Food Logging. This tool uses computer vision and ML to analyze photos of meals taken within the app. The AI automatically identifies ingredients and generates meal descriptions, significantly reducing the manual effort required for carb counting and data entry.

Furthermore, DexCom has launched a proprietary generative AI platform powered by Google Cloud’s Vertex AI and Gemini models. This platform analyzes individual health data patterns to offer personalized “Weekly Insights”—that is, recommendations based on the user’s glucose trends, activity levels, and sleep patterns.

Figure 3: The Dexcom G7 15-Day CGM has recently received FDA clearance for people aged 18 years and older with diabetes. (Source: DexCom Inc.) Eko Health Sensora platform

Eko Health has achieved a major milestone with the FDA clearance of its Eko Foundation Analysis Software with Transformers (EFAST) algorithm, the first-ever cardiac “foundation model” designed for clinical use. This AI has been integrated into the Sensora platform, Eko’s digital stethoscope, which helps clinicians detect signs of potential cardiac diseases quickly and accurately (Figure 4).

The EFAST algorithm integrates several advanced AI features that transform the diagnostic process. Instead of being trained for a single purpose, the system utilizes a cardiac foundation-model architecture trained on over 4 million heart-sound and electrocardiogram recordings. This large dataset allows the AI to learn a general representation of cardiac health, which is then fine-tuned to detect specific conditions, such as structural heart murmurs and atrial fibrillation, with high specificity.

Figure 4: Using Eko Health’s Sensora platform, clinicians receive automated alerts for structural heart murmurs and low ejection fraction, a sign of a weakened heart pump. (Source: Eko Health Inc.)

The post AI/ML integration in medical systems appeared first on EDN.

This week’s latest iteration of Google’s yearly developer event reiterated the company’s significant AI commitment. What’s different from messaging and examples past? Maturity.

One of the technologies showcased in the most recent edition of my previous-year retrospective series, published on New Year’s Day, was agentic AI. An overview excerpt from that earlier coverage follows:

Here’s what Wikipedia says about AI agents in its topic intro:

“In the context of generative artificial intelligence, AI agents (also referred to as compound AI systems or agentic AI) are a class of intelligent agents distinguished by their ability to operate autonomously in complex environments. Agentic AI tools prioritize decision-making over content creation and do not require human prompts or continuous oversight.”

And what about the aforementioned broader category of intelligent agents, of which AI agents are a subset? Glad you asked:

“In artificial intelligence, an intelligent agent is an entity that perceives its environment, takes actions autonomously to achieve goals, and may improve its performance through machine learning or by acquiring knowledge. AI textbooks define artificial intelligence as the “study and design of intelligent agents,” emphasizing that goal-directed behavior is central to intelligence. A specialized subset of intelligent agents, agentic AI (also known as an AI agent or simply agent), expands this concept by proactively pursuing goals, making decisions, and taking actions over extended periods.”

A recent post on Google’s Cloud Blog included, I thought, I concise summary of the aspiration:

“Agentic workflows represent the next logical step in AI, where models don’t just respond to a single prompt but execute complex, multi-step tasks. An AI agent might be asked to “plan a trip to Paris,” requiring it to perform dozens of interconnected operations: browsing for flights, checking hotel availability, comparing reviews, and mapping locations. Each of these steps is an inference operation, creating a cascade of requests that must be orchestrated across different systems.”

I suggested there that last year’s rapid evolution of agentic AI technology and products based on it wasn’t a one-off; that the maturation and proliferation trends would undoubtedly continue in the coming year and beyond. We’re nearing the 2026 mid-point and, judging from what Google showcased at yesterday’s keynote, I wasn’t offbase with my earlier perpetuation prediction:

But I’m getting ahead of myself…

Android Show: I/O Edition 2026

Google extended the trend it initiated last year by delivering a separate Android-specific showcase one week ahead of the main event:

Company representatives covered a lot of ground in only a bit more than a half hour, including pending enhancements to Android Auto and “Continue On”, an in-beta conceptual clone of Apple’s Handoff. But two other topics particularly caught my eye. Generally speaking, Google is fundamentally integrating Gemini Intelligence even more than previously into the core of both Android and its Chrome browser, including both anticipatory awareness of what you might need next and the agentic “chops” to independently (potentially) tackle such tasks on your behalf.

The central reason why I find this trend interesting is contextual in nature. Both Amazon (again) and OpenAI are reportedly working on smartphones based on brand new AI-based—specifically agentic, generative and personalized—operating systems. Going “clean slate” from a software standpoint does have at least some advantages, conceptually speaking at least, but it also tends to result in a “heavy lift” with respect to application development, internally and especially from a third-party standpoint. Conversely, Google’s building on a longstanding Android foundation.

Consider that contrast, too, in the context of the other key Android Show tidbit that I want to pass along today. Confirming longstanding rumor, Google announced that it is seriously re-engaging in the tablet market with Android (where, to clarify, it remains a “player” today, primarily courtesy of its Samsung partnership, albeit on a limited basis versus Apple iPad alternatives), as well as expanding Android into computing form factors that were traditionally serviced by Chrome OS, all with a new operating system version code-named “Aluminum”.

The coexistence of the two operating systems had always been awkward at best. They’re both built on a Linux foundation, but that’s kind of like saying that a Trabant and a Ferrari both hail from a Ford Model T heritage. I’m not trying here to infer a vehicle-analogous comparison between the two operating systems with respect to “sleekness”, price or anything like that, only generally proffering that they’re notably dissimilar. Different code bases, different development teams and schedules…over time, Android and Chrome OS had increasingly diverged, to their shared detriment.

And what does “Aluminum” mean for Chrome OS fortunes long-term? Unclear; the latter’s only notable success has been in the education market, but it’s been a notable success there, so Google needs to be careful about how it hand-holds these key customers during the transition (which I’d suggest is a matter not of if, but when). Event-delivered reassurances included that support-timeframe schedules for existing Chrome OS-based products would continue to be honored in full, that new Chrome OS-based products were still in the development pipeline from partners, and that at least some existing Chrome OS-based hardware would be upgradeable to whatever the marketing moniker for “Aluminum” ends up being. That said, if new Chrome OS hardware is still being announced when the decade turns in a few years, I’ll be shocked.

Foundation AI evolutions

Now for the main event. AI has been front and center in Google I/O messaging for a while now, as The Verge and I joked about two years back:

@verge

Pretty sure Google is focusing on AI at this year’s I/O. #google #googleio #ai #tech #technews #techtok

♬ original sound – The Verge

And it was more of the same this year. For those of you who’ve been wondering what the term “foundation model” (or variants of that name) means, I’ll start out with a Wikipedia-sourced definition:

In artificial intelligence, a foundation model (FM), also known as large x model (LxM, where “x” is a variable representing any text, image, sound, etc.), is a machine learning or deep learning model trained on vast datasets so that it can be applied across a wide range of use cases. Generative AI applications like large language models (LLM) are common examples of foundation models.

Building foundation models is often highly resource-intensive, with the most advanced models costing hundreds of millions of dollars to cover the expenses of acquiring, curating, and processing massive datasets, as well as the compute power required for training. These costs stem from the need for sophisticated infrastructure, extended training times, and advanced hardware, such as GPUs.

This all in contrast to dataset- and application-specific models. Wikipedia again:

Adapting an existing foundation model for a specific task or using it directly is far less costly, as it leverages pre-trained capabilities and typically requires only fine-tuning on smaller, task-specific datasets.

Last year at I/O, Google shared updates on v2.5 of its Gemini model family (standard, Flash and Pro), which had been introduced a few months earlier. Gemini v3 subsequently arrived last November. And now we’re up to Gemini family v3.5. Commensurate with the update, another term is in circulation for us to sort out: “Frontier model”. NVIDIA with the definition this time:

Frontier models are the most advanced AI models available at a given moment, trained on massive datasets to deliver state-of-the-art performance across many tasks, representing the leading edge of AI capability. They typically power advanced reasoning, image and text generation, and agentic workflows.

Translation: a fancy way of saying “next generation”. Gotta love those marketeers.

More generally, snark aside, I admittedly was particularly gob smacked by this subset of the event-opening keynote remarks by CEO Sundar Pichai:

These stories of how people are using AI are the best measure of progress. To understand the scale at which people are adopting AI, there is another great proxy — tokens, the fundamental units of data our models process, many representing a problem being solved.

Two years ago, we were processing 9.7 trillion tokens a month across our surfaces — a huge number. Last year at I/O, that grew to roughly 480 trillion tokens. Fast forward to today, that number jumped 7x to over 3.2 quadrillion per month. [Editor note: token maxxing? Likely, to a degree. Still…]

It tells an important story about our products and how others are building as well — especially developers and enterprises:

• Over 8.5 million developers are now building new apps and experiences with our models monthly.
• Our model APIs are now processing roughly 19 billion tokens per minute.
• Over the past 12 months, over 375 Google Cloud customers each processed more than one trillion tokens, representing incredible demand for AI from across industries.

Today we have 13 products with over a billion users each. Five of those have more than 3 billion users. [Editor note: and they’re all AI-enhanced, if not AI-centric]

Multimodal and agentic enhancements

Back in December 2024, within a broader attempt to forecast the year to come, I opined:

Large language models (LLMs), which I rightly showcased at the very top of my 2023 retrospective list, are increasingly impressive in their capabilities. But they’re also, admittedly somewhat simplistically speaking, “one-trick ponies”. As their name implies, they’re language-based from both input (typed) and output (displayed) standpoints. If you want to speak to one, you need to first run the audio through a separate speech-to-text model (or standalone algorithm); the same goes for spitting a response back at you through a set of speakers. Analogies to images and video clips, and other sensory and output data, are apt.

Granted, this approach is at least somewhat analogous to human beings’ cerebral cortexes, which are roughly subdivided into areas optimized for language, vision and other processing functions. Still, given that humans are fundamentally multisensory in both input and output schema, any AI model that undershoots this reality will be inherently limited. That’s where newer multimodal models come in. Vision language models (VLMs), for example, augment language with equally innate still and video image perception and generation capabilities. And large multimodal models (LMMs) are even more input- and output-diverse. Think of them as the deep learning analogies to the legacy sensor fusion techniques applied to traditional processing algorithms, which I ironically alluded to in my 2022 retrospective.

Enter the new Gemini Omni multimodal model:

Last year, Nano Banana brought Gemini’s intelligence to image generation and editing. Since then, it’s helped millions of people restore old photos, design from sketches and visualize ideas in ways that weren’t possible before. From the start we built Gemini to be natively multimodal from the ground up, and now we’re taking the next step.

We’re introducing Gemini Omni, where Gemini’s ability to reason meets the ability to create. Omni is our new model that can create anything from any input — starting with video. With Omni, you can combine images, audio, video and text as input and generate high-quality videos grounded in Gemini’s real-world knowledge. You can also easily edit your videos through conversation.

Today, we’re rolling out the first model in the Omni family: Gemini Omni Flash, to the Gemini app, Google Flow and YouTube Shorts. In time we will support output modalities like image and audio.

And what about burgeoning agentic AI assistants such as OpenClaw? How’s that saying go—”Imitation is the sincerest form of flattery”—albeit this time with innate Google services and account-data access?

We’re also introducing Gemini Spark, a 24/7 personal AI agent that helps you navigate your digital life. Spark represents a big shift for Gemini, transforming it from an assistant that can answer your questions into an active partner that does real work on your behalf and under your direction.

Gemini Spark runs on Gemini 3.5 and uses the Antigravity harness. It’s deeply integrated with the Workspace tools you rely on daily, like Gmail, Docs, Slides and more. Even better, because it is a cloud-based agent, Spark keeps working in the background even when you close your laptop or lock your phone. That combination means Spark is ready to take complex tasks off your plate so you can be more present for what matters most.

“Intelligent Eyewear”

Last but not least, a few words about head-located wearables, including those with integrated displays. Google seems to be reluctant to refer to them as “smart glasses” (or VR headsets, for that matter). Gee, I wonder why? And why? Snark off (again). As regular readers may already recall, I’ve been following this market quite closely in recent years, even personally investing in a few trendsetting product examples. And we’ve in-parallel been hearing about (and I’ve been writing about) Google’s Android XR operating system and application suite for augmented, virtual and hybrid reality systems for a while now, too.

Well, the reality behind the hype is finally coming to market starting this fall. Supposedly. Conceptually, they sound a lot like Meta’s counterparts (albeit perhaps a bit sleeker) which I’d suggest have been meaningful from an implementation standpoint since at least the October 2023 unveil of the second-generation AI Glasses. That said, Meta’s success has to date been held back by (among other factors) a dearth of third-party support. Here’s a reality calibration: even if Google and partners’ competitive devices are no better off in this regard, their inherent coordination with the aforementioned “Google services and account data” will still give them a “leg up”. More generally, you’ve got to admit this was one heck of a compelling live demo suite:

We shall see.

Wrapping up

There was plenty more interesting news released at the Tuesday keynote and more broadly across the two-day event (which is still underway as I type these words mid-day on Wednesday). Browse other writeups on the Google event portal page, along with coverage at 9to5Google and elsewhere. And then share your thoughts with me and your fellow readers in the comments!

The post Google I/O 2026: Agentic AI gets serious appeared first on EDN.

Portable power station design apparently isn’t easy. So suggests this engineer’s most recent two case studies, not to mention the long-term history (kudos to Kurt Vonnegut for the inspiration).

My first writeup last month discussed why one battery (or a few) in a larger cluster always seems to drain faster than the others, and how this imbalance would affect the system powered by that multi-battery cluster. The answer depended in part on whether the batteries were connected in a series, parallel or a hybrid combo of the two, to “boost the effective voltage (serial) and/or increase overall system runtime (parallel)”. Here’s how I concluded that piece:

I’m covering today only situations where the installed batteries are either non-rechargeable or are removed for recharging. Multi-battery packs recharged in situ (while installed inside a portable power unit, for example) translate to an even more complicated scenario involving, among other factors, the critical importance (and difficulty) of balancing the various cells within the likely series/parallel cluster.

At that point in time, for readers who wanted to know more about this topic, I referenced a Vitron Energy white paper I’d previously recommended in that same piece for other reasons, which also explored this topic at length. Here’s an example of what I’m talking about in the form of a quote from that same white paper:

If a large battery bank is needed, we do not recommend that you construct the battery bank out of numerous series/parallel 12V lead acid batteries. The maximum is at around 3 (or 4) paralleled strings. The reason for this is that with a large battery bank like this, it becomes tricky to create a balanced battery bank. In a large series/parallel battery bank, an imbalance is created because of wiring variations and slight differences in battery internal resistance.

And later in that section, as further elaboration:

When creating a lead-acid battery bank with a higher voltage, like 24 or 48V you will need to connect multiple 12V batteries in series. But there is one problem with connecting batteries in series, and this is that batteries are not electrically identical. They have slight differences in internal resistance. So, when a series string of batteries is charged, this difference in resistance will cause a variance in terminal voltages on each battery. Their voltages become “unbalanced”. This “unbalance” will increase over time and will lead to one of the batteries being constantly overcharged while the other battery is constantly undercharged. This will result in a premature failure of one of the batteries in the series string.

Again, I commend the entire white paper to your attention, not only because it delves in greater detail into the topics discussed in the two excerpts I selected but also because in doing so, it covers not only lead-acid but also lithium-based and other emerging chemistries such as “flow”.

I wrapped up that prior writeup by saying that “I’ll likely have more to say about these topics in future posts as well.” I didn’t necessarily think at the time that I’d be revisiting rechargeable in situ batteries this quickly…but then again, I also didn’t think that, a short time later, I’d personally experience what I still suspect was the outcome of an unbalanced multi-cell battery bank (along with a couple of other functional hiccups, details of which I’ll also share).

The DOA DELTA 3 Smart Extra Battery

I’ve had my EcoFlow DELTA 3 “stack”, the combo of a DELTA 3 Plus and its companion Smart Extra Battery, for one day shy of a year as I write these words in late April:

Until recently, all’s been well. It had only received a couple of firmware updates, all of which had been drama-free. And although it won’t seemingly power my refrigerator reliably, Xcel Energy’s increasingly frequent (or at least so it seems) power outages have provided plenty of other opportunities for me to tap into its stored-electron stash. The most recent outage (again as I write this …another is sooner-or-later-likely-sooner inevitable) in mid-March thankfully lasted only a bit more than seven hours, not several days, but alas, the “stack” didn’t survive it.

During the outage, I’d dragged upstairs its DELTA 2 base unit-plus-smart extra battery “stack” siblings to run an interior lights and recharge flashlights and various mobile devices:

When the utility company-sourced premises electricity came back up just after midnight, I took the DELTA 2 “stack” back downstairs to the workbench in the furnace room, its normal on-standby location. I’d left the DELTA 3 gear there; the base unit’s front panel display was now illuminated but that of the smart extra battery wasn’t, nor seemingly was the latter more broadly functional any longer. And looking more closely, I noticed an “Error 726” indicator on the base unit display that I’d never seen before:

I hit up Google search for suggestions, which were scant, dubious (I don’t think “turn off the unit and wait a few hours for the cells to rebalance themselves” makes much if any sense, particularly given that the only way to turn the unit off is to unplug it first) and more generally indeterminate save for the revelation that Error 726 indicates that “cell voltage differences are too great”. My next and increasingly common step was to publish a Reddit post. One respondent pointed me toward some Facebook group traffic that I’d already come across. Another noted that he/she had experienced the same issue after a recent firmware update, which I’d also done. And a third offered a “Possible it’s a bad cell” suggestion. Hold that thought.

Base unit BMS reset attempts were ineffective; it was also running the latest-available firmware:

I hooked up a fan to one of the AC outputs to drain the base unit’s batteries, in the hope that a full recharge might resurrect its cognizance of the smart extra battery. No dice; the base unit worked fine standalone but threw an Error 726 with the smart extra battery connected. So, I reached out to EcoFlow technical support, who confirmed that the smart extra battery had gone bad and offered to send me a replacement in exchange for my failed unit.

Two weeks later, the new smart extra battery was in my hands. Connecting it to the base unit initially resulted in the generation of another error code in the latter, the nebulous “Error 014”:

After a brief panic, and acting on a hunch, I checked to see if the base unit needed a firmware update before the two devices could be sympatico. Indeed, that was the case, two update cycles’ worth, in fact:

And, at least as of today, the setup once again seems to be working OK, leaving me with one lingering question: what went wrong in the first place?

• Was it a hardware failure, such as (but not necessarily) an unrecoverable cell-imbalance issue, in my specific original smart extra battery?
• Was there a broader fundamental hardware flaw in initial smart extra battery units, suggested by the prompt to do a firmware update (which I hadn’t seen before) when I plugged in the replacement?
• Did a firmware update to the base unit initiate the failure sequence in the first place, as the comments of one of the respondents to my Reddit post indicates might be the case?
• Or were the firmware-update timings (both prior to the original-unit failure and after installation of its replacement) purely coincidental?

Reader theories (and broader thoughts) are as-always welcome in the comments! And now, speaking of botched firmware updates…

A DELTA 2 double-whammy

Regular readers may recall that I’ve had issues after doing firmware updates on EcoFlow gear before. Generally speaking, I’ve therefore subsequently waited for an appropriate period, combing Reddit and relevant Facebook groups for posted evidence of others’ troubles, before taking the plunge myself. However, when I got prompted for an update to the DELTA 2 in late February, I (over)confidently decided to plunge ahead absent any preparatory research:

After all, the previous update I’d done to the DELTA 2 back in late December had gone well. I was so overoptimistic, in fact, that I updated my RIVER 2 at the same time:

The RIVER 2 survived the update just fine. The DELTA 2 on the other hand…

I admittedly didn’t immediately notice the issue, because it only happened occasionally. With the combo “awake”, everything seemed to be fine (well, mostly…keep reading). But at some random point after the units displays turned off (I stuck with the default “5 minute” setting), the power LED on the smart extra battery would extinguish, the base unit’s fan would kick on and perpetually run at low speed, and it would unceasingly (generally) or cyclically (briefly) draw ~20W of power from the AC outlet connected to it:

Punch either unit’s front panel power switch and the displays would wake up, the fan would stop and the trickle charge would cease…until after the displays turned back off again, that is. Lather, rinse, repeat. The trickle-charge behavior particularly worried me, because I didn’t want to end up with an overcharged, overheated and potentially exploded-and-burning battery situation on my hands. So, after several cycles of BMS resets and draining-then-recharging the battery sets, all of which “fixed” the issue only temporarily, I reached out to EcoFlow tech support once again with the proactive suggestion that a firmware downgrade might be in order.

They agreed. I never received (again, keep reading) their first attempt to “push” me a rollback from v1.0.2.176 to firmware v1.0.2.163, but the second attempt was successful:

The DELTA 2 “stack” is once again stable, at least from a charging standpoint. And although it took a while for another invitation to update back to firmware v1.0.2.176 to appear, leaving me wondering if either my “rollback” package had been customized to suppress the subsequent update or EcoFlow had pulled firmware v1.0.2.176 completely:

Turns out I just didn’t wait long enough (and yes, I declined when the update invitation eventually arrived):

Unfortunately, EcoFlow doesn’t publish firmware version histories for its devices, so I couldn’t check for an answer to my question that way. And customer service’s silence in (non-)response to my repeated queries about this particular topic weren’t helpful either (in contrast, I’m compelled to note, to their overall general excellent support).

Connectivity troubles

While I was in communication with EcoFlow, I also brought up an unrelated DELTA 2 issue that I’d been having on-and-off, albeit seemingly more frequently with the passage of time, with the base unit. After some random time period, hours-to-days after I’d established Wi-Fi connectivity, it’d drop Wi-Fi and revert to Bluetooth-only communication with my controlling smartphone:

Other times, even when Wi-Fi was supposedly still operational:

the DELTA 2 base unit, therefore entire “stack”, became “invisible” when I was attempting to reach it from outside my LAN via the EcoFlow “cloud” intermediary:

EcoFlow tech support suggested that the IoT module inside the unit might be gradually failing. I almost didn’t bother pressing the issue further—returning the unit for repair or replacement is something of a hassle, further complicated by the fact that the included flammable batteries mean that I can’t just drop it off at a FedEx Office location but need to arrange for front-door pickup, and then there’s the delay for a replacement unit to arrive—while the loss of Wi-Fi connectivity is annoying, it’s not a functional “death sentence”.

But then EcoFlow confirmed what I’d already suspected, that the IoT module also implements the Bluetooth subsystem, whose functional loss would completely sever further communication with the unit. Couple that with the fact that I’ve been promised a brand-new (not refurbished) replacement, with a zero-cycle fresh battery pack, and it was an offer I couldn’t refuse. I’m awaiting a return-shipping label as I type these words; I’ll report back on the status of the replacement unit via a posted comment on this post once published.

Design is hard

You may have already noticed a commonality to both primary issues noted in this writeup, as well as those in prior EcoFlow problem-themed coverage from me: the smart extra battery. I’m guessing that it’s relatively uncommon for base unit owners to also have this additional-charge (not to mention additional credit card charge) storage peripheral. As such, the prevalence of user problems is also likely to be uncommon. Therefore, I suspect, it’s relatively easy for smart extra battery firmware-related issues in particular to slip through any EcoFlow pre-release testing cracks.

In no way am I making excuses for the various EcoFlow issues I’ve come across; I’m just striving to be pragmatic about root causes. As I noted at the beginning of this writeup, battery pack design is fundamentally challenging. Make a device increasingly “smart” and the level of difficulty further ramps up. Is EcoFlow unique in this regard? I don’t know. I welcome feedback from owners of other manufacturers’ portable power stations (as well as from both EcoFlow and other manufacturers’ representatives themselves) regarding their comparative reliability. And I’d also appreciate insights from other EcoFlow owners re the commonality-or-not of their experiences. Sound off with your thoughts in the comments, please!

—Brian Dipert is the associate editor, as well as a contributing editor, at EDN.

Related Content

• Single-battery failures in multi-battery arrangements: diagnosing selective cell derangements
• EcoFlow’s DELTA 3 Plus and Smart Extra Battery: Product line impermanence curiosity
• EcoFlow’s Delta 2: Abundant Stored Energy (and Charging Options) for You
• Portable power station battery capacity extension: Curious coordination
• Firmware-upgrade functional defection and resurrection

The post More EcoFlow woes: So it goes appeared first on EDN.

Manufacturing is at a pivotal moment. Global supply-chain volatility, increasing energy costs, workforce shortages, and growing expectations for quality and customization are forcing factories to rethink how they operate. Traditional automation, optimized for predictability and repetition, struggles to cope with today’s variability and speed of change.

The smart factory represents a decisive shift: production environments that can sense, interpret, and adapt in real time. Central to this shift are three tightly connected technology domains: machine vision, intelligent sensing, and edge AI. Together, they enable factories not just to collect data, but to turn it into insight and action where it matters most.

Figure 1 The notion of smart factory marks a decisive shift in modern manufacturing. Source: Renesas

The limits of conventional automation

Conventional automation systems excel at executing predefined logic. However, they are inherently reactive. When processes drift, materials vary, or equipment degrades, intervention is often manual, time‑consuming, and costly.

Key pressures accelerating the move toward smarter automation include:

• Greater product diversity driven by mass customization
• Higher quality expectations that allow little tolerance for defects
• Skilled labor shortages across engineering and maintenance roles
• Soaring downtime costs, particularly in highly automated lines

Addressing these challenges requires automation systems that are more perceptive and context-aware systems capable of learning from data rather than simply enforcing rules.

Below is a quick recap of smart factory’s three key design building blocks: machine vision, intelligent sensing, and edge AI.

Machine vision: From inspection to interpretation

Machine vision is one of the most visible pillars of the smart factory. Once limited to basic presence checks or rigid defect criteria, today’s vision systems can interpret complex scenes and adapt to variation.

Seeing beyond pass or fail

Traditional, rule-based vision systems perform well under tightly controlled conditions but tend to break down when lighting, materials, or product designs change. Modern vision approaches increasingly incorporate learning-based techniques that recognize patterns instead of relying on fixed thresholds.

Figure 2 Modern vision systems recognize patterns instead of relying on fixed thresholds. Source: Renesas

This evolution enables machines to distinguish acceptable variation from true defects, adapt to new product versions with minimal retraining, and provide richer information for downstream decision-making.

Broader roles on the factory floor

Machine vision now plays a central role in:

• In-line quality assurance, detecting cosmetic, structural, and assembly issues
• Robot guidance, enabling flexible pick-and-place and assembly operations
• Traceability, supporting serialization and regulatory compliance
• Safety monitoring, detecting unsafe conditions or human proximity

As processing moves closer to where images are captured, vision becomes more responsive and resilient, key traits for real-time factory environments.

Figure 3 Machine vision technology is quickly acquiring the key traits required in real-time factory environments. Source: Renesas

Intelligent sensing: Adding awareness to automation

While machine vision provides visual insight, intelligent sensing fills in the rest of the picture. Parameters such as vibration, temperature, current, torque, pressure, and acoustics reveal what is happening inside machines and processes.

From measurement to meaning

Intelligent sensors are no longer passive components. Increasingly, they embed local processing and diagnostics, enabling them to filter and contextualize raw signals, detect subtle behavioral changes, and reduce unnecessary data transmission.

Instead of reporting isolated values, sensors can now indicate conditions such as early wear, imbalance, or inefficiency.

The power of sensor fusion

True process understanding emerges when multiple sensor types are combined. By correlating visual data with physical and environmental measurements, factories gain a far more reliable and nuanced view of operations.

For example, a visual anomaly combined with abnormal vibration data may indicate tool degradation rather than a material flaw. This holistic view reduces false alarms and accelerates corrective action.

Edge AI: Intelligence at the point of action

Edge AI ties machine vision and intelligent sensing together, enabling factories to interpret complex data locally, without relying on constant cloud connectivity.

Why the edge matters

Manufacturing environments demand capabilities that centralized systems struggle to provide:

• Low-latency decision-making for time-critical control
• Operational autonomy in environments with limited connectivity
• Data sovereignty and IP protection
• Scalable deployment across many machines and lines

Edge AI meets these needs by bringing inference and decision logic directly to machines.

Figure 4 Edge AI, the third key building block in smart factory designs, ties machine vision and intelligent sensing. Source: Renesas

Practical impact on operations

With edge AI, factories become more intelligent and proactive in their operations. Instead of reacting to problems after they occur, systems can predict potential failures in advance and help avoid costly disruptions. Processes can also be adjusted in real time to account for changes in materials or environmental conditions, ensuring consistent quality and efficiency.

In addition, AI-driven systems can identify unusual patterns and anomalies that were not explicitly programmed, enabling earlier detection of issues. At the same time, more intuitive and responsive human–machine interactions improve safety and usability on the shop floor. Altogether, this represents a clear shift from reactive control toward adaptive, self-optimizing operations.

Convergence: Creating intelligence through integration

The greatest gains emerge when machine vision, intelligent sensing, and edge AI are designed as a unified system rather than isolated capabilities.

Consider a high-mix production line:

• Machine vision identifies subtle quality deviations
• Intelligent sensors monitor mechanical and electrical behavior
• Edge AI correlates these inputs to identify emerging issues

Instead of scrapping products or stopping the line, the system can adjust in real time, maintaining quality while maximizing throughput. This distributed intelligence also simplifies factory architectures. Decisions are made close to the process, improving responsiveness and system robustness.

Designing for sustainable smart factories

Achieving this level of intelligence is not just a technical challenge, it is a system and ecosystem challenge. Manufacturers need platforms that simplify integration across sensing, processing, connectivity, and security, while supporting long product lifecycles typical of industrial environments.

As adoption accelerates, successful smart factory strategies share several traits:

• Scalability, allowing intelligence to be added incrementally
• Interoperability, avoiding vendor lock-in
• Lifecycle support, including long-term availability and maintenance
• Energy-efficient design, balancing performance with sustainability

Smart factories built on these principles are better equipped to adapt, not just to current challenges, but to future uncertainty.

In the final analysis, smart factory is not defined by a single technology, but by how technologies work together. Machine vision gives machines eyes. Intelligent sensing provides awareness. Edge AI delivers understanding.

With the right enablement and ecosystem support, manufacturers can move beyond reactive automation toward systems that continuously learn, adapt, and improve. In doing so, they transform data into decisions, and factories into resilient, future-ready operations.

Suad Jusuf is director of product marketing at Renesas Electronics. His work centers on defining distinctive value, empowering differentiation, and accelerating customer success through integrated MCU/MPU platforms, AI tools, and system‑level enablement and offerings.

Special Section: Smart Factory

• Rethinking machine vision in industrial automation
• Smart factory: The rise of PoE in industrial environments
• Precision lasers boost safety and efficiency in smart factories
• Tale of 3 sensors operating in smart factory environments
• From edge AI to physical AI in smart factories: A shift in how machines perceive and act
• Robots: Why AI alone will not deliver the next leap in automation
• How emerging robotics standards will shape next-gen automation

The post How machine vision, intelligent sensing, and edge AI are powering smart factory appeared first on EDN.

Three digital multiphase controllers from AOS enable Intel IMVP9.3 Vcore power delivery in high-performance mobile systems. When paired with the company’s DrMOS and Smart Power Stage devices, the AOZ71049QI, AOZ71149QI, and AOZ71146QI form a complete power solution for Intel Panther Lake and Wildcat Lake mobile processor architectures.

The buck controllers use AOS’s advanced transient modulation (A2TM), a hybrid approach that combines digital tuning with analog efficiency. By integrating variable-frequency hysteretic peak current-mode control with advanced phase current sensing, they deliver fast transient response and balanced current sharing across both transient and DC loads. They also maintain low quiescent power across all Intel IMVP9.3 power states, helping maximize battery life in laptops and notebooks. Key features are summarized below:

• Flexible configurations: Up to 4+2+1+2 phase outputs for Core (IA), Graphics (GT), Auxiliary (SA), and LPCORE domains
• Low quiescent current: 5.9 mA at PS0 in 3+2+1+1 configurations
• Power management: Autonomous phase shedding and auto-DCM to reduce power loss
• Compatibility: Supports industry-standard DrMOS and driver + MOSFET power stages from multiple vendors
• Acoustic noise suppression: Integrated features reduce audible noise under dynamic load conditions

The AOZ71049QI, AOZ71149QI and AOZ71146QI are available in production volume, with a lead time of 12 to 16 weeks. Prices start at $2.66 in 1000-piece quantities.

Alpha & Omega Semiconductor 

The post Multiphase controllers optimize mobile Vcore power appeared first on EDN.

Lumissil’s IS32FL3776 matrix LED driver brings expressive intelligent signal displays (ISDs) to software-defined exterior automotive lighting. With 36 constant-current channels providing 60 mA each, it drives dynamic LED light matrices up to 36×6 with as many as 216 individually addressable LEDs.

Automotive ISD systems use matrix LED patterns to communicate vehicle intent, safety status, driver-assistance cues, and brand identity. The IS32FL3776 enables compact LED designs used in RGB mini LED displays, full-width front light strips, grille lamps, automated driving system marker lamps, and other expressive vehicle lighting functions.

The driver features high-resolution, high-frequency dithered PWM for fine brightness adjustment and smooth animations without flicker or camera banding. For improved system efficiency and thermal performance, the IS32FL3776 uses DCFB adaptive control to optimize the LED supply rail while maintaining sufficient headroom for proper current regulation. A software-configurable architecture supports either internal operation or external PMOS drive for power sequencing in larger matrix configurations.

The IS32FL3776 is available for sampling and volume production, with evaluation hardware and reference designs provided to facilitate system development.

IS32FL3776 product page 

Lumissil Microsystems 

The post LED driver animates exterior vehicle lighting appeared first on EDN.

Microchip’s PIC18-Q20 MCUs integrate up to two I3C peripherals and Multi-Voltage I/O (MVIO) in 14- and 20-pin packages as small as 3×3 mm. Well suited for sensor interfacing, real-time control, and connectivity applications, they simplify communication across multiple voltage domains with minimal external circuitry.

Compared to I2C, I3C provides higher data rates and lower power consumption while remaining backward compatible with legacy systems. The MCUs operate across three independent voltage domains, with MVIO-enabled pins supporting I3C communication down to 1.0 V. Additional integration includes a 10-bit ADC with computation, capacitive touch sensing, and an 8-bit signal routing port for flexible peripheral interconnect.

The PIC18-Q20 series can process sensor data, manage low-latency interrupts, and perform system status reporting, reducing the workload on a host MCU in larger systems. These devices are supported by Microchip’s hardware and software development ecosystem, including the PIC18F16Q20 Curiosity Nano Evaluation Kit for rapid prototyping.

Now in production, the PIC18-Q20 MCUs are available from Microchip and its authorized distributors.

PIC18-Q20 product page 

Microchip Technology 

The post MCUs bridge I3C across voltage domains appeared first on EDN.

Toshiba is sampling the TB9M040FTG motor control device, which integrates an MCU and motor driver for controlling small automotive motors. Part of the SmartMCD series, it supports single-channel motor drive currents up to 2 A, enabling direct drive of three-phase brushless DC motors used in electric valves, HVAC dampers, flaps, and grille shutters.

In addition to an Arm Cortex-M23 processor core running at up to 40 MHz and the motor driver, the TB9M040FTG incorporates flash memory, a 5-V high-side driver for power-supply functions, and a power supply that operates at automotive battery voltage levels. It also integrates a LIN transceiver for ECU communication.

The device features a hardware vector engine that offloads field-oriented control (FOC) processing, helping reduce CPU load and software size. Back-EMF detection enables sensorless square-wave control.

All functions are integrated into a compact VQFN36 package, reducing component count in automotive equipment. The TB9M040FTG is compliant with AEC-Q100 Grade 0 and ASIL-B requirements.

TB9M040FTG product page

Toshiba Electronic Devices & Storage 

The post Motor MCU integrates driver and control functions appeared first on EDN.

The SiFive Performance P570 Gen 3 is a RISC-V out-of-order superscalar vector processor IP designed for scalable performance. SiFive says it delivers a substantial performance improvement over the P550 Gen 1, along with a comprehensive set of mandatory and optional RVA23 profiles.

The IP can serve as the control processor in embedded IoT devices with full networking stacks or as the main applications processor in consumer devices running operating systems such as Android and enterprise-grade Linux. Its vector unit also supports AI model execution and inference on edge devices.

Multicore configurations scale to 16 cores across four clusters with shared L3 and optional L2 cache, a RISC-V-compliant interrupt architecture, and fine-grain power-management control. The P570 supports mandatory RVA23 requirements, including Hypervisor and Vector extensions, along with optional security and management extensions, RISC-V Vector Crypto, and FP16/BF16 capabilities for AI acceleration.

The Performance P570 Gen 3 IP is available now. Visit the product page for configuration and customization details.

P570 Gen 3 product page 

SiFive

The post CPU IP processes mixed scalar and vector workloads appeared first on EDN.

The healthcare industry is progressively moving from a centralized, clinical model to a more patient-centric approach, requiring monitoring solutions that are portable, wearable, and patient-focused. This process involves significant technical and hardware challenges. Designers must find a way to maximize diagnostic accuracy and reliability in a clinical setting while also keeping power consumption, size, and long-term reliability in mind.

The design of a medical device usually follows a modular approach. This means that each part, from the first signal capture to the last communication protocol, must be optimized for speed and efficiency. This article will provide insights into some of the most relevant sensors employed in medical devices as well as the associated technologies, including analog front ends (AFEs), power management devices, and wireless system-on-chips (SoCs) for connectivity.

Pressure sensors

With the wide range of medical devices, from wearable glucose monitors to computerized tomography (CT) scan equipment, there is also a variety of sensors incorporated into these devices. These include pressure and temperature sensors as well as biosensors and accelerometers.

Pressure sensors are used in a wide range of medical equipment, from non-invasive blood pressure monitors to specialized airflow sensors in ventilators. Besides common medical requirements, such as reliability and high sensitivity, these sensors must provide robustness and endurance. Medical-grade pressure sensors must exhibit high linearity and long-term stability, maintaining calibration over weeks or months of continuous operation.

For example, TDK Corporation offers a wide portfolio of piezoresistive pressure sensor dies well-suited for high-precision measurements in the medical sector. Based on advanced silicon MEMS technology, these sensors are grouped into three main categories: absolute, gauge, and differential pressure.

As for the piezoresistive pressure measurement methods, sensor dies are available with frontside and backside absolute measurement and gauge-differential measurement. The frontside configuration, where the electronics are directly exposed, is preferred for dry, non-aggressive gases. The backside design allows the sensor to handle wet media or non-aggressive fluids because the sensitive electronic components are shielded on the opposite side of the pressure-sensing diaphragm. Finally, the gauge configuration is well-suited for physiological measurements relative to ambient.

The C39 series are highly miniaturized dies (with an area of 0.65 × 0.65 mm) with frontside absolute pressure measurement up to 1.2 bar. Able to operate over a temperature range from −40°C to 150°C, these sensors are optimized for high burst pressure and feature narrow sensitivity tolerances and high signal stability. As such, they are suited for integration into high-density wearable medical devices.

Sensors for medical imaging

Imaging technology has made significant advances in the last few years. CT scans used for the diagnosis and monitoring of various conditions, including cancer and cardiovascular diseases, have evolved with the introduction of the photon-counting CT (PCCT).

The main difference between these two techniques lies in how sensors (“detectors”) process X-rays. Conventional CT uses indirect energy-integrating detectors. X-rays hit a scintillator, convert it to light, and then to electricity. In practice, they measure the total energy accumulated, losing individual photon data.

PCCT instead uses direct-conversion sensors that convert X-rays directly into electrical pulses, counting every single photon and measuring its specific energy. This eliminates electronic noise, improves spatial resolution, and allows for precise tissue differentiation at a lower radiation dose.

Ams Osram, now part of Infineon Technologies AG, introduced a system-in-package (SiP) sensor module specifically designed for photon-counting detectors. This sensor, shown in Figure 1, enables a significant reduction in the radiation dose and diagnostic images with higher resolution.

As the company states, the AS5920M module features a 9× reduction of the module’s detector pixel size compared with traditional CT systems. Moreover, more modules can be combined in an array arrangement, increasing the detection area according to the desired CT application.

Figure 1: The AS5920M is a four-sided buttable SiP sensor module engineered for photon-counting detectors (Source: ams Osram)

At the 2025 annual meeting of the American Society for Radiation Oncology, Siemens Healthineers presented the Naeotom Alpha.Prime PCCT scanner (Figure 2) based on cadmium telluride crystal detectors that significantly improve image resolution and contrast. The company introduced the world’s first PCCT scanner in 2021.

Figure 2: Siemens Healthineers’ Naeotom Alpha.Prime PCCT scanner (Source: Siemens Healthineers AG) Embedding AI in sensors

The integration of embedded AI cores directly into biosensors is changing the architecture of medical diagnostics. Previously, devices were limited to a traditional sensing process, wherein all raw data was transmitted to a central processor for analysis. With the direct integration of edge intelligence, sensors can now process data locally, exactly where it is sourced.

The main benefit of this architecture is efficiency, as the device transmits only processed results or alerts. This approach significantly reduces the system’s power consumption, latency, and required bandwidth.

STMicroelectronics has introduced a high-accuracy biosensor that integrates a vertical AFE (vAFE) for biopotential signals (typically cardio and neurological parameters) with a low-power, three-axis accelerometer with AI and anti-aliasing. The ST1VAFE3BX’s vAFE features programmable gain and input impedance and includes a 12-bit ADC.

Providing output data at a rate up to 3,200 Hz, the biosensor is well-suited for biopotential measurement of heart, brain, and muscular activities. The compact size (2 × 2 mm) and reduced power consumption (48.1 µA during normal operation, which can be cut to just 2.6 µA in power-saving mode) suit it for wearables designed for predictive healthcare.

The biosensor features ST’s proprietary machine-learning core (MLC) and finite-state machine (FSM), which allow designers to develop decision-making rules and algorithms to be deployed directly on the chip. The AI-assisted capabilities enable the sensor to autonomously manage motion and activity detection.

This AI feature decreases the interactions with the host controller, reducing the overall power consumption and latency while extending battery life. MLC and FSM can be implemented using ST’s software development tools such as MEMS Studio, which is part of the ST Edge AI Suite.

High-precision AFE

The integrity of a medical device is defined by the quality of its input data. AFEs are components required for interfacing with the human body. They are essential for all types of medical sensors that produce analog signals and therefore require further processing, such as conditioning, amplification, filtering, and digital conversion.

AFEs bridge the gap between physical measurements, typically available in analog form, and the compute device that processes them in digital form. In medical devices, AFEs are required for any sensor that measures physical parameters.

AFEs operate by extracting small-amplitude physiological signals from the environment, which are often noisy or subject to electromagnetic interference. As a result, to achieve medical-grade results, the AFE must provide a high signal-to-noise ratio and low leakage currents.

Among the sensors that require an AFE are biosensors, such as those used in continuous glucose monitoring (CGM) and electrocardiogram patches. Onsemi’s CEM102 is an AFE specifically designed for CGM and similar applications. Based on an amperometric measurement that senses very low currents, the device features a small form factor and low power consumption. These features suit the CEM102 for miniaturized and battery-operated medical devices.

The CEM102 can be operated with a supply voltage ranging from 1.3 to 3.6 V—typically a single 1.5-V silver oxide battery or a standard 3-V coin cell. It supports up to four electrodes, integrates a high-resolution ADC and several DACs for bias setting and a factory-trimmed system, and can be interfaced with a host controller, such as the onsemi RSL15, a secure Bluetooth 5.2 wireless microcontroller (MCU) for connecting to an external device or terminal.

Power management

In the design of compact wearables, such as hearing aids, power management represents one of the most challenging constraints. Designers must select power management integrated circuits (PMICs) with high efficiency, thus preserving the energy provided by small battery cells.

Onsemi’s HPM10 battery-charge controller is a high-performance PMIC engineered to recharge batteries in miniaturized medical devices, typically hearing aids and cochlear implant devices. The device supports different rechargeable battery technologies, including lithium-ion and silver-zinc, and can detect zinc-air and nickel-metal hydride disposable batteries.

The HPM10 also provides a charger communication interface to communicate the state of the charging process to the hearing-aid charger. Other information available on this interface includes the battery voltage levels, current levels, temperature, and battery failures.

Connectivity

A medical device is more effective if it can communicate data to clinicians or electronic health-record systems. Several connectivity protocols are available, and their selection is based on the application’s range and data throughput requirements.

Low-power Bluetooth SoCs are the industry standard for wearables, providing a reliable and efficient link to smartphones or home gateways. For high-bandwidth clinical environments, such as hospitals or clinics, integrating Wi-Fi 6 with Bluetooth Low Energy (LE) represents a suitable connectivity solution.

For example, Silicon Labs’ Series 2 BG29 family of wireless SoCs is designed to provide Bluetooth LE connectivity in an extremely small form factor. The BG29 device’s small size (2.6 × 2.8 mm) suits it for applications such as wearable health and medical devices and battery-operated sensors. The device integrates a DC/DC boost converter supporting a wide voltage range, a Coulomb counter for accurate battery monitoring, 1 MB of flash, 256 kB of RAM, and security features.

Figure 3: Silicon Labs’ BG29 is available in compact QFN and WLCSP packages. (Source: Silicon Laboratories)

NXP Semiconductors is collaborating with Silex Technology, a provider of wireless connectivity and smart edge solutions for the medical and industrial sectors. Silex focuses on wireless solutions for medical applications requiring high longevity, cybersecurity features, and high reliability. Patient monitors, medical wearables, and other connected devices often operate in hospitals where several Wi-Fi access points are available.

Silex integrates NXP’s Wi-Fi SoCs in its Wi-Fi 6 + Bluetooth 5.3 and 5.4 module solutions, including NXP’s IW611 Wi-Fi 6 SoC and RW610 Wi-Fi 6 wireless MCU.

The post Designers guide: Sensors for medical devices appeared first on EDN.

It’s always gratifying when a simple and successful design idea luckily turns out to have additional applications that you didn’t originally envision.  Here’s an example.

Wow the engineering world with your unique design: Design Ideas Submission Guide

A while back, the design shown in Figure 1 was accepted for publication:

Figure 1 U1 plus R1 through R5 current steering networks convert a 0/20mA input into a 4/20mA output.

Later, the same circuit, when wired up differently as shown in Figure 2, turned out to be an equally good fit in a different job:

Figure 2 This 4/20mA current loop converter integrates an OFF/ON field contact.

A recent Design Idea by another frequent contributor, Jayapal Ramalingam, addressed the problem of convenient calibration of precision current loop receivers in industrial applications. His design comprises a linear control input that expedites calibration and testing.   He explains that it helps to:

…”calibrate the analog input modules of distributed control systems (DCSs) and programmable logic controllers (PLCs) by simulating process signals.”…

This inspired me to wonder if a different approach to the same calibration problem might also be useful.  I imagined a design in which the three standard analog test current loop levels: 0, 4mA, and 20mA, were accurately preset and quickly accessed by flipping a switch. I then proceeded to ponder whether that same friendly little converter circuit might work in such an application.

Figure 3 shows the result:

Figure 3 The three-position, center-off, DPDT switch S1 converts this current converter (verbiage redundancy pun-intentional) into a convenient current calibrator.

Not only did it fit, but the calibration procedure for the new role is just as quick, simple, and easy to accomplish in a single pass as it was before.

1. Set S1 to the 4mA position.
2. Tweak 4mA adj for 4mA output (as measured, for example, with a precision DMM).
3. Set S1 to the 20mA position.
4. Adjust 20mA adj for 20mA output (ditto).

So, it turns out that the same circuit thriftily fits three related, yet different, applications – a triple-duty design trifecta.

Stephen Woodward‘s relationship with EDN’s DI column goes back quite a long way. Over 200 submissions have been accepted since his first contribution back in 1974.  They have included best Design Idea of the year in 1974 and 2001.

Related Content 

• Silly simple precision 0/20mA to 4/20mA converter
• Double-duty current loop transmitter
• Linearly variable two-wire loop current generator
• Full circle current loops: 4mA-20mA to 0mA-20mA
• 4/20mA to 0/20mA loop current converter for grounded loads

The post Triple-duty current loop calibrator appeared first on EDN.