EDN Network
A short primer on optoelectronic transistors, their apps
Optoelectronics, the use of electronic devices for the detection and control of light, has become a preferred means for broadband and transoceanic communication networks. Below is a sneak peek at the basics of optoelectronic transistors, their different types, applications, and the future trends they entail.
What is an optoelectronic transistor
Optoelectronic transistors are semiconductor devices that combine the best of both worlds: electronic transistors that can control and generate light. They take advantage of the properties of semiconducting materials to regulate the flow of both electrons and photons, paving the way for a range of exciting possibilities in future technologies.
An optoelectronic device is essentially a light-related electronic gadget. The transistor is a component of optoelectronics which makes the functioning of the optoelectronic devices efficient.
Optoelectronic transistor working principle
At the heart of a transistor lies a heterojunction. This brings together two semiconducting materials together with complementary electronic and optical properties. This junction forms a device that can efficiently convert one form of signal into another and vice versa. The key principle behind their functioning is the ability to control the density of charge carriers in the semiconductor, thus regulating the flow of both electrons and photons.
Various types of optoelectronic transistor devices and applications
- Photodiode
When light strikes the junction of a semiconductor photodiode, it produces a voltage or current. It’s made up of an active P-N junction that is biased in the opposite direction. An electron or hole pair is produced in a semiconductor when a photon with sufficient energy strikes the material. An electric field is created at the junction by the diffusion of electrons.
Applications
Photodiodes are usually used in applications such as cameras, medical equipment, and safety equipment like fire detectors. They are also used in automotive as headlight dimmers, twilight dimmers, sunlight detectors, and climate detectors.
Additionally, aircraft X-rays have gained significant importance in the past years. For instance, in 2022, the Transportation Security Administration seized more than 6,542 weapons from airport patrons, which is a record-high figure since the agency’s creation. At airport security checks, 88% of the firearms seized were loaded. Hence, aircraft X-ray is also an important application that uses photodiode.
- Solar cell
An electronic device referred to as a solar cell or photovoltaic cell directly transforms solar energy into electricity. A solar cell creates a voltage and a current to produce electricity when sunlight strikes it. Photons, which make up sunlight, are emitted from the sun. Photons transfer energy to loose electrons when they strike silicon atoms in a solar cell. These high-energy electrons subsequently flow to an external circuit.
Two layers that are struck collectively make up the solar cell. The electrons are prepared to jump from the first layer to the second layer since the first layer is laden with them. Due to the removal of some electrons, the second layer is prepared to accept additional electrons. Solar cells have the benefit of not having a fuel supply or price issue. These require very little upkeep and are quite dependable.
Applications
Solar cells can be utilized for a variety of solar energy-based projects, including rural electrification, telecommunication systems, ocean navigation aids, electric power generation in space, remote monitoring and control systems, and more.
- Light-emitting diode (LED)
A photon is produced when electrons and holes combine to form a light-emitting diode, a P-N semiconductor diode. The diode emits narrow-spectrum light that is incoherent when it’s electronically biased in the forward direction. The electrons and holes within the LED recombine when a voltage is supplied to its leads, releasing energy in the form of photons.
Electroluminescence is the name given to this phenomenon. It’s the process through which electrical energy is changed into light. The energy band gap of the substance determines the color of the light.
Applications
LEDs have covered a wide range of applications such as phones, digital watches, and automobiles. Its use is also growing in the aviation industry. Additionally, traffic signals, optical communication, and indicator lamps are some of the important applications of LEDs. So, the market for LEDs is set to grow at CARG of 12% by the end of 2035.
- Optical fiber
An optical fiber uses total internal reflection to transmit light along its axis. It’s a dielectric waveguide that is cylindrical in shape. A core is enclosed by a cladding layer, both of which are constructed of dielectric materials.
Applications
Due to its flexibility and thinness, optical fiber is used in a variety of tools to view internal body parts by passing through body cavities. Optical fiber is used in surgery, endoscopic, microscope and biomedical lasers. Moreover, high-speed, high-bandwidth HDTV transmissions are delivered through such cables. Fiber optic cable is more affordable than copper wires.
- Laser diode
High-quality, coherent, directional light comes from lasers, which are sources of light amplified by radiation that is triggered to emit light. It functions in a stimulated emission environment. A laser diode’s purpose is to transform electrical energy into light energy, similar to infrared diodes and LEDs.
Applications
The most widely produced form of laser, laser diodes have a wide range of uses, including fiber optic communications, barcode readers, laser pointers, reading/recording of CD, DVD, and Blu-ray discs, laser printing, laser scanning, and beam lighting.
Furthermore, the preference for green optoelectronics is growing. There is also a growing influence in manufacturing paper-based optoelectronic devices. The paper has played an important role in human civilization as the continuous record bearer of contemporary civilization. As paper production technology has advanced, paper production has increased continuously. Every year, about 299 million tons of paper are produced worldwide.
The primary raw material in paper is cellulose. A diluted solution of cellulose fibers is dehydrated, filtered, compressed, and heated to create traditional paper. Numerous varieties of nano paper have been used in research in recent years. One of them is in optoelectronics. A discrete semiconductor system can sense, correct, amplify, toggle, maintain voltage stability, and manipulate data, among other functions. The organic optoelectronic device is supported well by the movable foundation, which is also essential to the device’s customizable interface.
That has opened opportunities for the manufacturing of organic devices. For instance, the use of organic light-emitting diodes (OLEDs) for projection and illumination has risen. OLED provides several advantages over conventional LEDs, including a wide field of vision, quick response times, and light weight.
The two main methods utilized in paper-based OLED research are building a buffer layer on a platform made of regular paper and using cellulose to create a different kind of platform. The OLED market is estimated to grow at a CAGR of 22% by the end of 2035.
Future trends and prospects of optoelectronic transistors
Optoelectronic transistors have potential across several fields, thanks to their seamless integration of electronic and optical functionalities. Below are some of the fields where these transistors are making notable advancements.
- Optical computing
By leveraging transistors, we can create speed and energy for efficient optical computing systems. For manipulating electrons, these transistors harness the power of light to process amounts of data in parallel, providing notable advantages in terms of speed and energy efficiency when compared to traditional electronic circuits.
- Optical communications
Optoelectronic transistors are becoming increasingly important in communication networks as the need for high-speed data transmission grows exponentially. They have the potential to enable the development of optical switches, routers and modulators, facilitating faster and more dependable data transfer.
- Photovoltaics
The use of optoelectronic transistors in the field of solar energy is increasing. The integration of light absorption, charge generation, and transmission through a single device is paving the way for efficient solar cells. These solar cells are expected to revolutionize the renewable energy sector with easy, stable, and higher conversion efficiency.
- Imaging technologies
Optoelectronic transistors have applications in the sensing and imaging technology sector as well. In biomedical imaging devices, these transistors boost weak optical signals by amplifying them. This helps in elevating the performance of the imaging devices.
Ranjana Jain is a content writer at Research Nester.
Related Content
- Terahertz Optical Transistors Beat Silicon
- Universal purpose optoelectronic logic elements
- Unlocking Compound Semis’ Economy-Driving Potential
- Optoelectronics set to benefit from BiCMOS manufacturing developments
- Universal purpose optoelectronic logic element with input optical switching of AND/NAND, OR/NOR and XOR/XNOR functions
The post A short primer on optoelectronic transistors, their apps appeared first on EDN.
Cadence develops next-gen Xtensa LX processor
Representing the 8th generation of the Tensilica Xtensa LX processor family from Cadence, the LX8 improves system-level performance by as much as 50%. The 32-bit RISC processor IP serves as a foundation for processor and system-level designs, including DSP, multiprocessor, interconnect, and system-level IP products.
Among the LX8’s multiple enhancements is a flexible L2 memory option, which can be configured at initialization to be a fixed-address memory, L2 cache, or a combination of both. Cadence reports performance improvements of 50% or more for cache-based subsystems compared to the Xtensa LX7 processor.
Other advancements include improved branch prediction and enhanced Arm AMBA interfaces, such as AMBA 4 AXI and APB manager interface. An integrated DMA (iDMA) controller improves 3D DMA transfers found in complex DSP algorithms. The iDMA also adds compression/decompression support and expands the physically addressable memory to 40 bits. Further, the Xtensa LX8 accommodates up to 128 interrupts.
The Tensilica Xtensa LX8 processor is currently shipping to early access customers, with general availability expected in the late third quarter of 2023.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post Cadence develops next-gen Xtensa LX processor appeared first on EDN.
LTE Cat 4 module runs under Linux OS
The SC696S multimode LTE Cat 4 module from Quectel empowers a wide array of applications with its built-in Linux OS and both Wi-Fi and Bluetooth capabilities. It also supports 1×1 MIMO technology for IoT connectivity and includes a multi-constellation GNSS receiver for accurate fixes in any environment.
Processing power for the SC696S smart module is derived from a Qualcomm QCM6125 SoC with a 64-bit octa-core Kryo 260 CPU and Adreno 610 graphics processing unit. The device offers worldwide LTE, UMTS/HSPA+, and GSM/GPRS/EDGE coverage. Onboard interfaces include LCM, camera, touch panel, UART, USB, I2C, I2S, and SPI interfaces. It also accommodates a maximum of six cameras, two of which can work concurrently.
The SC696S operates over a temperature range of -35°C to +75°C, enabling deployment in challenging environments. Applications for the SC696S range from smart home devices and wearables to industrial equipment, robots, and point-of-sale devices.
Housed in a 43×44×2.85 LCC+LGA package, the SC696S LTE Cat 4 module is offered in three variants: SC696S-EM, SC696S-NA, and SC696S-WF (Wi-Fi/Bluetooth only). It is sold through Quectel’s distributor network.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post LTE Cat 4 module runs under Linux OS appeared first on EDN.
Near-field flash LiDAR caters to diverse applications
Mojave from PreAct is a flash LiDAR sensor that meets the near-field 3D sensing requirements of both automotive and non-automotive applications. Based on continuous-wave time-of-flight technology, the modular LiDAR sensor offers software-defined capabilities and a depth accuracy error of less than 2%.
Able to define and track objects in darkness or intense direct sunlight, Mojave can be used in numerous applications. Specific use cases include elevator passenger monitoring, retail, patient monitoring in medical facilities, security cameras, robotics, smart cities, and education and university research.
The Mojave sensor’s field of view is 70° horizontal, 52.5° vertical, and 87.5° deep. It has a resolution of 320×240 pixels and a maximum frame rate of 18 fps. The sensor’s range at 10% reflectivity is 0.3 m to 4.5 m. At 94% reflectivity, the range extends from 0.3 m to 14 m.
“We created the sensor to allow companies to monitor volume and movement through high-density point clouds, which gives them the information they need to adjust their services without the ‘creepy’ factor of watching individuals on camera,” said Paul Drysch, CEO of PreAct. “In addition, you get much more useful data with a point cloud, such as precise object location and volume.”
With a retail price tag of $350, the Mojave flash LiDAR sensor will be available in September 2023 and sold through Digi-Key Electronics and Amazon. Engineering samples will be available August 16, 2023. Products can be preordered by contacting PreAct.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post Near-field flash LiDAR caters to diverse applications appeared first on EDN.
JEDEC crafts base standard for CXL memory modules
JEDEC has published the JESD317: Compute Express Link (CXL) Memory Module Base Standard to simplify system design and ease device specification. JESD317 defines the specifications for interface parameters, signaling protocols, environmental requirements, packaging, and other features as reference for specific target implementations of CXL-attached memory modules.
The JESD317 base standard leverages established industry-standard specifications, including the Storage Networking Industry Association’s EDSFF, as well as PCIe and CXL. According to JEDEC, it enables system providers to accommodate certain known devices, while leaving room for innovation. The base standard also supports system device slots designed to be compatible with both CXL-attached memory modules and conventional PCIe-attached NVMe drives.
The JESD317 document is available for free download here. Registration is required.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post JEDEC crafts base standard for CXL memory modules appeared first on EDN.
RGB device transforms dynamic in-vehicle lighting
An RGB LED module, the ams Osram Osire E3731i enables the creation of dynamic color and motion effects across hundreds of LEDs for automotive interior lighting. Using the company’s license-free Open System Protocol (OSP), the device’s digital core can communicate with any microcontroller over a standard serial bus interface.
The module’s trio of LEDs are co-packaged with an embedded IC that provides drivers for each LED, a serial bus interface, temperature sensor, and on-chip memory. OSP allows an external microcontroller to transmit instructions to modulate color and brightness individually to each LED. Up to 1000 LEDs can be connected in a daisy chain and controlled by a single microcontroller.
The Osire E3731i is characterized at the factory, and its optical performance data is stored in on-chip memory. This makes it easier for automotive manufacturers to perform end-of-line calibration of interior lighting systems, while ensuring optical uniformity and consistency across LED arrays.
The OSIRE E3731i comes in a surface-mount package and is AEC-Q102-003 qualified. Register for the Open System Protocol 1.0 by using the link to the product page below.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post RGB device transforms dynamic in-vehicle lighting appeared first on EDN.
VS Code gaining traction in programming MCUs, MPUs
Microsoft Visual Studio Code, commonly known as VS Code, has another embedded processor company in its fold: top MCU supplier Renesas, which ships more than 3.5 billion units annually. The Japanese chipmaker will embrace VS Code to program its microcontrollers (MCUs) and microprocessors (MPUs).
Figure 1 Embedded developers can now design and debug software for Renesas embedded processors in VS Code. Source: Renesas
In March 2023, STMicroelectronics made a similar announcement by adopting VS Code for its STM32 microcontrollers. Like Renesas, ST will make available tool extensions that bring the advantages of VS Code to its MCUs.
VS Code, a popular integrated development environment (IDE) and code editor, is gaining traction for its ease of use and flexible features that simplify and accelerate code editing. And it simplifies and accelerates code editing across a variety of platforms and operating systems. That includes Windows as well as Linux x64 and macOS on Apple devices.
Unlike Arduino IDE—widely seen as training wheels for simple projects and a straightforward tool for non-developers to get started in creating the Internet of Things (IoT) and embedded applications—VS Code is a highly extensible code editor for embedded software development. It’s open source, and users can download it free of charge, including access to the source code.
Figure 2 VS Code enables developers working on high-level and consumer applications to easily create embedded solutions. Source: Microsoft
Both Renesas and ST are adding tool extensions for their embedded processors to the Microsoft VS Code website. The tool extensions for Renesas MCUs and MPUs are available on the Microsoft VS Code website and at www.renesas.com/software-tool/renesas-extension-of-vscode.
Likewise, both embedded processor suppliers are offering VS Code support alongside their respective developer platforms. VS Code support will complement the e2studio IDE platform of Renesas as well as ST’s Eclipse-based STM32CubeIDE environment. So, VS Code will allow developers to edit, build, program, run, and debug e2studio IDE and STM32CubeIDE projects, respectively.
VS Code is becoming a preferred environment for high-level software developers, academics, enthusiasts, and makers alike for creating embedded applications. Its adoption by leading embedded processor suppliers demonstrates its merits for creating efficient embedded solutions.
Related Content
- How To Buy Embedded Processors
- Embedded Processors of Tomorrow
- Embedded Processors by the Numbers
- The basics of programming embedded processors
- Real-Time Edge Processing, AI Play Major Embedded Roles
The post VS Code gaining traction in programming MCUs, MPUs appeared first on EDN.
Adventures with a remote heart monitor
On the heels of my fourth stroke this past May, I was subjected to a lot of testing. One of the tests required me to wear a portable electrocardiogram (ECG) heart monitor for one month.
Some years ago, around the time of my third heart attack, I believe, I wore such a monitor for one month and it was an endurance test to do so. I had this heavy, cucumber shaped thing glued to my chest wall, then another and then another and when it was all over, my relief was beyond my ability to phrase.
Each cucumber shaped sensor I’d been wearing back then had recorded and stored my heart wave shapes for its own time span of usage and then the whole set of sensors with their collected data were sent back for subsequent analysis. This time though, things were different, see Figure 1.
Figure 1 The new, improved heart monitor with a much smaller sensor and a Bluetooth link to a nearby cellphone equipped with the appropriate app. Source: John Dunn
I still had something glued to my chest wall, a “sensor”, but it was very much smaller than the ones before. This sensor had a Bluetooth link to a nearby cell phone equipped with an app that provided constant ECG monitoring to a remote facility for immediate analysis. Also, if I experienced any of a list of symptoms, I could report those symptoms in real time for correlation to ECG wave shapes.
Cool stuff, I thought.
Things weren’t perfect though. As simple as the game plan was, actually figuring out how it all worked was a challenge. The instruction book seemed like little more than a mnemonic memory aid for its author and the person from whom I sought clarification at customer service was ill-trained and linguistically inept.
Still, after I got things figured out, the process worked. I later got back the ECG results themselves in which things like this showed up (Figure 2).
Figure 2 My ECG results from the remote cardiac monitor with abnormal QRS complexes but no atrial fibrillation, a pattern that would cause alarm. Source: John Dunn
Scary looking, isn’t it?
This pattern, others like it and others very much unalike were to be found here and there, but my doctor saw no reason for alarm because throughout that month, there was never any sign of atrial fibrillation. Seeing any of that going on would have raised some concerns.
Hooray!!! I’ve gotten back to normal life.
John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).
Related Content
- Avoiding blood pressure measurement errors
- Avoiding blood pressure measurement errors – Part 2
- How to design an optical heart rate sensor into a wearable device’s wristband
- Analog Devices introduces a single-lead heart-rate monitor AFE
- Optical heart-rate measurement’s top 5 challenges
- Medical devices: Where wireless charging has real impact
The post Adventures with a remote heart monitor appeared first on EDN.
Extending the resolution of a peripheral DAC
The 8-bit resolution of a peripheral DAC (such as the ATtiny family, for instance) is often insufficient. Let’s see how the problem can be solved using on-chip resources.
The circuit in Figure 1 shows a method to improve the resolution of a peripheral DAC. The circuit also reduces output resistance of the DAC and can reduce its offset (which is rather pronounced for ATtinyx17).
Figure 1: Circuit used to improve the resolution of a peripheral DAC while also reducing the output resistance and offset of the DAC.
Wow the engineering world with your unique design: Design Ideas Submission Guide
This simple approach is fully static (needs no additional capacitors with their imperfections and charge refreshing), it preserves the monotonicity of the original DAC, but may require some tuning to achieve the best linearity.
To add one more bit you can use any free port of your MPU, in Figure 1, this port is marked as Pxn (x=A … C, n=0 … 7). (And yes, you can add more than one bit this way.)
Note: The port Pxn should be configured as an output before using it as an extra bit for the DAC. Also have in mind those very useful instructions in AVR ASM, which can selectively change one bit in a row.
The bit of the Pxn can be used as a new most significant bit (MSb), or a new least significant bit (LSb)—a detailed description of the last case is given below.
To maintain the integrity of the output, both the DAC0 and the Pxn should have the same headroom on their outputs. The easiest way to secure this is to choose +VDD as VREF for the DAC. Of course, the value of +VDD has to be well-regulated in this case.
The op amp should provide sufficient speed and precision where the total added error at the output of the op amp is:
Er = Vos * (1 + R5 / Req),
where Vos is the offset voltage of op amp and Req = R1||R3 (approximately).
The value of Er should be at least less than 1 LSb of the modified DAC.
The minimal voltage on DAC0 OUT is about 0.2 V (for ATtinyx17). If this offset is okay for your application, you can exclude resistors R3 and R4, and use only one +E supply, nevertheless, the op amp should have: a rail-to-rail output, a sufficient precision, and be able to work with near zero input.
To reduce the minimal value of e0 to zero, you must zero the output with the resistors R3 and R4 while DAC0 gets its zero code.
The values of all resistors can be calculated as follows:
R1 ≥ 5.6k (the requirement from the datasheet of ATtinyx17).
Let N bits be the original DAC resolution. Then:
R2 = R1 * 2^(N+1) (some final tuning may provide the better result).
Since the minimal voltage on DAC0 OUT is about 0.2 V,
R3 + (R4/2) = 5*R1 / 0.2 (If VDD = E = 5 V)
R3 = 0.8*(5*R1 / 0.2)
R4 = R3 / 2 (R4 is a multi-turn pot).
The resistor R5 can easily conform the DAC output to the values expected by an application.
Some efforts may be required in the program code to synchronize the DAC0 OUT and Pxn, albeit there are applications which can tolerate the de-synchronization if it is not very large (less than tens of microseconds).
Note: the output of the op amp is inverted compared to the DAC0 OUT (hence the sign “-” before e0 in Figure 1). You can cope with this by adding an inverter (op amp) to the output, or by making a change in the code, which may be a better solution.
You should not expect that the previous maximal value of the conversion rate would be left unchanged. Every next added bit means a resistance that is twice as large as well as some added capacitance in the summing node.
So, the settling time will be unavoidably larger with every expansion bit added.
To reduce a parasitic capacitance in the inverting node of the op amp the resistors R1, R2, R3, and R5 should be placed close to the inverting input.
Sometimes to optimize the performance of the DAC (to compensate the amplifier), the capacitor C3 (10-40 pF) has to be connected in parallel with R5.
—Peter Demchenko studied math at the University of Vilnius and has worked in software development.
Related Content
- A safe adjustable regulator
- Shunt circuit clips large transients or regulates voltage
- LED current regulator has low dropout
- Simple switchable filter
- Squeeze extra resolution from an 8-bit DAC with Shannon decoder idea
- Double µC’s PWM frequency & resolution
The post Extending the resolution of a peripheral DAC appeared first on EDN.
Dissecting a battery-backed LED light bulb
Teardowns of various LED light bulb flavors are consistently among the most popular pieces I write. This, I think, is a to-date complete list (not counting LED illumination sources that aren’t bulb-shaped, like touch-activated and motion-sensing panels):
I’ve also treated the topic of LED illumination more conceptually; take, for example, this piece from more than seven years ago that analyzed LED and CFL light bulbs as potential successors to incandescents. Or take this one from a couple of years later (and a half-decade ago this month) that discussed how, as LED light bulb technology matures, manufacturers are variously differentiating their products in search of competitive isolation and profits. Such innovations are fundamentally enabled by LEDs’ inherent low power consumption and heat dissipation, along with their inherent reliance on a DC voltage source. Several of these differentiated offerings (color temperature, multi-color, network connectivity) have also found their way to my teardown table, while others (integrated speakers, candelabra and other shapes) are still awaiting their turns in the analysis spotlight.
But one variant I didn’t foresee at all back in 2018, albeit obvious in retrospect (then again, what isn’t?) is a light bulb with an integrated rechargeable battery that enables ongoing illumination (at least for a while) when premises power is down. I came across just such a device at the beginning of this year, on sale at the time for $8.99 at my local Ace Hardware store. It’s a Feit Electric A21 E26 LED smart bulb, with soft white color temperature and 60-watt incandescent equivalence. Here are a couple of stock photos to start, along with a promo video:
Strangely, I actually can’t even find this particular product on Feit Electric’s website any longer, only a smaller (and more standard) A19-sized product variant:
Let’s as usual begin with some packaging shots:
One thing you might have noticed right away is that the packaging on my unit doesn’t look anything like the stock image from Ace Hardware’s site. The divergence grows even more stark when you get the light bulb (and associated sliver of literature, both as-usual also accompanied by a 0.75″ (19.1 mm) diameter U.S. penny for size comparison purposes) out of the box:
The stock photo shows what I assume are passive air vent holes midway all around the light bulb, but no such holes exist in my unit. Neither does mine contain a promotional “Feit Electric” stamp between the holes and the base; the markings on mine are all-the-way-around not to mention, and thankfully more informative:
Next, let’s remove the rubberized cover from the on/off switch (which finds use when the bulb is running from its integrated battery) shown in the prior photo:
A quick look at the base end, conventional in appearance this time versus the added-contact one seen with the three-way LED bulb back in March:
And now, aided by a hacksaw, let’s get inside that globe:
Everything looks standard so far…aside from that cylindrical object sticking out of the center!
Across the multiple LED teardowns I’ve done to date, I’ve encountered multiple interface implementation options for communicating between (and powering) the LED array “plate” and the PCB embedded in the base. This time it’s a five-wire harness, which I disconnect next:
Next off are the three screws holding the LED array plate in place:
And now, aided by a diminutive flat head screwdriver, the plate itself achieves liftoff:
Next to go is the cylindrical object in the center which, yes, is the battery. Let’s first get that white plastic protective cap off the top:
Now for the battery itself, which lifts out easily once I disconnect its two-wire harness connection to the PCB inside the base:
The specs are blatantly stamped on the outside; I’ll be holding onto it for future-project use (as, apparently, some folks are also doing with “disposable” vape unit embedded battery cells):
One more metal plate to go before the circuitry inside comes into full view:
Voila!
Two more screws to go (detail-oriented readers have already noticed that the end of one of the three earlier screw posts snapped off at some prior point in the teardown process):
And at this point, the PCB itself achieves liftoff, or at least starts to:
What’s holding it back, as anyone who’s already seen any previous teardowns in this series knows, is the multi-wire connection to the now-twisted-off base (two wires in most cases, three for the three-way):
The black “neutral” wire press-connects to the base during bulb assembly, while the red “hot” wire is crimped to the center “hot” contact and easily detached:
At this point, the PCB comes right out of the chassis:
Disconnect the other end of the five-wire harness that originally fed the LED array plate:
At the top is the on/off switch, to its left in the upper left quadrant is the connector for the five-wire harness, and below and to the left of it is, at left, the two-wire connector that feeds the battery (for recharging) and is fed by the battery (when the bulb is running off DC):
Underneath the large green capacitor in the lower left quadrant are even more passives, etc.:
Now for the PCB underside:
Finally, flipping the PCB back over, I’ll conclude with shots of all four sides:
As I glance over the circuitry, I’m struck by how different it would look were this a battery-backed incandescent-based bulb (putting aside how impractical such a product would be in real-life usage, due to incandescent illumination’s pathetically low efficiency). As mentioned at the beginning of this piece, one key reason why enhanced-featured LED light bulbs exist, aside from the LEDs’ inherent high efficiency which frees up power budget sourced from the socket for other uses, is that everything past the initial AC-to-DC conversion stage is DC-powered, including the embedded battery-charging circuitry in this case. Were this an incandescent with battery backup you’d need two inefficient conversion stages: AC-to-DC to charge the battery, and DC-to-AC to drive the filaments from the battery when premises power was absent. Yuck!
Anything strike you as particularly interesting in this design? Sound off in the comments!
—Brian Dipert is the Editor-in-Chief of the Edge AI and Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.
Related Content
- Teardown: What killed this LED bulb?
- Teardown: Zigbee-controlled LED light bulb
- Teardown: Wi-Fi LED light bulb
- Teardown: Bluetooth-enhanced LED bulb
- Teardown: Cutting into a multicolor LED light bulb
- Freeing a three-way LED light bulb’s insides from their captivity
- The LED: incandescent light bulb heir apparent
The post Dissecting a battery-backed LED light bulb appeared first on EDN.
GMSL technology eyes a place on ADAS bandwagon
Gigabit multimedia serial link (GMSL), introduced in 2004 as a 1 Mbps single-channel serializer and deserializer (SerDes) pipe, has come a long way in its ability to transport everything over a single wire while bridging video, peripherals, data, reference clock, timing, and power. It’s now aiming to accelerate the adoption of advanced driver assistance systems (ADAS) and in-vehicle infotainment (IVI), boasting as many as 7 displays by 2027.
GMSL is a SerDes IC for high-speed communication and offers multi-stream support over a single cable. According to Analog Devices, its GMSL technology has been utilized by 25 OEMs across several IVI and camera connectivity applications.
Figure 1 GMSL bridges video, peripherals, data, reference clock, timing, and power over a single cable. Source: Analog Devices
“The PHY we use for transmitting video from sensor to a system-on-chip (SoC) to a display is very robust and can withstand noise and cable temperature,” said Balagopal Mayampurath, VP and GM for serial links at Analog Devices. That allows automakers to fine-tune the car for specific users, advanced sensing and processing, and connectivity to the outside world.
GMSL started in 2004 with 1 Gbps single channel connectivity and has expanded to 12 Gbps across three generations. Each generation is backward compatible with the previous generation, which minimizes software overhead besides promoting sustainable reuse.
Compared to GMSL2, which offers a speed of 6 Gbps and has been widely employed in the automotive industry, GMSL3 provides 12 Gbps in the forward channel and 187 Mbps in the reverse channel. Moreover, the third-generation GMSL technology supports uncompressed 4K resolution at 90 frames per second, delivering smooth video streams.
Analog Devices claims that it’s starting to find GMSL applications in other areas, such as industrial automation, aerospace, digital healthcare, and intelligent infrastructure. For instance, Leopard Imaging has employed GMSL3 technology in its embedded vision solutions compatible with different artificial intelligence (AI) platforms. With the GMSL3 interface, the company’s cameras can operate at full MIPI CSI-2/D-PHY v1.2 specifications as well as integrate three 4K streams to transmit over one power-over-coax (PoC) cable.
Figure 2 GMSL technology has been incorporated into cameras for embedded vision applications. Source: Leopard Imaging
GMSL in ADAS, IVI
In automotive, GMSL, the connectivity solution for ADAS and IVI applications, facilitates multiple high-definition displays. “Beyond 2027, there will be more displays in the car,” Mayampurath said. He adds that technologies like GSML will be critical in high-resolution implementation to transport and synchronize error-free video from high-resolution cameras to a processor.
“If you look at where automotive cabinet experience is headed, it’s clear that users are looking for an immersive experience in terms of audio, video, and connectivity,” he said. “They are looking for some level of personalization hooked to their smartphones.” In other words, a driver gets into the car and automatically adjusts to his or her settings.
Figure 3 GMSL3 supports ADAS and L2+ autonomous driving while facilitating high-definition displays for a more immersive cabin experience. Source: Analog Devices
Then there are safety features like support for ASIL B and real-time link diagnostic capability. It also meets stringent EMI/EMC requirements and offers connectivity with ultra-low latency and high data bandwidth to support the growth of sensors for ADAS.
GMSL in the car of the future
Today, most automakers offer Level 2 or Level 2 Plus autonomous driving capabilities with platforms using six cameras and one radar. And, as automakers increase the level of automated driving, they could go to as many as 18 cameras and 5 radars.
The pace of innovation has picked up because the automotive market is moving fast, going through an inflection point with next-generation infotainment applications. That encompasses higher resolution displays and connectivity of different types.
“Change is happening as OEMs move toward sensor fusion model, where data is brought from different sensors and the fused data is employed for automated driving,” Mayampurath said. “The idea is to reduce the amount and number of cables by aggregating sensor data on zonal hubs, and then use higher bandwidth connectivity to go to central compute.”
At this automotive technology junction, GMSL promises to reduce design and development efforts by drastically minimizing wiring and system complexity. It also claims to enable a new level of functional safety performance for high-speed video links.
“We see GMSL as a foundation technology to a rich cabin experience for its ability to offer immersive audio, high-definition video, and noise cancelation,” said Mayampurath. That makes GMSL a designer-centric, scalable SerDes technology vying for a significant role in accelerating the adoption of ADAS and automated driving.
Related Content
- The evolving automotive industry
- ADAS takes greater control in 2015
- 6 Trends on ‘Perception’ for ADAS/AV
- The Future of Automotive Connectivity
- Using GMSL SerDes Devices in a Dual Automotive (ECU) Application
The post GMSL technology eyes a place on ADAS bandwagon appeared first on EDN.
The saga of chip technology transfer reaches France
Four senior executives at the French semiconductor outfit Ommic SAS are currently under investigation for sharing confidential technology with entities in China and Russia while skirting sanctions and export controls. Two French and two Chinese citizens working at the company have been placed under formal investigation since March this year after the allegation surfaced in the French newspaper Le Parisien.
The French industry minister Roland Lescure said that when the government warns about the risk of industrial espionage, it’s not like playing James Bond. Ommic’s technology portfolio includes gallium nitride (GaN) semiconductors, widely used in defense and aerospace electronics for their ability to operate on high voltages and sustain high temperatures.
Ommic, located in Limeil-Brévannes, France, 20 km outside Paris, has around 100 employees, and it focuses on metal organic chemical vapor deposition (MOCVD) epitaxial growth, wafer processing and fabrication, and monolithic microwave integrated circuits (MMICs) specializing in gallium arsenide (GaAs) and GaN semiconductors.
Le Parisien, which first reported on the case, also discovered that a Beijing-based Chinese businessman with ties to China’s defense industry had bought 94% of Ommic’s shares via a French-based investment fund in 2018. French authorities have stripped the majority stake of this Chinese investor while temporarily placing the company under state control before its sale to Macom.
The Lowell, Massachusetts-based chip company, which designs and manufactures RF, microwave, analog and mixed-signal, and optical semiconductors, paid $42.5 million to acquire Ommic in May 2023. “We look forward to building upon the existing team’s expertise in material science, semiconductor wafer processing and millimeter-wave MMIC design,” said Macom’s president and CEO Stephen G. Daly.
According to news reports, investigators have uncovered nearly $13 million worth of suspected technology exports. Here, the company’s French manager is suspected of personally delivering chips to Russian clients. Moreover, semiconductors products are suspected to be exported to armament manufacturers in China with the help of forged paperwork.
The semiconductor technology restrictions and export controls have taken a new turn with this Bloomberg news report about the transfer of French technology to China, especially when it’s about Ommic’s wafer processing technology. Semiconductor wafer processing and emerging technologies like GaN materials are critical to China’s ambitions of cultivating a robust chip industry.
Related Content
- China Struggles to Crack Chip Manufacturing
- EU Imposes Restrictions on Chip Exports to Russia
- U.S. Ban on Huawei Seen Widening China Chip War
- Commerce Adds Limits on Exports of Chip Tech to China
- U.S. Chip Sanctions ‘Put Temporary Checkmate on China’
The post The saga of chip technology transfer reaches France appeared first on EDN.
Broadband amp software optimizes output signals
A software package for the R&S BBA300 RF amplifier allows users to set the bias point and high power output via a straightforward web GUI. The BBA300-PK1 software option provides access to the amplifier’s extensive parameter sets for use in a variety of applications, from EMI immunity and RF component testing to medical and scientific particle accelerators.
Models in the BBA300 series of broadband amplifiers offer continuous frequency ranges from 380 MHz to 6 GHz at up to 300 W of output power. Units support amplitude, frequency, phase, pulse, and complex OFDM modulation.
The BBA300-PK1 software introduces two tools for optimizing the output signal to meet a wide range of test requirements. The first tool allows bias point adjustment between Class A and Class B. Shifting to Class AB enables pulsed signals to be reproduced accurately at high power, while improving efficiency.
The second tool offers a high power mode that provides high maximum power with a well-matched RF path, while VSWR mode furnishes rated power with high tolerance to load mismatches. VSWR mode is useful in EMC applications because it maintains rated power up to a VSWR of 6:1 before the amplifier gradually reduces its rated power to 50% for self-protection in the event of an open or short circuit.
For more information on the BBA300 broadband amplifier, as well as the BBA300-PK1 software option, use the link to the product page below.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post Broadband amp software optimizes output signals appeared first on EDN.
RF analyzer touts software-defined configuration
Keysight’s N9912C handheld RF analyzer is customizable with over 20 downloadable software applications and license-key activated frequency and bandwidth options. Operating at up to 10 GHz, the instrument can be configured with vector network analyzer (VNA), cable and antenna tester (CAT), and spectrum analyzer (SA) functions. Mix and match VNA, CAT, and SA options with the desired maximum frequency to get precisely the capabilities you need in a single field device.
The N9912C, part of the company’s FieldFox lineup, performs spectrum and network analysis down to 3 kHz and up to 10 GHz to test and troubleshoot a wide range of high-frequency and wireless applications. It allows the capture of elusive signals with a gap-free, real-time bandwidth of up to 40 MHz and measures all four S-parameters simultaneously with a 115-dB range. In addition, the N9912C can be used to perform over-the-air measurements for 5G NR and LTE. The analyzer uses GPS/GNSS for geolocation and timestamping.
The battery-operated analyzer weighs 3.4 kg (7.4 lb) and comes with a rechargeable Li-ion battery, AC/DC adapter, power cord, and carrying case. Request a price quote using the link to the product page below.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post RF analyzer touts software-defined configuration appeared first on EDN.
Upgraded analyzer performs laser diode measurements
Anritsu has expanded the functions of its MS9740B optical spectrum analyzer for testing pulsed laser diode chips during production. The measurement function was added to meet the increased production of high-power laser diode (LD) chips, driven by higher communication bit rates and longer LiDAR detection ranges.
By eliminating the need for a trigger signal, the MS9740B analyzer accelerates the optical spectrum evaluation of LD chips. The expanded capability of the MS9740B also ensures measurement reproducibility of ±1.4 dB for side-mode suppression ratio (SMSR). Low SMSR variation improves LD chip yield and production efficiency.
Intended for production environments, the MS9740B optical spectrum analyzer provides a high dynamic range of up to 70 dB (±1 nm from peak wavelength) and fast sweep speeds of <0.2 s over a wavelength range of 600 nm to 1750 nm. Anritsu says that the MS9740B reduces measurement times by as much as half compared to the previous model.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post Upgraded analyzer performs laser diode measurements appeared first on EDN.
Android wireless module packs a processing punch
With compute power of up to 48 TOPS, Quectel’s SG885G-WF Android smart module satisfies high-performance industrial and consumer IoT applications. The module is powered by a Qualcomm QCS8550 SoC with built-in octa-core Kryo CPU, Adreno 740 GPU, Adreno 8550 VPU, and Hexagon DSP. It also supports Wi-Fi 7, Bluetooth 5.3, and 2×2 Wi-Fi MIMO for enhanced IoT connectivity.
The SG885-WF module provides robust video capabilities to ensure smooth and high-quality video processing. Video encoding at 4K and 8K resolution is handled at a rate of 120 fps and 30 fps, respectively. Video decoding at the same 4K and 8K resolution is performed at 240 fps and 60 fps, respectively.
Running under the Android operating system, the SG885G-WF is well-suited for a wide range of IoT and M2M applications, such as videoconferencing systems, live streaming devices, edge computing, robots, drones, and AR/VR. It packs 12 Gbytes of LPDDR5X and 256 Gbytes of flash memory. Onboard interfaces include LCM, camera, PCIe, UART, USB, I2C, and SPI.
The SG885G-WF Android smart module comes in a 49×51×4.25-mm LGA package and is sold through Quectel’s distributor network.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post Android wireless module packs a processing punch appeared first on EDN.
Demo board manages power for space-grade SoC
A reference design from Renesas pairs its radiation-hardened components for power management with AMD’s space-grade Versal XQRVC1902 adaptive SoC. Developed in collaboration with AMD, the ISLVERSALDEMO2Z reference design provides all the power rails required by the Versal device, including a low 0.80-V core voltage supply that can source up to 140 A.
As core voltages decrease and currents increase for FPGAs and ASICs, it has become more difficult to meet the stringent power requirements of these devices to ensure that they operate error-free. This is especially critical in space missions where power availability is limited and systems are exposed to extreme temperatures and radiation for an extended period of time.
The demo board’s ICs support a wide range of power rails for next-generation space avionics systems that require tight voltage tolerances, high current, and efficient power conversion, while withstanding the harsh environment of space. All of the power management devices have been tested and verified to withstand exposure to high levels of radiation and come in small-footprint packages. These Intersil-brand components include the ISL73847SEH dual-output PWM controller, the ISL73041SEH and ISL71441M GaN FET half-bridge drivers, and the ISL73007SEH POL regulator.
All products are available now, and the ISLVERSALDEMO2Z reference design is available on request in limited quantities. Contact the sales team or visit the Renesas website.
Find more datasheets on products like this one at Datasheets.com, searchable by category, part #, description, manufacturer, and more.
The post Demo board manages power for space-grade SoC appeared first on EDN.
How solar inverters are cashing in on silicon carbide
Silicon carbide (SiC) power semiconductors, making headlines for their vital role in electric vehicle (EV) inverters and charging infrastructure designs, are also steading making headway in large-scale renewable energy installations like solar inverters. This wide bandgap (WBG) semiconductor technology is turning solar inverters—ranging from utility to residential—into smaller, lighter and more efficient systems while minimizing energy loss and reducing overall system cost.
Case in point: onsemi has secured $1.95 billion in long-term supply agreements for providing SiC-based power semiconductor solutions to leading solar inverter manufacturers. Earlier this year, Navitas Semiconductor announced that solar inverter manufacturer KATEK is adopting its GeneSiC power semiconductors for improved efficiency, size, weight, and cost.
Figure 1 GeneSiC MOSFETs claim to offer 30% longer short-circuit withstand time and stable threshold voltage for easy paralleling. Source: Navitas
The Munich, Germany-based KATEK develops and produces power electronics for grid inverters, energy storage, and control technology for photovoltaic and fuel cell systems. Its Steca solar inverters convert DC power from a string of solar panels into 4.6-kW AC power for use in the home, returning to the grid, or being stored locally for later use.
Solar inverters convert direct current (DC) electricity solar panels generate to grid-compatible alternating current (AC). In the conversion process, however, some energy is lost as heat. Here, SiC semiconductors, though more expensive than silicon solutions, offer higher switching speeds and efficiency, allowing transformers, capacitors, heat sinks, and ultimately, packages to be smaller.
When SiC devices facilitate a significant increase in switching frequency, it shrinks the size and weight of passive components, which optimizes the size and weight compared to legacy silicon-based inverters. That, according to Infineon’s senior executive Peter Wawer, saves cost at the system level.
Fronius Solar Energy, a solar inverter supplier based in Munich, Germany, has incorporated Infineon’s CoolSiC MOSFETs in its power modules. The company’s Symo GEN24 Plus solar inverters facilitate power for direct use in the household while also offering an interface for energy storage systems.
Figure 2 Fronius claims that the use of 1,200-V CoolSiC MOSFETs has significantly improved the functionality of its Symo GEN24 Plus solar inverter while its size remains comparable. Source: Infineon
According to “Renewables 2022,” a report published by the International Energy Agency (IEA), installed solar power capacity is expected to exceed that of natural gas in 2026 and coal by 2027. That will make solar power the world’s largest energy source while also recording a 3x increase in installed capacity from 2022-2027.
Next, research company BloombergNEF (BNEF) claims that the global levelized solar electricity cost is now 40% lower than coal and natural gas. That makes solar power one of the fastest-growing markets. And SiC devices seem prominent in this energy revolution encompassing power grids and energy storage systems.
Related Content
- Wafer supply deals herald the upcoming SiC boom
- SiC and resurgence of semiconductor vertical integration
- Chipmakers’ quest for wafers supply affirms SiC adoption
- Silicon carbide (SiC) and the road to 800-V electric vehicles
- APEC 2023: SiC moving into mainstream, cost major barrier
The post How solar inverters are cashing in on silicon carbide appeared first on EDN.
Lightning as an energy harvesting source?
We’re always looking to harvest energy from diverse, nominally “free” sources such as wind, water, solar, and even less-dense possibilities such as vibration and friction. Then there are lightning strikes which are potential energy sources are wasted, as well as often being destructive. (Lighting strikes are surprisingly common; the European Meteosat Third Generation satellite launched in December with cameras which can track and record lightning strikes, even the smallest and fastest ones, day and night, over more than 80 percent of Earth’s surface.)
Could that wasted-energy situation change? Unlikely – but as we know, it’s “never say never” when it comes to technology and advances. Back in January, a European-based team published an interesting paper in Nature Photonics with the surprisingly simple title “Laser-guided lightning.”) It detailed how they blasted repeated pulses from a high-power laser to guide lightning strikes that were as far as two miles away down to a relatively small, grounded rod near their Swiss-mountain setup, Figure 1 [1].
Figure 1 a) Layout of the experimental setup on top of the Säntis Mountain in Switzerland. b) Photography of the experiment with the second harmonic of the laser beam used to visualize the laser path. Source: “Laser Guided Lightning”, Nature photonics, 2023
The likely explanation is that the laser pulses super-heated the air, which caused the air to become electrically conductive along the path of the laser, Figure 2.
Figure 2 Snapshots of the lightning event of 24 July 2021 recorded in the presence of the laser. Source: “Laser Guided Lightning”, Nature photonics, 2023
This ability to perhaps direct a lightning strike brings up an obvious question: why not channel this energy to some sort of energy storage system (ESS)? After all, that energy is otherwise wasted since it is truly and literally grounded. (I’ve even seen so-called “experts” suggest this as a no/low-cost energy source.)
If only it were that easy. A good first question is this: how much energy and power is there in a lightning strike? The answer spans a wide range, but according to Tess Light in the Space and Remote Sensing group at Los Alamos National Laboratory, lightning is “both incredibly powerful and crazy fast” and each strike delivers about fifty thousand amps in just microseconds, with a megavolt punch.
Furthermore, while each strike delivers around five or ten gigajoules of energy (one GJ = 109 joules), much of that energy is lost in heating the air and so is not “capturable” electrical energy. (To give you some sense of scale, a large EV battery has a capacity of about 100 kilowatt hours, or 3.6 × 108 joules.) So even though lightning does seem to have a lot of energy, it’s really not that much and you’d need a lot of strikes.
Why not make it simple, and use a totally passive, simple lightning rod? The answer has to do with the lightning rod’s range of effectiveness.
Although the exact number is a function of atmospheric, ground, and other conditions, a general guide is that a standard rod can attract lightning within a radius equal to its height. Thus, you’d need a large “farm” of lightning rods with some combination of a large amount of land area and tall rods. The laser-based scheme increases the capture area and decreases the rod height requirements.
There’s yet another issue: Is directing lightning via lasers worthwhile from an energy-balance viewpoint? Even if it is technically viable, the laser scheme has its own major drawback, as it uses a lot of power itself. The researchers used a Yb:YAG laser emitting picosecond pulses at 500 MJ energy with a wavelength of 1,030 nm and a 1-kHz repetition rate. In other words, the energy cost of driving the lasers was greater than that of the captured lightning.
This imbalance is somewhat analogous to the recent fusion “success” report from the National Ignition Facility (NIF) in California. They focused about 2 MJs of energy from hundreds of lasers onto a tiny capsule of fusion fuel, sparking an explosion that produced about 3 MJ of energy. While this is a breakthrough in many ways and took years (and $) to happen, the laser-drive subsystem itself is only 1% efficient, so about 200 MJ was needed as the total project input energy—not a winning equation.
Even if the power-balance was favorable for harvesting lightning, it still doesn’t address another intractable problem between captured lightning and any known ESS. How do you get that much power (the rate of energy transfer) into the battery? No existing battery or ESS could survive that enormous power surge—and that’s what it is.
Of course, the next step in such research almost always involves “bigger” and “more money.” The lead researcher was quoted in news reports as saying that they would like to increase the distance that the laser can guide the lightning to hundreds of meters. That is theoretically possible with a bigger, more powerful laser, but the prototype laser used in this demonstration cost 2 billion euros. (If that’s out of your budget, you may have to stick with something simpler, such as a four-transistor DIY lightning detector, see “Simple analog-centric circuits expand STEM perspectives.”)
There’s an interesting historical link to this entire lightning-capture story. Before current-flow electricity as we use today existed, Benjamin Franklin actually captured some lightning energy in a Leyden jar (an early type of capacitor) during his famous kite-flying experiment. This is not myth or legend, as it was properly documented at the time (the Wikipedia entry “Kite experiment” has a clear discussion and credible reference links).
Franklin split the kite’s “downlead” by using wet hemp string as a conductive lead to the Leyden jar and an insulating silk thread on the other branch to his son who was flying the kite, who stayed in a shed so as to not get wet. He succeeded in showing the relationship between static electricity and lightning before electricity (static or otherwise) was understood or had any practical uses.
As for harvesting lightning today, it looks like it’s the same story: no matter how attractive it may seem at first, large-scale energy harvesting doesn’t come easy, and there is no “free lunch” out there.
Bill Schweber is an EE who has written three textbooks, hundreds of technical articles, opinion columns, and product features.
Related Content
- Lightning strike becomes EMP weapon
- Innovative IC can’t trap lightning, but can let you know it’s coming
- Lightning rod ball
- Forget Tesla coils and check out Marx generators
The post Lightning as an energy harvesting source? appeared first on EDN.
Network-on-chip (NoC) interconnect topologies explained
Today’s complex system-on-chip (SoC) designs can contain between tens to hundreds of IP blocks. Each IP block may have its own data width and clock frequency and employ one of the standard SoC interface protocols: OCP, APB, AHB, AXI, STBus, and DTL. Connecting all these IPs is a significant challenge.
Functional IP blocks connect to the network-on-chip (NoC) via sockets. In the case of an initiator IP, the socket serializes and packetizes the data generated by the IP, assigns an ID to the packet, and dispatches it into the network. When the packet arrives at its destination IP, the associated socket extracts the data from the packet and transforms it into the protocol required by the IP. A large number of packets can be in flight throughout the network at any given time.
Topological types and target applications
The term topology refers to how the constituent parts of something are interrelated or arranged. In these discussions, the topology of an NoC refers to how the IP nodes are connected.
NoCs are incredibly versatile. As illustrated in the figure below, in addition to an NoC implementation of a crossbar—this is not the same as a traditional crossbar switch implementation—other topologies commonly deployed in today’s SoCs include 1D stars, rings and trees, along with 2D meshes and toruses. More complex structures, including 3D cubes and 4D hypercubes, are also possible but are rarely employed in practice.
High-level representations are shown for common NoC topologies. Source: Arteris
When comparing different NoC topologies, a typical mega metric is the concept of wire cost, which essentially embraces all the expenses, power, area, and congestion associated with the interconnect portion of the NoC. Bearing this in mind, a high-level summary of the various NoC topologies illustrated in the above figure is as follows:
NoC crossbar topology
Advantages:
- Provides a direct and simultaneous connection between all nodes.
- No contention for shared resources.
- High throughput and low latency.
Disadvantages:
- High wire cost in terms of power for larger systems.
- Scalability issues due to the limited number of ports in the switch.
Comparison:
- Lowest latency and highest throughput.
- Highest power and wire costs.
Use cases:
- High-performance computing systems, such as supercomputers or data centers, require low latency and high throughput for transferring large amounts of data.
- High-end graphics processing units (GPUs) and other high-speed processing units require high bandwidth and low latency communication.
- Some commercial interconnect providers offer crossbar interconnects as general-purpose solutions. However, topologies like mesh or tree may be more appropriate for small embedded systems.
Star topology
Advantages:
- Simplicity
Disadvantages:
- Offers only a single point of failure in the central hub or switch—if the hub or switch fails, the entire system will fail.
- The total bandwidth of a star topology is limited by the capacity of the central hub or switch. If the processing elements require more bandwidth than the hub or switch can provide, the system’s performance will be limited.
- While a star topology is scalable to a certain extent, it may become difficult and expensive to scale beyond a certain number of processing elements.
- As messages must pass through the central hub or switch, a star topology may have higher latency than a mesh or torus topology.
Use cases:
- Can be a good choice for small embedded systems or consumer electronics with limited processing elements and low bandwidth requirements. However, it may not be suitable for systems with high bandwidth requirements or where fault tolerance and scalability are critical.
Ring topology
Advantages:
- Low power and wire costs.
- Provides a guaranteed path between nodes.
Disadvantages:
- Low scalability due to the fixed number of nodes in the ring.
- High latency caused by messages needing to traverse the entire ring.
Comparison:
- Lower power and wire cost compared to mesh and torus topologies.
- Higher latency than the mesh and tree topologies.
Use cases:
- Smaller designs with limited numbers of nodes, such as embedded systems or sensor networks.
- Systems in which messages must be delivered in a specific order, such as token rings in real-time systems.
Tree topology
Advantages:
- Scalable with low power and wire costs for larger systems.
- Fault tolerance due to having multiple paths between nodes.
Disadvantages:
- High contention and congestion at the root of the tree.
- Higher latency than mesh topology.
Comparison:
- Lower power and wire costs than crossbar topology.
- Lower latency than ring topology.
Use cases:
- Hierarchical systems include systems with an initiator-target structure or a central processing unit (CPU) with multiple peripheral devices.
- Clustered systems in which nodes are grouped and must communicate with their group leaders.
- Compared to a mesh topology, a tree topology can be more cost-effective and easier for small systems to implement, as they require fewer wires and switches. Tree topology can also be more energy efficient, as data can be routed directly to its destination without passing through multiple nodes.
Mesh topology
Advantages:
- Scalable with low power and wire costs for larger systems.
- Fault tolerant due to having multiple paths between nodes.
Disadvantages:
- High contention and congestion in the middle of the mesh.
- Higher latency than the crossbar topology.
Comparison:
- Lower latency and higher throughput than ring and tree topologies.
- Lower power and wire cost than the crossbar topology.
Use cases:
- Systems with many nodes and moderate communication requirements.
- For consumer electronics or small embedded systems, a mesh topology can be better than a crossbar topology. This is because a mesh topology is generally more cost-effective and simpler to implement than a crossbar topology. It uses fewer wires and switches, making it well-suited for smaller systems with limited resources.
- Multi-core processors where each core is connected to its nearest neighbors for inter-core communication.
- Mesh topologies can be beneficial for large-scale data center networks where scalability and fault tolerance are essential.
- Mesh topologies can also be a good option for AI applications. A mesh topology provides good fault tolerance and scalability and can adapt to different communication patterns and workloads.
Torus topology
Advantages:
- Provides a natural and regular layout for 2D or 3D chip designs.
- High fault tolerance is due to the multiple paths between nodes.
- Low latency and high throughput for small- and medium-sized systems.
Disadvantages:
- High power and wire costs for larger systems.
- More complex implementation compared to other topologies.
Comparison:
- Lower power and wire cost than the crossbar topology.
- Lower latency than the ring topology.
- Higher throughput than the mesh and tree topologies.
Use cases:
- Systems with regular 2D or 3D grid structures, such as graphics processors or video processing units.
- Large-scale systems with high fault-tolerance requirements, such as data centers or cloud computing systems.
- The torus is an alternative topology for AI applications as it provides redundant paths for data transmission and ensures high availability and reliability. The torus topology can also provide scalability, as it can be easily expanded by adding more nodes.
No “one size fits all”
While simple designs may still be satisfied with a single network, complex SoCs small and large will benefit from implementing NoCs with similar or disparate topologies.
Some SoCs employ hierarchical tree structures and multiple separate trees instead of a single tree topology. In some cases, one portion of the chip, such as a machine learning (ML) inference engine, may take advantage of a mesh topology. At the same time, other areas may be better served by one or more different topologies.
The main thing to remember is that each SoC presents unique interconnect challenges. NoCs offer the most powerful and versatile interconnect solutions available. Some SoCs can benefit from a mixture of NoC topologies. Most importantly, there is no one-size-fits-all solution when it comes to NoCs.
Selecting the best NoC topology or mix of topologies for a specific application is not always a simple task. Your IP partner can offer advice on the selection of the best technology for your design and its NoC requirements.
Andy Nightingale, VP of product marketing at Arteris, has over 35 years of experience in the high-tech industry, including 23 years spent on various engineering and product management positions at Arm.
Related Content
- SoC Interconnect: Don’t DIY!
- Network on chip eases IP connect
- What is the future for Network-on-Chip?
- The network-on-chip interconnect is the SoC
- Arteris spins packet-based network-on-chip IP
The post Network-on-chip (NoC) interconnect topologies explained appeared first on EDN.
Pages
