Always-On Battery Life Improvement with ML Chip

Devices that must always remain on must conserve power in every way possible to extend their battery life. Their design starts with the lowest possible system power and every mode of operation must consume the bare minimum power necessary for operation. Now, with AML100, an analog machine learning or ML chip from Aspinity, it is possible to cut down the system power by up to 95%, even when the system always remains on. AML100 consumes less than 100 µA of always-on system power. This opens new types of products for biometric monitoring, preventive and predictive maintenance, commercial and home security, and voice-first systems, all of which are systems that continuously must remain switched on.

The movement of data to and from a system consumes power. One of the most effective ways of reducing power consumption is, therefore, minimizing the amount and movement of data through a system. The AML100 transfers the machine learning workload to the analog domain where it consumes ultra-low levels of power. The chip determines the relevancy of data with highly accurate and near-zero power. By intelligently reducing the data at the sensor, while it is still in the analog mode, the tiny ML chip keeps its digital components in low-power mode. Only when it detects important data, does the chip allows the analog data to enter the digital domain. This eliminates the extra power consumption in digitizing, processing, and transmission of irrelevant analog data.

The AML100 consists of an array of independent analog blocks configurable to be fully programmable with software. This allows the chip to support a wide range of functions that include sensor interfacing and machine learning. It is possible to program the device in the field, using software updates, or with newer algorithms that target other always-on applications. When it is in always-sensing mode, the chip consumes a paltry 20 µA, and it can support four analog sensors in different combinations like accelerometers, microphones, and so on.

At present, Aspinity is producing the AML100 chip in sampling numbers for key customers. The chip has dimensions of 7 x 7 mm and is housed in a 48-pin QFN package. Aspinity has slated the volume production of this chip for the fourth quarter of 2022 and is presently offering two evaluation kits with software. One of the kits is for glass breakage and T3/T4 alarm tone detection, while the other is for voice detection with preroll collection and delivery. Other kits with software for other applications are also available from Aspinity on request.

AML100 is the first product in the AnalogML family from Aspinity. It detects sensor-driven events from raw, analog sensors by classifying the data. It allows developers to design edge-processing devices with significantly low power consumption, those that are always on. The device has a unique RAMP or Reconfigurable Analog Modular Processor technology platform that allows the AML100 to reduce the always-on system power by more than 95%. This enables designers to build ultra-low power always-on solutions with edge-processing techniques for biomedical monitoring, predictive and preventive maintenance for industrial equipment, acoustic event monitoring applications, and voice-driven systems.

Advantages of Additive Manufacturing

Additive manufacturing, like those from 3-D printers, allows businesses to develop functional prototypes quickly and cost-effectively. They may require these products for testing or for running a limited production line, allowing quick modifications when necessary. This is possible because these printers allow effortless electronic transport of computer models and designs. There are many benefits of additive manufacturing.

Designs most often require modifications and redesign. With additive manufacturing, designers have the freedom to design and innovate. They can test their designs quickly. This is one of the most important aspects of making innovative designs. Designers can follow the creative freedom in the production process without thinking about time and or cost penalties. This offers substantial benefits over the traditional methods of manufacturing and machining. For instance, over 60% of designs undergoing tooling and machining also undergo modifications while in production. This quickly builds up an increase in cost and delays. With additive manufacturing, the movement away from static design gives engineers the ability to try multiple versions or iterations simultaneously while accruing minimal additional costs.

The freedom to design and innovate on the fly without incurring penalties offers designers significant rewards like better quality products, compressed production schedules, more product designs, and more products, all leading to greater revenue generation. Regular traditional methods of manufacturing and production are subtractive processes that remove unwanted material to achieve the final design. On the other hand, additive manufacturing can build the same part by adding only the required material.

One of the greatest benefits of additive manufacturing is streamlining the traditional methods of manufacturing and production. Compressing the traditional methods also means a significant reduction in environmental footprints. Taking into account the mining process for steel and its retooling process during traditional manufacturing, it is obvious that additive manufacturing is a sustainable alternative.

Traditional manufacturing requires tremendous amounts of energy, while additive manufacturing requires only a relatively small amount. Additionally, waste products from traditional manufacturing require subsequent disposal. Additive manufacturing produces very little waste, as the process uses only the needed materials. An additional advantage of additive manufacturing is it can produce lightweight components for vehicles and aircraft, which further mitigates harmful fuel emissions.

For instance, with additive manufacturing, it is possible to build solid parts with semi-hollow honeycomb interiors. Such structures offer an excellent strength-to-weight ratio, which is equivalent to or better than the original solid part. These components can be as much as 60% lighter than the original parts that traditional subtractive manufacturing methods can produce. This can have a tremendous impact on fuel consumption and the costs of the final design.

Using additive manufacturing also reduces the risk involved and increases predictability, resulting in improving the bottom line of a company. As the manufacturer can try new designs and test prototypes quickly, digital additive manufacturing modifies the earlier unpredictable methods of production and turns them into predictable ones.

Most manufacturers use additive manufacturing as a bridge between technologies. They use additive technology to quickly reach a stable design that traditional manufacturing can then take over for meeting higher volumes of production.

New Requirements for Miniature Motors

Innovations in the field of robotics are resulting in the emergence of smarter and smaller robotic designs. Sensor technologies and vision systems use robotic applications in warehousing, medical, process automation, and security fields. Disruptive technologies are creating newer opportunities for solving unique challenges with miniature motors. These include the robotic market for efficient and safe navigation through warehouses, predictable control of surgical tools, and the necessary endurance for completing lengthy security missions.

With industries transitioning to applications requiring collaborative robotics, they need systems that are more compact, dexterous, and mobile. Tasks that earlier required handling by human hands are driving the need for miniaturized motors for mimicking both the capability and size of the hands that accomplished the work.

For instance, multiple jointed solutions representing the torso, elbow, arm, wrist, etc. require small, power-dense motors for reducing the overall weight and size. Such compact solutions not only improve usability but also improves autonomy and safety, resulting in faster reaction times due to lower inertia. Therefore, robotic grippers, exoskeleton, prosthetic arms, and humanoid robots require small, high-power density motors. Power density is the amount of power a motor generates per unit of its volume. A motor that generates greater amounts of power in a small package, has a higher power density. This is an important factor when there is a space constraint, or where a high level of output is necessary when a limited space.

Manufacturers can miniaturize motors with high power densities. Alternately, they can increase the capability of current designs. Both options are critical in reducing the space that motion elements occupy. High efficiency is necessary to obtain the maximum power possible from a given design. Here, BLDC or brushless DC motors and slot-less motor designs in combination with efficient planetary gearboxes can offer powerful solutions in small packages. Brushless solutions are flexible enough for engineering them to meet customer requirements like long and skinny designs, or short, flat, low-profile configurations.

Smooth operation and dynamic response can result in these miniature motors being dexterous and agile. Slot-less BLDC motors achieve this by eliminating detent torque, thereby providing precise dynamic motion with their lower inertia. Applications requiring high dynamics, such as pick-and-place systems and delta robots, must be able to accelerate/decelerate quickly and constantly. Coreless DC motors and stepper motors with disc magnets are suitable for applications requiring critical characteristics like high acceleration as they have very low inertia.

Ironless brushed DC motors with their high efficiency, are the best choice for battery-powered mobile applications to extend their operational life between charges. Several robotic applications now run on battery power, thereby requiring motors with high efficiency for longer running times. Other applications require high torque at low speeds, and it is possible to achieve this by matching the motor with a high-efficiency gearbox.

Some applications that are inhospitable to humans may need robot systems capable of enduring difficult environmental conditions. This may include tremendous vibration and shock. With proper motor construction, it is possible to improve their reliability and durability when operating under such conditions.

What is Moisture Sensing?

In agriculture, where plants require watering, people often use time-controlled watering methods. While this method irrigates plants in fixed time intervals, there is no way to assess whether there is an actual need for watering. Most often, this leads to either over-watering or under-watering. Depending on weather conditions, over-watering may cause harmful water-logging, while under-watering may lead to dry stress for plants. People often mitigate the amount of water flow by using a rain sensor or controlling the water delivery based on online weather information.

Using a sensor to sense the amount of moisture in the soil and control the watering works much better. Not only does the latter method allow optimal water supply to the plants, but it also substantially reduces water consumption. Threshold levels can be set using various strategies. Any experienced gardener can recognize the start of dry stress when they notice the plants wilting slightly, or when the leaf edges start rolling.

Excessive watering does not increase the moisture in the soil, rather, it results in saturation. By delaying watering for a while, the excess water usually drains off into the subsoil. Most gardeners set the lower threshold to about 60% of the saturation level. They observe the plants and the moisture trend during the early phases to adjust the threshold levels to allow an economical and optimal automatic watering. It is necessary to position the sensor properly in the soil near the root area. For drip irrigation, it is possible to achieve a good soil moisture cycle by placing the sensor somewhere where it is neither too far nor too close to the drip location.

For working with moisture sensors, it is necessary to consider sensor selection and integration. This is because moisture sensors have two functions in a watering system. The first is they provide information about the current status of the watering. The second is they help to economically use water as a resource. Many plants are intolerant to dry soil as they are to water-logging. Moreover, while there are numerous types of moisture sensors, they have different ways of working and their life spans vary widely.

The presence of moisture in the soil can have different definitions. There is the volumetric water content, which represents the amount of water in the total amount of soil. In natural soil, the maximum volumetric water content is about 50-60 % and represents the amount of water filling all the airspace in the soil. Organic materials and peat can hold more water.

The relative mass of water in the soil is its gravimetric water content. This is determined chiefly by weighing the soil sample before and after drying. As it requires a laboratory to do the measurement, this method is not suitable for continuous monitoring in the field.

A variety of principles of physical measurements form the basis of many types of electrical sensors for measuring soil moisture. The most inexpensive is the measurement of electrical conductivity. Next are low-frequency capacitive sensors. High-frequency capacitive sensors are more expensive. Then there are tensiometers that measure the soil moisture tension.

What are Power Factor Controllers?

Connecting an increasing number of electrically-powered devices to the grid is leading to a substantial distortion of the electrical grid. This, in turn, is causing problems in the distribution of the electrical network. Therefore, most engineers resort to advanced power factor correction circuitry in power supply designs that can meet power factor standards strictly for mitigating these issues.

Most power factor correction methods popularly use the boost PFC topology. However, with the advent of wide band-gap semiconductors, like silicon carbide and gallium nitride, it is becoming easier to implement bridge-less topologies also, including the column PFC. With advanced column controllers, it is now possible to simplify the control over complex designs of the interleaved column PFC.

At present, the interleaved boost PFC is the most common topology that engineers use for power factor correction. They use a rectifying diode bridge for converting AC voltage to DC. A boost converter then steps up the DC voltage to a higher value, while converting it to a sinusoidal waveform. This has the effect of reducing the ripple on the output voltage while offering a sinusoidal waveform for the current.

Although it is possible to achieve power factor correction with only a single boost converter, engineers often use two or more converters in parallel. Each of these converters is given a phase shift to improve its efficiency and reduce the ripple on the input current. This topology is known as interleaving.

With new families of semiconductors, especially the silicon carbide type, creating power switches offers substantial improvements in their thermal and electrical characteristics. Using the new type of semiconductors, it is becoming possible to integrate the rectification and boost stages, along with two switching branches for operating at different frequencies. This is the bridge-less column PFC topology.

One of the two branches is the slow branch, and it commutates at the grid frequency, typically 50 or 60 Hz. This branch operates with traditional silicon switches, while it is primarily responsible for input voltage rectification. The second branch is the fast branch and is responsible for stepping up the voltage. Switching at very high frequencies like 100 kHz, this branch places great thermal and electrical strain on the semiconductor switches. For safe and efficient performance, engineers prefer to use wide band-gap semiconductor switches, such as GaN and SiC MOSFETs, in the second branch.

The bridge-less column PFC topology improves the performance in comparison with the interleaved boost converter. But the control circuitry is more complex due to the presence of additional active switches. Therefore, engineers often integrate the column controller to mitigate the issue.

It is possible to add more high-frequency branches for improving the efficiency of the bridge-less column PFC. Such additions help in reducing the ripple on the output voltage of the converter while distributing the power requirements equally among the branches. Such an arrangement minimizes the overall costs while reducing the layout.

Although it is possible to reach general conclusions about each topology by comparing their performance, this largely depends on the device selection and its operating parameters. Therefore, designers must be careful in considering the design for implementation.

Mobile Screen Over Your Eyes

It is no longer necessary to hold a mobile with the hands. How? Thanks to AR or Artificial Reality eyeglasses, it is now possible to transfer the screen of the mobile device to the lens of a pair of eyeglasses. Although this technology was around for a while, the glasses were rather cumbersome and bulky.

Now, Trilite Technologies of Vienna, Australia, has a newer approach to AR glasses that make them look and feel just like normal glasses. According to their CEO, Dr. Peter Weigand, so far, there have been three types of light engine technologies.

The first was the LCoS technology. This is a panel-based technology, and it requires optics with illumination. It is necessary to have a nice, homogeneous, and smooth illumination, and a waveguide must carry the input image. This is not a very efficient technique, and it has a number of optical elements, making it bulky.

The other was the MicroLED display technology. This is semiconductor-based and far superior to a reflective display as it emits its own light. However, it is still a challenge to make the display visible in outdoor applications. And, the two-dimensional display does not scale up when moving to higher FOV or Fields of View and higher resolutions.

The third was the Laser beam scanner technology. This has the highest level of miniaturization. Typically, it has an RGB laser module with three separately mounted lasers as the red, blue, and green light sources. Optics follows the laser module to merge the three beams of lasers into a single ray. A set of MEMS mirrors follows, generating the image scans for the eyeglass display. Two mirrors are necessary, one for the X- and the other for the Y-axis.

According to Weigand, the latest generation of these scanners uses a single MEMS mirror that can move in both x and y-direction. This two-dimensional mirror helps to achieve a lighter and smaller product.

Electronics create the image for display by modulating the lasers. Coupling the image to an optical waveguide allows it to be sent to the display. For this, the laser scanner uses relay optics, a rather large optical element. Coupling the laser beam scanner into the input coupler of the waveguide directly, allows the display engine to be made to a small size. The entire arrangement contains the collimating optics, the MEMS mirrors, and the three lasers.

Trilite Technologies is able to make very small scanners because of its design philosophy. They have designed their scanner such that software rather than hardware handles many of the scanning functions. The other significant contribution to the small size comes from using a single two-axis MEMS mirror rather than one mirror for each axis.

The waveguide contains the optical input coupler as an integral part. This coupler has a pattern of microstructure gratings on its surface, allowing light to enter. The output side, where the light emerges from the waveguide, also has a similar structure. The waveguide conveys the image to the lens and, at the same time, combines the incoming with the generated digital light, allowing the user to see both the digital image and the real-world scene through the eyeglass lens.

What are Floating Sensors?

Floating sensors support applications for environmental monitoring and agriculture. Designed by researchers from the University of Washington, floating sensors typically spread just like seeds of the dandelion plant do, when a drone drops them from a height. The sensors are battery-free devices, hovering over 100 meters. The sensors have electronics on board, including a capacitor for storing overnight charge, sensors, and a microcontroller for running the system. The entire structure resides in a flexible body.

The evolution of dandelions allows them to disperse their seeds further than a kilometer in the air. Although for valuable wireless sensors, it is not a good idea to drop them from great heights. However, the researchers did just that by creating a tiny device that can carry the sensor, with the wind blowing it at it tumbles towards the ground.

Just like the dandelion seeds do, the sensors too, float in the breeze. As the device is about 30 times heavier than a dandelion seed weighing one milligram is, it can travel only up to a distance of about 100 meters on a windy day. The researchers had to mimic the shape of the dandelion seeds as it was necessary to ensure that the device landed with its solar panels facing skywards.

The structure of dandelion seeds has a central point where little bristles stick out. These tend to slow down their fall. The researchers took a 2-D projection of the seed and used it to create the base design for the structure of their floating sensors. When they added more weight, the bristles started to bend inwards. The researchers then added a ring structure to make the bristles stiffer, and take up more area, allowing it to slow down the fall. The team tested more than 75 designs with various sizes and patterns using laser micro-machining.

The sensor can share data related to pressure, temperature, humidity, and light up to a distance of 60 meters. The researchers have added a capacitor to the design of their floating sensors, allowing it to store some charge for the night. As an experiment, the researchers used a drone to drop sensors from a height of 20 meters, sending the sensors sideways to about 100 meters towards a parking area.

According to the researchers, from an engineering point of view, imitating dandelion seeds allows for achieving some amazing capabilities. Although dandelion plants cannot move, they can disperse their seeds up to a kilometer away, provided the right conditions exist. The team has been trying for a similar achievement by automating the deployment of wireless sensors to create a network. Conventional methods of studying climate changes or monitoring the environment over really large geographic areas can be very expensive and time-consuming. Dandelion seeds and their dispersion methods provided the team with the necessary inspiration to create sensors that can disperse in the wind, and automate this process.

The team had to look at nature again to get good coverage over the area of interest. They mimicked the random process followed by plants to disperse their seeds. The researchers designed a large array of different structures to make them float for different periods.

Monitoring Battery Health

The prolific use of battery-powered instruments for regular use in the consumer and industrial fields requires monitoring battery health for proper functioning. Usually, a battery health monitoring system uses a microcontroller and a software user interface. This arrangement monitors all the batteries in a battery bank 24×7 and identifies weak batteries before they actually fail. This helps to improve the overall performance of the system. Stationary applications such as data centers commonly use such battery health monitoring systems.

In vehicles too, it is necessary to have precise and reliable information about the state of health and state of charge of the battery. Battery health is sensitive to temperature, and conventional trucks and buses with diesel engines also frequently fail during winter and autumn. Now, vehicle fleets use solutions for monitoring battery health and the fleet manager does this in a centralized manner.

Analog Devices Inc. presents a solution for monitoring the state of health of primary batteries. The LTC3337 from Analog Devices provides information such as battery cell impedance, voltage, discharge, and temperature. The data from LTC3337 is not only accurate, but the readings are in real-time.

For monitoring the state of health of the battery in real-time, the user must place the LTC3337 in series with the battery terminals. Analog Devices ensure that the series voltage drop is negligibly small when the IC is in series with the battery. Analog Devices has integrated an infinite coulomb counter with a dynamic range to tally all the accumulated battery discharges. LTC3337 stores this information in an internal register which the user can access through an I2C interface. The user can program a discharge alarm with a threshold based on this state of charge. As soon as the state of charge crosses this threshold, the IC generates an interrupt at its IRQ pin. The accuracy of the coulomb counter is constant down to a no-load condition on the battery.

Analog Devices has designed the LTC3337 to be compatible with a wide range of primary batteries with varying voltages. For this, the user can select the peak input current limit of the LTC3337 from 5 mA to 100 mA.

The user can calculate the coulombs from either the BAT IN or BAT OUT pin of the LTC3337—the AVCC pin connection decides this. Some applications require using supercapacitors at the output of the IC. Analog Devices has provided a BAL pin for connecting a stack for supercapacitors for the purpose.

Analog Devices offers LTC3337 as an LFCSP or Lead Frame Chip Scale Package with 12 leads. There is an exposed pad for improving its thermal performance.

The LTC3337 can withstand a voltage range of 5.5 VDC to 8.0 VDC at its input. Its quiescent current is as low as 100 nA. The user can preset the peak input current limits depending on the type of the primary battery. The presents are 5, 10, 15, 20, 25, 50, 75, and 100 mA levels.

LTC3337 is meant for monitoring the state of health of batteries in low-power systems powered by primary batteries. It is very helpful for batteries providing backup and supplies in keep-alive scenarios.

Optical Microphone Watches Sound

A research team from Carnegie Mellon University has claimed to have developed an optical microphone or camera system that can monitor sound vibrations very precisely. The precision is so high that the camera can capture separate audio of individual guitars playing simultaneously. That allows the camera to reconstruct the music faithfully and accurately from a single instrument even when it is playing in an orchestra or band.

Even using the most directed and high-powered microphones it is not possible to totally eliminate neighboring sounds, effects of acoustics, and ambient noise when capturing audio. The research team has used a novel approach. They used two cameras along with a laser beam. This allows them to sense the high-speed but low-amplitude vibrations from the surface of the instrument. The team uses these vibrations for reconstructing sound. The unique arrangement allows them to isolate the audio and capture it without a microphone and with no interference.

The team claims to have invented a new way of seeing sound. The camera system is innovative, represents a new device for imaging, and makes it possible to see things that are not visible ordinarily to the naked eye. The team has successfully completed several demonstrations for showcasing the effectiveness of sensing vibrations and reconstructing sound faithfully and with quality.

During their demonstrations, the team was able to successfully capture isolated audio from separate guitars that were playing together, and the audio of individual speakers that were playing assorted music at the same time. For instance, they analyzed vibrations from a tuning fork. They also captured the vibrations on a bag of burritos placed near a speaker thereby capturing the sound from the speaker.

The team significantly improves the work done earlier for capturing sound by computer vision. Where earlier researchers used high-speed cameras for producing a high-quality recording, the present researchers used ordinary cameras costing only a fraction. The dual-camera is necessary for capturing vibrations from moving objects, such as the movements of the instrument when the musician is playing it, while at the same time, sensing individual sounds from many other points.

The team claims they have improved the optical microphone to make it more usable and practical. They claim to have improved the quality while reducing the expenses.

According to the team, the system operates with two types of shutters—a global shutter and a rolling shutter. An algorithm analyzes the difference in speckle patterns between the two streams of video. It converts the differences into vibrations for the reconstruction of the sound.

A speckle pattern is a result that coherent light or laser generates after its reflection off a rough surface. By aiming a laser beam at the vibrating surface, the team created the speckle pattern. This speckle pattern changes with the changes on the surface as it vibrates. The rolling shutter rapidly scans the speckle pattern from top to bottom and produces the image by stacking rows of pixels one on top of the other. At the same time, a global shutter captures the entire speckle pattern in a single instance.

How Piezoelectric Accelerometers Work

Vibration and shock testing typically require piezoelectric accelerometers. This is because these devices are ideal for measuring high-frequency acceleration signals generated by pyrotechnic shocks, equipment and machinery vibrations, impulse or impact forces, pneumatic or hydraulic perturbations, and so on.

Piezoelectric accelerometers rely on the piezoelectric effect. Generally speaking, when subject to mechanical stress, most piezoelectric materials produce electricity. A similar effect also happens conversely, as applying an electric field to a piezoelectric material can deform it mechanically to a small extent. Details of this phenomenon are quite interesting.

When no mechanical stress is present, the location of the negative and positive charges are such as to balance each other, making the molecules electrically neutral.

The application of a mechanical force deforms the structure and displaces the balance of the positive and negative charges. This leads the molecules to create many small dipoles in the material. The result is the appearance of some fixed charges on the surface of the piezoelectric material. The amount of electrical charges present is proportional to the force applied.

Piezoelectric substances belong to a class of dielectric materials. Being insulating in nature, they are very poor conductors of electricity. However, depositing two metal electrodes on the opposite surfaces of a piezoelectric material makes it possible to produce electricity from the electric field that the piezoelectric effect produces.

However, the electric current that the piezoelectric effect produces from a static force can last only a short period. Such a current flow continues only until free electrons cancel the electric field from the piezoelectric effect.

Removing the external force causes the material to return to its original shape. However, this process now causes a piezoelectric effect in the reverse direction, causing a current flow in the opposite direction.

Most piezoelectric accelerometers constitute a piezoelectric element that mechanically connects a known quantity of mass (proof mass) to the accelerometer body. As the mechanism accelerates due to external forces, the proof mass tends to lag behind due to its inertia. This deforms the piezoelectric element, thereby producing a charge output. The input acceleration produces a proportional amount of charge.

Piezoelectric accelerometers vary in their mechanical designs. Fundamentally, there are three designs, working in the compression mode, shear mode, and flexural mode. The sensor performance depends on the mechanical configuration. It impacts the sensitivity, bandwidth, temperature response of the sensor, and the susceptibility of the sensor to the base strain.

Just as in a MEMS accelerometer, Newton’s second law of motion is also the basis of the piezoelectric accelerometer. This allows modeling the piezoelectric element and the proof mass as a mass-damper-spring arrangement. A second-order differential equation of motion best describes the mass displacement. The mechanical system has a resonance behavior that specifies the upper-frequency limit of the accelerometer.

The amplifier following the sensor defines the lower frequency limit of the piezoelectric accelerometer. Such accelerometers are not capable of true DC response, and hence incapable of performing true static measurements. With a proper design, a piezoelectric accelerometer can respond to frequencies lower than 1 Hz, but cannot produce an output at 0 Hz or true DC.