Advantech, a leading industrial embedded AI solution provider, is pleased to release the RSB-3810. A 2.5” Pico-ITX SBC that adopts MediaTek’s flagship chipset Genio 1200. This solution enables responsive AI inference capabilities up to 4.8 TOPS with 8-watt power efficiency. It also integrates MediaTek’s 5G and Wi-Fi 6 connectivity to empower IoT applications at the edge including robotics, and industrial IoT.
Powerful and Efficient with Multitasking Performance for IoT Applications
MediaTek’s Genio 1200 chipset is at the core of the RSB-3810 and boasts a robust octa-core CPU configuration. It features 4 x premium Arm Cortex-A78 and 4 x Cortex-A55 processors, all integrated into an advanced 6nm-class chip. The result is exceptional power efficiency, with the RSB-3810 consuming a mere 8 watts while effortlessly handling compute-intensive workloads. To enable seamless on-device AI processing for deep learning, Neural Network (NN) acceleration, and machine vision applications, the Genio 1200 chipset incorporates a dedicated dual-core AI processing unit (APU). This specialized unit delivers an impressive 4.8 TOPS of performance, enhancing the RSB-3810’s capabilities in AI-driven tasks. By effectively offloading tasks from the host CPU, this intelligent architecture achieves an optimal balance between system performance and power consumption.
1 x M.2 3052 Key B for 5G, 1 x M.2 2230 Key E Slot for Wi-Fi 6/BT
Rugged design: 0 ~ 60 °C / -40 ~ 85 °C / 3.5Grms
Rear I/O expandable with UIO40-Express I/O boards
Supports Ubuntu, Android, and ROS2 Suite
Seamless Image Processing with Ultra-Low Latency Transmission
The RSB-3810 offers exceptional camera input capabilities via 3 x MIPI-CSI and USB 3.0 ports, featuring an embedded starlight-grade ISP. This enables intelligent vision-based systems to operate effectively even in extremely low-light conditions. With the incorporation of the Mali-G57 chipset, the RSB-3810 supports H265 4K60 video capture and 4K90 image processing to accommodate a wide range of AI applications. Additionally, it facilitates multi-display setups with 1 x 4Kp60 HDMI, and 1 x dual-channel LVDS.
In terms of connectivity, the RSB-3810 provides the necessary I/O interfaces for advanced network and peripheral connections, including 1 x M.2 3052 Key B, and 1 x M.2 2230 Key E Slot. This allows seamless integration with MediaTek’s 5G and Wi-Fi 6/BT networking modules. Furthermore, GbE TSN (Time-Sensitive Networking) is supported as an application protocol for efficient data transmission in monitoring systems and equipment. These features make the RSB-3810 an ideal solution for edge computing applications in industrial IoT and surveillance domains.
“We designed the MediaTek Genio 1200 specifically to process highly demanding AI and performance-centric applications in the industrial IoT space,” said Mr. CK Wang, General Manager of IoT, MediaTek. “In addition to these benefits, the Genio chipset built into the RSB-3810 can also take advantage of the latest multimedia graphic support for 4K and multi-display products to offer even more impressive performance in heavy workloads.”
Flexible OS and Robot-Oriented Software Package Facilitate Integration
The RSB-3810 offers extensive support for various operating systems, including Ubuntu, and Android, catering to diverse development environments for AI and edge computing applications. In partnership with Canonical, the company behind Ubuntu, Advantech enhances enterprise-grade Ubuntu support and provides pre-loaded and certified services. This collaboration ensures efficient over-the-air security updates, enabling developers to focus on their core applications and reduce time-to-market.
As for robotics development, the RSB-3810 is designed to facilitate seamless integration with the ROS2 Suite. This comprehensive software package, built on Advantech’s AIM-Linux embedded software, is specifically tailored to support Robot Operating System (ROS) environments. With the inclusion of MediaTek’s NeuroPilot SDKs, edge intelligence is further enhanced through the utilization of APUs within the processor. The RSB-3810’s high level of software integration significantly reduces the time and resources required for robotics development, streamlining the process for developers.
Advantech’s AI-native 2.5” Pico-ITX RSB-3810 is available now. Please contact an Advantech sales office or authorized channel to find out more. For more information on Advantech’s Arm computing products and services, visit our solution page.
INTRODUCTION – Converting Continuous Signals To Discrete
By definition, an analog signal changes continuously over time and it has an infinite variety of amplitude values as time goes on. The best example of an analog physical signal is our voice. Whatever we speak we generate sound waves and this sound (message) travels through waves. The main characteristics of analog signals are amplitude, frequency, and phase.
If an alternating current (AC) line voltage is viewed on a conventional oscilloscope, the display shows a continuous sine waveform. On that curve, an instantaneous amplitude value might be 100, 99.8 or 99.875 volts. Given different resolutions, there are different real numbers or possible values.
Nowadays, we live in a wide digital world of numbers, texts and symbols, and similar discrete data which would seem to necessitate digital circuits and systems like arithmetic & logic computing and storage systems.
Digital techniques initially began with electro-mechanisms like relays and vacuum-tube logic circuitries and then gradually followed by discrete transistors, small-scale ICs, and finally large-scale and fast microprocessors with some billion transistors. Digital methods and computational applications are currently used for almost all parts of our life; like keeping track of money, and transferring words, sounds, images, and videos.
Utilizing digital signals and systems essentially have some advantages:
Digital signals are very easy to analyze and process
Although digital signals can be easily stored and accessed comfortably.
Digital systems normally produce less noise and consume less power than their analog counterparts.
The Analog-to-Digital converter has numerous applications. For example, it is used in a digital communication system (like a cellular phone) where the analog signal to be transmitted is digitized at the sending end by an A/D converter. It also forms an essential interface when it comes to analyzing analog data with a digital computer. Figure 1 shows the system diagram of the A/D converter.
Figure 1: The A/D converter
Essentially, most of the electronic sensors (sound, image, …) are in the analog form. Therefore, the A/D converter (as a part of digital processing) is used in all digital read-out test and measuring equipment; whether in a digital multimeter or a digital storage oscilloscope or even a pH meter.
We normally expect to utilize the advantages of converting the analog signals to digital form and it is mostly supposed to recover the signals into their original forms (analog) again at the final stage. The main process is referred to as A-to-D conversion or as digitization. The reverse of this process is called D-to-A conversion which means turning the digital information back into analog form. The A/D conversion process is generally more complex than the D/A conversion process.
A/D Conversion Method
The physical nature of an A/D converter is such that it divides a voltage range into discrete values, each of which is then represented by a binary number. There are many different ways in which an analog signal may be turned into a digital form. Several different techniques have evolved for dealing with different system requirements. These include flash, staircase, successive approximation, and delta-sigma forms.
In this article, we will only consider one of the most popular methods, called Pulse Code Modulation, or PCM for short. This concept of A/D conversion is simple: first, the voltage of the incoming waveform is measured at specific instances in time. Then, these measurements will be quantized and finally translated into some digital words. Figure 2 shows the block diagram of a PCM system.
Figure 2: The PCM System
Based on the theory of PCM, to convert an analog signal into a digital form there are three functional steps:
Sampling
Quantization
Digital Encoding
Sampling
The first step involves transforming the signal from continuous-time domain into discrete-time domain. This process is achieved by sampling technique. A sampler systematically extracts some amplitude values in some instants of time, like an electronic switch. It means sampling is done along the x-axis of signals.
This mechanism is shown graphically in Figure 3.
Figure 3: The Ideal sampling system (a) and (b) input/output waveforms
The result of ideal sampling is a series of impulses (or spikes) with amplitude values representing the input signal v(t) amplitude values at different instances. Figure 3 (a) shows that the analog signal is sampled by the electronic switch once every T seconds (for example every 0.5 s), resulting in a sampled amplitude sequence; which is known as a discrete-timesignal. Such discrete-time signals can have any amplitude values (similar to analog signals), but at specific points in time; as shown in Figure3 (b). At other times, the amplitudes are unknown. Briefly, such discrete-time signals are discrete in time, but still continuous in amplitude.
The sampler switch is also assumed to be ideal in which the value of the signal is taken at an instant (an infinitely small time). The time interval between samples is mostly held constant, i.e., the rate of sampling is constant. This condition is called uniform sampling.
The most important parameters in the sampling process are the sampling period T, and the sampling frequency or sampling rate which is defined as fs. The sampling frequency is given in units of ‘samples per second’ or ‘Hertz’.
The Nyquist-Shannon sampling theorem specifies the minimum sampling rate at which a continuous-time signal needs to be uniformly sampled so that the original signal can be completely recovered (or reconstructed) again by these samples alone. This rate is called the Nyquist rate.
If a continuous-time signal contains no frequency components higher than fmax (Hz), as it is represented in Figure 4 (a), then it can be completely determined by uniform samples taken at a rate fs (samples per second) whereas it is explained in Equation 1. To simplify notations, we define fmax = W.
Equation 1: The sampling frequency conditionEquation 2: The sampling period condition
For example, if a signal with a bandwidth up to 20 kHz (like the sound signal) needs to be digitized, then a sample rate of at least 40 kHz is required.
A better understanding of the sampling theorem could be obtained by looking at the sampling process from the frequency domain perspective. Essentially, the sampling process is equivalent to the chopping of the original (analog) signal in the time domain. It can be shown through the Fourier transform theory that this chopping (or switching) function introduces a large number of high-frequency components into the signal spectra.
The high-frequency components generated by sampling appear in a very regular fashion. In fact, every frequency component in the original signal spectrum is periodically repeated over the entire frequency axis. Notice that spectra mathematically extend to negative frequencies also! The period at which this replication occurs is determined by the sampling rate, fs. Replication of the signal spectrum for a bandlimited signal after sampling is shown in Figure 4 (b).
Figure 4: (a) Spectrum of the original analog signal (b) Spectra of the sampled signal
The Nyquist-Shannon sampling theorem states that if the sampling frequency is at least twice the highest frequency of the spectrum (fs = 2W), the replicated spectra do not overlap, and no distortion occurs. Thus, the original spectrum can be faithfully recovered again by suitable frequency filtering.
Consider the effect if the sampling frequency is less than twice the highest frequency component (fs < 2W). As shown in Figure 5, the replicated spectra overlap each other, causing distortion to the original spectrum. Under this circumstance, the original spectrum can never be recovered faithfully. This effect is known as aliasing.
Figure 5: Aliasing distortion in frequency
The aliasing effect in the time domain is shown graphically in Figure 3-1. Normally, in a sequence with a sampling rate greater than the Nyquist rate, if we connect the sample points together by some imaginary lines, it is supposed to make a shape that is similar to the original continuous waveform. This case is shown in Figure 6 (a). But, in the case of fs < 2W the number of sample points is not sufficient to restore the original waveform. Thus, the original waveform has completely disappeared in Figure 6 (b) and this is the real meaning of the alias.
Figure 6: Aliasing distortion in time
In order to prevent the aliasing effect, the input signal must be already frequency-band limited. That is, a low-pass filter (often referred to as an anti-alias filter) must be used before A/D conversion takes place.
The low pass filter (LPF) passes signals with a frequency lower than a selected cutoff frequency (fc) and eliminates the higher frequency components present in the input analog signal to ensure that the input signal to the sampler is free from the unwanted frequency components. The ideal filter has a flat passband and a sharp cutoff behavior.
The block diagram of an ideal LPF is shown in Figure 7 (a) and its transfer function in the frequency domain is depicted in Figure 7 (b).
Figure 7: (a) The block diagram of an ideal LPF (b) The Transfer function
The sampling system incorporating an ideal low-pass filter is shown in Figure 8 (a). Assume that the spectrum of an analog signal has some frequency components more than fs/2, as shown in Figure 8 (b). If the cutoff frequency of this filter is defined as half of the sampling frequency (fc = fs/2), those portions of the spectrum with higher frequencies and small magnitudes will be eliminated by the filter. Thus, the spectrum of the signal after low-pass filtering is shown in Figure 8 (c). Mostly there is no important information in those omitted frequency components and finally, the signal can be restored again by the remained spectrum.
Figure 8: (a) The sampling system including an ideal LPF (b) The spectrum of the input signal (c) The spectrum of the signal after filtering (d) The spectrum of the filtered signal after sampling
As it is depicted in Figure8 (d), the resulting replicated spectra of the discretely sampled signal [Xd(f)] do not overlap each other. Thus, no aliasing occurs.
Quantization
The quantization stage (Figure 9) rounds off the value of samples to the nearest discrete amplitude values in a set of Lquantumlevels. This process is lossy (not reversible) since it is not possible to determine again the exact value of the original fractional number from the rounded integer.
Figure 9: The L-level quantization stage
As well as the sampling process is done along the x-axis (time), and the quantization process is done along the y-axis (amplitude). The spacing between quantization levels is called the quantization width or the quantizer step size which is denoted by q. If the spacing between these levels is the same throughout the range of amplitudes, then it is called a uniform quantizer. A linear quantizer performs a uniform mapping between input and output values.
The following operations are performed in the Quantization stage:
The system assumes that input values, v(t), have amplitudes between Vmin and V Then, for inputs after sampling we have the variable v(kT) as it is explained in Equation 3.
Equation 3: The range of amplitudes
The system divides the range of amplitudes into L quantization levels, each of size q as described in Equation 4.
Equation 4: The value of quantization width
The number of quantum levels for a binary PCM system must equal some power of 2, as it explained in Equation 5 (We will see the reason in the next step).
Equation 5: The number of quantum levels
A simple example of quantization is the mechanism of rounding a fractional number to the nearest integer as it is described in Equation 6.
Equation 6: The quantization as rounding a fractional number to the nearest integer
Then, the quantized results come out from Equation 7.
Equation 7: Calculation of quantum levels
Figure 10 shows the characteristic function of the typical uniform quantizer relating to our example waveform in Figure 3.
Figure 10: the characteristic function of a uniform quantizer
In this graph, some new levels in equal distances with integer values are defined as quantized levels for output results. In this case, because of 8 levels of quantization between 0 and 1 volts, the quantity q equals to (1/8), equivalent to 0.125 volts of the amplitude.
If we apply Equation6 on the samples of v(kT) in Figure 3, we can collect the calculated data of quantization in Table 1.
Table 1: Quantization data of our example
The output results of Table 1 are plotted in Figure 11.
Figure 11: The plot of 3 waveforms (continuous, discrete, and quantized)
Figure 11 shows how the amplitude of the input samples [v(KT)] is mapped to the new quantization levels of amplitude at the output. Also, it is clear that the first continuous waveform v(t) is replaced by the new discrete waveform [vq(kT)].
It is important that the shape of the quantized-samples function vq(kT) (the blue curve) could be a discrete approximation for the original continuous waveform v(t) (the red one), but it also could be better by increasing the number of quantization levels to 16 or 32 and so on.
Summary
Essentially, analog signals are continuous in both time and amplitude. Analog signals, during the time interval within which they are measured, can have any amplitude (including zero) between the minimum and maximum limits, for every instant of time.
The amplitudes of discrete-time signals are determined only at certain time instants.
Digital signals are non-continuous. They are discrete both in time and amplitude. They jump from one allowed value to another as time moves on. There are a limited number of values.
In practice, computers are best able to work with what is known as digital signals.
The most common technique to change an analog signal to digital data is called pulse code modulation (PCM). A PCM encoder has the following three processes: Sampling, Quantization, and Encoding
The result of ideal sampling is a series of sampled values between the maximum and minimum of the input signal.
The number of measurements (amplitude samples) taken per second is known as the sampling frequency. The terms sample rate and sample frequency are often used interchangeably and are usually denoted by fs.
According to the Nyquist theorem, the sampling rate must be at least 2 times the highest frequency contained in the signal. It is also known as the minimum sampling rate and is given by: fs =2 fmax
If the sampling frequency is smaller than twice the highest frequency of the spectrum (fs < 2fmax) a unique form of distortion called alias distortion, is produced.
Anti-aliasing filters are always analog low-pass filters as they process the signal before it is sampled.
In the quantization stage, the system replaces each original sampled value with the nearest quantization level.
In the quantized signal, only a finite number of amplitudes are measured at distinct points in time.
The number of values, or voltage levels (L), in a digital system, is determined by the number of bits (n) in each binary number: Number of voltage levels = L = 2n
MYIR has launched a new ARM SoM MYC-YF13X CPU Module based on STM32MP135, the latest processor from ST, equipped with Arm Cortex-A7 core running up to 1GHz. The processor also features a dedicated LCD-TFT parallel display interface, a 16-bit parallel camera and dual Ethernet ports and is particularly suitable for applications such as entry-level industrial human-machine interfaces (HMI) and embedded devices for energy and power management. In addition to the STM32MP135 processor, the MYC-YF13X module has integrated DDR3L, and external memory and carries out a variety of peripheral and IO signals through the 1.0 mm pitch 148-pin stamp-hole (Castellated-Hole) expansion interface.
The MYC-YF13X CPU Module is a compact SoM form factor measuring only 39mm by 37mm. It is capable of running Linux and provided with software resources including kernel and driver source code, together with detailed user manual and documentations to help customer start their development rapidly.
MYC-YF13X CPU Module
Features of MYC-YF13X CPU Module
Dimensions: 39mm x 37mm
PCB Layers: 10-layer design
Power supply: 5V/1A
Working temperature: -40~85 Celsius (industrial grade)
Up to 1GHz STMicroelectronics STM32MP135 ARM Cortex-A7 processor (STM32MP135DAF7)
256/512MB DDR3L
256MB Nand FLASH/4GB eMMC
32Kbit EEPROM
0mm pitch 148-pin stamp hole expansion interface
– 2 x RGMII
– 2 x USB2.0
– 8 x UART
– 2 x SCI
– 2 x CAN FD
– 4 x I2S
– 5 x I2C
– 2 x ADC
– 1 x RGB
– 1 x Parallel Camera
– 2 x SAI
– Up to 108 GPIOs
Note: the peripheral signals brought out to the expansion interface are listed in maximum number. Some signals are reused. Please refer to the processor datasheet and CPU Module pin-out description file.
Linux OS
MYC-YF13X CPU Module Function Block Diagram
The MYD-YF13X Development Board is a starter kit for evaluating the MYC-YF13X CPU Module. It has a versatile base board to facilitate the expansion from the MYC-YT113X through the 1.0 mm pitch 148-pin stamp-hole (Castellated-Hole) interface, a rich set of peripherals and interfaces have been brought out such as RS232, RS485, two USB 2.0 HOST and one USB 2.0 OTG, two Gigabit Ethernet, CAN, one Micro SD card slot, one USB based Mini-PCIe 4G Module interface with one SIM card holder, LCD interface, Camera interface, Audio input and output as well as two extension headers.
MYD-YF13X Development Board Function Block Diagram
The MYC-YT113X CPU Module is provided with two standard configurations for Nand Flash or eMMC version options at cost-effective prices. The Nand Flash version (MYC-YF135-256N256D-100-I) is pricing at only $19/pc. Discount is to be offered for volume quantities. MYIR also provides OEM/ODM services to help customers accelerate their time to market and save costs.
Axiomtek – a world-renowned leader relentlessly devoted to the research, development, and manufacture of a series of innovative and reliable industrial computer products of high efficiency – is pleased to introduce the AIE110-XNX, its new edge AI developer kit powered by the NVIDIA® Jetson Xavier™ NX platform which features 6-core NVIDIA Carmel ARM® v8.2 64-bit CPU delivering up to 21 TOPS of accelerated AI computing performance and integrates an advanced 384-core NVIDIA Volta™ GPU with 48 Tensor Cores. Aimed to save effort on system integration and reduce complexity for embedded developers, the AIE110-XNX enables users to develop AI solutions on a friendly budget with no minimum order limit. The palm-sized edge AI developer kit is ideal for makers, learners, and system integrators to build more AI innovation with little effort.
“The AI boom is happening all over the world, and it is now available for everyone. Catching the AI wave with faster responding and more efficiency, Axiomtek’s AIE110-XNX can be quickly engaged to be the AIE100-903-FL-NX for smart city, the IP67-rated AIE800-904-FL for outdoor edge AI applications, the AIE900-XNX for robotics and AMR or self-service kiosk applications. It can also be flexibly transitioned to a new generation platform with upgraded AI performance such as NVIDIA® Jetson Orin™ NX, NVIDIA® Jetson Orin™ Nano, and more,” said Annie Fu, a Product Manager of the AIoT Team at Axiomtek. “In addition, the AIE110-XNX has reserved mounting holes on the bracket, which helps system integrators to easily start development right out of the box.”
The edge AI developer kit AIE110-XNX has a rugged design for harsh environments, allowing it to operate under a wide temperature range from -10°C to +60°C and vibration of up to 3 Grms. It has 8GB of LPDDR4x memory and 16GB eMMC onboard. One M.2 Key M 2280 SSD slot with a high-speed PCIe x4 NVMe interface and one Micro SD slot are available for extensive storage needs. One 15W GbE PoE port is available for both power and video transmission to easily deploy an IP camera. It also supports one PCI Express Mini Card slot for Wi-Fi, Bluetooth, LTE, and GPS. The AIE110-XNX supports the NVIDIA JetPack, which provides a full development environment for end-to-end accelerated development on Nvidia Jetson modules, to foster the development of AI-assisted operations across industries.
1 PCI Express Mini Card for Wi-Fi/Bluetooth/LTE/GPS
JetPack supported
Supports 24/7 secure remote monitoring, control, and OTA deployment empowered by Allxon
The AIE110-XNX supports device monitoring and management services by Allxon, bringing comprehensive remote management onto edge AI devices. Axiomtek’s edge AI developer kit AIE110-XNX is now available for purchase. For more product information or pricing, please visit Axiomtek’s global website at www.axiomtek.com or contact our sales representatives at info@axiomtek.com.tw
New Super Compact AONDenoise™ Solution Enables Customizable Noise Cancellation in Ultra-Low Power, Real-Time Processing Devices
AONDevices, a provider of super low-power, high-performance on-device AI processors and full stack solutions, today announced the launch of its groundbreaking AONDenoise™ – a compact, efficient edge AI denoising technology. Designed for applications that require minimal power and latency, AONDenoise is one of the smallest denoising algorithms available, making it an ideal AI speech enhancement for use in hearing aids, wireless earbuds, smartphones, and wearables.
Traditionally, AI speech enhancement is performed using arrays of multiple microphones, beamforming and large Digital Signal Processing (DSP) algorithms. This method, currently implemented in today’s smartphones and other devices, requires huge amounts of memory and power, which is costly. Through its state-of-the-art, AI-based denoiser, AONDevices is addressing these concerns – and leading the way forward into a new era of edge AI.
Known for its innovation in super low-power edge AI human-machine interfaces, AONDevices is expanding its focus to include innovations in human-human interfaces. With just 50K parameters, the AONDenoise solution is one of the most compact denoising algorithms on the market. It is multiple orders of magnitude smaller than traditional DSP noise reduction methods and at least an order of magnitude smaller than current AI-based denoisers. The AONDenoise solution addresses the need for customizable noise cancellation in ultra-low power, real-time processing devices, such as chips used for hearing aids. The algorithm can target a wide range of noises, such as chatter in a crowded restaurant, wind noise, dogs barking, babies crying, or keyboards clicking.
The AONDenoise solution has the potential to revolutionize various applications, including:
Hearing Aids: Customizable noise cancellation for improved performance in different environments.
Wireless Earbuds: Optimal audio quality in transparency mode, conversation mode, and voice calls.
Smartphones: Clearer voice quality during calls, especially in outdoor settings.
Podcasting: Crystal-clear audio streaming even in noisy surroundings.
Leveraging AI and deep learning neural networks, the AONDenoise solution can distinguish human speech from background noise. To maximize the efficiency of this neural network, AONDevices has developed a dedicated software tool suite. This allows both in-house and customer development to customize and train the system to handle specific noise environments or a broader range of noises.
AONDevices’ new Technical Director, Dr. Youhong Lu, commented that AONDenoise is poised to significantly enhance the next generation of wearable and mobile devices.
“AONDenoise is a truly remarkable accomplishment in AI speech enhancement,” said Dr. Lu. “A small yet incredibly effective algorithm that works with just a single microphone will positively impact many lives. I have been incredibly impressed by AONDevices’ achievements and am excited for the upcoming technology revolution in which AI delivers better user experiences.” Dr. Lu recently joined the company after having previously served in high-level audio engineering roles for industry leaders including Conexant, Goodix, Plantronics, Verizon, and Microsoft.
The AONDenoise solution is now available for integration into third-party processors. For customers seeking dedicated hardware solutions, AONDevices is preparing to launch the denoiser hardware IP and the associated next-generation chip – the AON2100™ – in the coming months. For more information, contact info@aondevices.com
iWave adds to the iW-RainboW-G50M family: The Solderable NXP i.MX 91-based LGA System on Module (SoM). The SoM incorporates the newest member of the i.MX 9 applications processor series and is built on the OSM v1.1 solderable SoM standard, providing extensive interfaces in a rugged and compact form factor.
The Rainbow-G50M family provides for hardware and software commonality and ensures scalability based on the requirement and application. Evaluation Kits of the System on Module will be ready to purchase in June 2023.
The i.MX 91 SoC delivers an optimized blend of security, performance and scalability, and is a perfect foundation block for cost-optimized Linux® edge devices. The i.MX 91 family features an Arm® Cortex®-A55 running at up to 1.4GHz and an integrated EdgeLock® Secure Enclave, together with interfaces such as dual Gigabit Ethernet, dual USB ports, essential I/Os to fit in products used in smart factory, smart home, medical devices, and smart metering.
Key features of OSM System on Module
NXP i.MX 91 SoC
2GB LPDDR4 RAM
16GB eMMC Flash
2 x 1Gbps Ethernet Controllers (1 with TSN)
1 x SD(4bit)
1 x USB 2.0 OTG, 4 x USB 4.0 Host
Wi-Fi 6 & Bluetooth 5.3 Connectivity
15.4 Connectivity
Size-L Form Factor: 45mm x 45mm
Solderable LGA Package in OSM v1.1 Standard
10+ Years Product Longevity Program
The System on Module is built on a 45mm x 45mm OSM Size-L standard with the provision for 662 contacts, offering the highest pin-to-area ratio across SoM standards. With the ability to directly solder the SoM onto the carrier card, the SoM ensures high levels of robustness and is ideal for products prone to vibrations.
“The i.MX 91 applications processor family from NXP provides an ideal mix of high performance, pricing, and security; fit for entry-level Linux edge processing solutions for applications such as EV Charging Stations, Industrial Gateways, and HMI Displays,” said Immanuel Rathinam, Vice-President of System on Modules Unit at iWave.” The i.MX 9x family of i.MX 91 and i.MX 93 provides customers scalability with reduced development cost and time to market and provides consumers with efficient power management and advanced security features on the edge
The i.MX 91-based System on Module is integrated on a carrier board, which is positioned as a Single Board Computer and also doubles up as an evaluation kit. The production-ready Single Board Computer is built on a Pico-ITX form factor and integrates the necessary interfaces for applications such as smart home hubs, metering gateways, and industrial automation solutions.
Key features of the Single Board Computer
NXP i.MX 91 SoC
2GB LPDDR4 RAM, 16GB eMMC
Wi-Fi 6 & Bluetooth 5.3 Connectivity
15.4 Connectivity
GNSS receiver Module – GPS/GLONASS/Galileo/BeiDou
Dual 1000/100/10 Mbps Ethernet
USB 2.0 OTG (microAB Receptacle Connector) x 1
Dual USB 2.0 Host (TypeA) x 1
USB Header x 1 & USB Type-C Connector x 1
Micro SD
5mm Audio In & Out Jack through I2S Codec
2 Connector Key B
Expansion Connector – UART, CAN, ADC, Tamper, PWM
RS232 Header x 1
The System on Module and Single Board Computer are go-to-market and production ready, with all documentation, necessary software drivers, and BSP available for customers. iWave maintains a product longevity program and ensures the availability of the modules for 10+ years.
Atezr, an innovative and trailblazing brand within the laser engraving industry, recently unveiled a new series of cutting-edge laser engravers called L2, representing the second generation of their product line. The L2 models are available in three variants: 10W, 20W, and 35W. Notably, these engravers possess numerous technical advancements compared to their predecessors.
One of the standout improvements is a substantial increase in speed, with a maximum engraving speed of 54000 mm/min. This technology compresses the laser spot size from 0.15mm x 0.15mm to 0.06mm x 0.1mm, allowing effortless cutting of 20mm Pine Wood and 15mm Black Acrylic in a single pass. This speed is more than three times faster than the previous generation, resulting in a 40-60% reduction in project completion times. Additionally, each L2 model features a resume function that detects power loss and enables the machine to seamlessly resume its work from where it left off once power is restored. With these remarkable features and more, Atezr’s second-generation laser engravers are the ideal choice for individuals seeking to enhance their productivity and efficiency.
This cutting-edge engraver features a high-quality, all-aluminum frame that guarantees stability and long-lasting performance. With an expanded working area of 430mm x 430mm, you have the freedom to explore greater versatility in your projects. The innovative optical axis transmission structure of our engraver incorporates an integrated lead screw, replacing the previously combined coupling mechanism. This enhancement not only enhances accuracy but also boosts durability. By integrating the stepping motor and lead screw, seamless and precise movement of the laser, resulting in flawless cutting and engraving is guaranteed.
The LED display provides precise information on the laser module’s power consumption, facilitating easy monitoring during operation. Moreover, second-generation Atezr machines are equipped with a 4.3-inch touchscreen that can be moved, enabling users to conveniently manage various settings, including activating the flame detector and adjusting the auto-focusing, all in one centralized location.
In addition, numerous functional features have been incorporated to enhance performance. The inclusion of an air-assisted automatic door opening system not only ensures smarter heat dissipation but also reduces noise by more than three times compared to previous models. The machines’ double-layer air filtration cover enables improved dispersal of dust, preventing the accumulation of debris in the workspace. Furthermore, the laser preview function accurately positions the engraving material for precise results. A notable feature of second-generation Atezr machines is their automatic focal length adjustment, achieved by raising or lowering the laser module within a range of 10-40 mm based on the detected material thickness. This feature prevents blurry engravings and enables an increased cutting capacity of 4 mm. Compared to other engraving machines with similar power, this represents an improvement of over 30%.
The second-generation Atezr laser engravers are equipped with safety measures that include a gyroscope with slope protection. This feature automatically halts the machine’s operation when the angle between the X- and Y-planes and the horizontal Z-plane exceeds 15°-20°, ensuring user safety. Furthermore, these engravers are equipped with a flame detector that actively monitors the working area. By providing real-time monitoring, the flame detector effectively prevents sparks from igniting into flames and alerts the user promptly. For more information about Atezr laser engravers, visit the announcement page.
InfirayXinfrared introduced the compact P2 Thermal camera dongle, which received a positive response from the market, prompting Infiray to introduce the P2 PRO model. This upgraded version features a detachable close-up lens designed for macro imaging tasks. When the close-up lens is attached, the P2 camera’s capabilities are enhanced, allowing for effective analysis of modern miniature SMT components on printed circuit boards. With a resolution of 256x192px is able to see in detail the thermal profile of the object under test.
Features
Infiray Xinfrared’s P2 Pro represents a remarkable achievement in shrinking thermal imaging technology to cater to users seeking to equip their mobile phones with thermal imaging capabilities. By utilizing a mobile phone as the host device, the camera design not only reduces the overall Bill of Materials (BoM) but also takes advantage of the substantial processing power and high-quality displays commonly found in modern mobile phones. The accuracy of the thermal camera is ±2℃ which is more than enough for most applications. The refresh rate is 25Hz.
Build Quality and Connection
The P2 PRO is encased in aluminum, delivering a sense of premium quality to the user. At the top of the camera’s case, a USB Type C male connector is firmly secured. The camera itself lacks buttons or indicators, with the standout feature being the sizable LWIR transparent window located on the front. This window not only safeguards the camera lens situated behind it but also provides protection against dust and liquids. While the camera is not officially waterproof, the presence of a sealed window in place of an exposed imaging core lens is highly advantageous.
Macro Lens
The P2 PRO model offers users an enhanced temperature measurement capability of -20 to +600C, along with the flexibility to attach a meticulously designed close-up (Macro) lens for intricate object examination. The close-up lens assembly is thoughtfully shaped to align with the front casing of the P2 PRO and securely held in place using a magnet. This clever design proves to be highly effective, allowing for effortless attachment and removal of the close-up lens. The focus distance of the close-up lens is 25mm. For optimal performance in long-wavelength infrared (LWIR) applications, the close-up lens is crafted from Chalcogenide IR glass with an anti-reflective (AR) coating.
Compatible with USB 2.0
The P2 PRO incorporates the latest Infiray Xinfrared Tiny1 series thermal imaging core, providing integrators with multiple interface options for the host device, such as USB. Within the core, you’ll find the lens, microbolometer, housekeeping electronics, and USB interface. The host system takes on the more resource-intensive tasks of processing, measuring, and displaying. The camera connects to the host mobile phone using a USB-C connector, while compatibility with USB 2.0 is achievable by using appropriate OTG connector adapters.
The image collection, processing, and analysis software are installed on a typical Android mobile phone. The P2 PRO software delivers the necessary imagery and image analysis functions to the user, eliminating the need for a larger computing device like a PC. It’s worth noting that for phones lacking a USB-Type C interface, USB adapters will be required.
The P2 Pro camera package is meticulously designed, resembling the packaging of a mobile phone. Inside the box, a removable card envelope holds documentation and a small Infiray leatherette bag for the camera’s protection during travel. Adjacent to the camera, there is a card box containing a USB Type C extension cable, which is a nice addition to a camera dongle meant to be directly mounted on a mobile phone. Removing the camera from its foam insert is made easy with a ribbon pull, adding a thoughtful touch by Infiray Xinfrared.
Specifications
Imaging Core: Infiray Tiny1
Resolution: 256 x 192 Pixels
Pixel Size: 12um
Frame Rate: 25fps
Field of view: 56.0°x 42.2°
NETD: <50mK @25C with F1.0 Lens
Focus Distance (Fixed lens): ~200mm to Infinity
Focus Distance (with close-up lens): 25mm
Measurement Temperature range: -10C to +550C in two ranges
Measurement accuracy: +/- 2C or +/- 2% whichever is greater
Span and Level adjustment: Automatic (Brightness and Contrast controls are included)
Interface to Host: USB C with USB 2.0 data compatibility
Operating Systems Supported: Android only
SDK availability: A Tiny1 series core SDK is available
The smaller Infiray Xinfrared box, containing the “P2 PRO Micro Teleconverter,” follows a similar design. The box features gold embossed writing and a velvet-covered foam insert that holds the Close-up lens assembly. Just like the camera, a ribbon lifting system is employed for effortless extraction. However, this box lacks documentation, which would have been helpful for understanding the specifications of the close-up lens.
Software
The P2 Pro camera is designed as a mobile phone dongle, specifically for use with an Android mobile phone as the host device. Therefore, a compatible mobile phone is required to utilize the camera dongle. However, it might be possible to use the P2 Pro with a PC if suitable host software is obtained, although this aspect is not covered in this review.
To establish communication between the camera dongle and the mobile phone, the mobile phone must have the Infiray Host system app called “P2 PRO” installed, which can be downloaded from the Google Play Store. Once installed, running the app prompts an animation instructing the user to insert the P2 Pro camera into the phone’s USB port. The app then requests the necessary access permissions.
Sample Photos/Video
After the initial setup, a thermal image should appear on the mobile phone’s display, indicating that the camera and phone are functioning together. The camera receives power and commands from the host phone, while the P2 PRO app receives the USB data stream required to generate the thermal image on the phone’s display.
The P2 Pro camera utilizes a Flat Field Correction (FFC) flag to maintain image quality and perform self-calibration for temperature measurements. Periodic clicking sounds can be heard from the camera dongle as the microbolometer warms up, followed by occasional clicks every few minutes.
Once the setup is complete, the P2 Pro camera and its accompanying software app are ready for use. Attaching the close-up lens (Macro lens) is a simple process of aligning it correctly with the shape of the P2 Pro camera and allowing the magnetic retention system to securely hold the close-up lens assembly in place on the camera’s front face. Removing the close-up lens, it is a matter of pulling it away from the camera body.
Video
Conclusion
The Infiray Xinfrared P2 PRO thermal imaging camera dongle is a meticulously designed solution suitable for various applications, including general thermal imaging and specialized tasks such as electronics, research and development (R&D), automotive, and repairs that involve temperature measurement or imaging. The camera’s build quality is exceptional, featuring a robust cast aluminum casing, a reliable USB C connector, and an integrated large lens protection window.
The ZimaBoard positions itself as a Hackable Single Board Server, which seems to be the first of its kind. It’s a highly customizable board to a significant degree as we will see below. Think of the ZimaBoard as a malleable material like clay, that can be shaped and customized according to your specific needs and preferences. Moreover, it boasts an abundance of ports, providing ample connectivity options. The product page shows that it’s suitable for Personal NAS, VPN, Streaming, Software Router, Media server, and Smart Homes.
Features
The ZimaBoard is marketed as a Hackable Media Server, which is actually one of its features in this case. It comes pre-installed with CasaOS, which has its pros and cons. On the positive side, it offers a sleek and modern UI. CasaOS includes various pre-installed apps and features, such as a file manager, app store, automation tools for IoT devices, and an intuitive settings panel. The ZimaBoard allows for extensive customization and boasts numerous ports for connectivity. It is suitable for a range of purposes, and its compact size adds to its appeal.
Additionally, you are able to install additional dockers if you cannot find what you’re looking for on the built-in App Store. However, it is worth noting that CasaOS, being Debian-based, may not offer the same level of customization as Windows or other Linux distros. But here’s the good news: since the board is based on x86 architecture, it is compatible with most modern operating systems. This means you can even install Android, Libreelec, or OpenWrt, in addition to the more obvious choices of Linux and Windows.
PCI Expansion Port
With its 2 x SATA ports, the ZimaBoard offers virtually limitless storage capacity. However, whether or not the two ports are sufficient is a matter of personal preference. Some may argue that they require at least four ports. Although the ZimaBoard does not include pre-installed graphics or network cards, it has the capability to support them once they are installed. To enable this, the ZimaBoard provides a single PCIe 2.0 port expansion port. With proper installation, you can easily add any x 4 compatible board to expand the capabilities of the ZimaBoard. This is what enables the expansion mentioned above. This offers you the freedom to use the board as you wish, and there are no limitations to what it can achieve.
Compatible OS: Linux / Windows / OpenWrt / pfSense / Andorid / Libreelec
Connectivity
Whether you need to set up your VPN, access files remotely, or simply stream media, having Ethernet ports is essential. It offers 2 x Ethernet ports for network connectivity which may be more than enough for the average user. Additionally, it offers 1 x Mini-DisplayPort 1.2 4K @60Hz, which is another aspect that contributes to the “hackability” of the ZimaBoard. Instead of simply including a DVI, Thunderbolt, or HDMI port, they offer you the choice, emphasizing the DIY nature of the board, right? With a maximum compatibility of 4K @60Hz, the ZimaBoard is perfect for high-resolution media display. However, I wouldn’t recommend using it for gaming since you would typically prefer a refresh rate higher than 60Hz.
Video
Flavors
The ZimaBoard proudly showcases its compact dimensions. To be precise, it measures a mere 138.7mm in width, 81.4mm in depth, and 34.9mm in height. In comparison to a typical NAS server, it is truly miniature. The ZimaBoard is equipped with the Intel Celeron N3450 Quad Core 1.1-2.2GHz CPU, which is also found in its 832 models.
The CPU can be boosted up to 2.2GHz, which is more than sufficient for your home media server or cloud storage needs. The model number “832,” features 8GB of RAM. The 432 model features 4GB of RAM, while the 216 model is equipped with 2GB of RAM. Please note that the RAM is not upgradable. In terms of storage, the “832” model provides you with 32GB of storage. Additionally, it features Intel HD Graphics 500 with a frequency range of 200 MHz to 700 MHz.
Conclusion
The ZimaBoard 832 is available at a price of $199.90, which one can consider quite affordable when compared to other SBS with similar specifications. However, the pricing varies for the 216 or 832 models because of the difference in RAM size. In a nutshell, the ZimaBoard people may have varying opinions about the ZimaBoard. However, one thing is clear, you can use the ZimaBoard as your personal NAS storage, VPN, media server, or router. For more information, visit the product page.
ICOP, a manufacturer of embedded devices, has recently launched the ICOP NX8MM-35, a powerful single-board computer (SBC) designed for various industrial applications. The SBC integrates the NXP i.MX 8M Mini, is a competent and efficient internal module that ensures high performance and power efficiency.
The NXP i.MX 8M Mini is an embedded multicore applications processor specifically designed for industrial use cases. This module combines a quad-core ARM Cortex-A53 CPU with a Cortex-M4 microcontroller, enabling multitasking and efficient power management. With its advanced graphics capabilities, the i.MX 8M Mini supports rich multimedia experiences, making it ideal for applications such as industrial control and home automation.
The NXP i.MX 8M Mini module is part of the company’s EdgeVerse edge computing platform designed with a commercial use case. The EdgeVerse platform primarily offers edge computing capabilities through security and software libraries, including machine learning solutions. Some of the other EdgeVerse processors are Layerscape and S32 automotive processing platform.
Specifications of ICOP NX8MM-35 SBC:
CPU: NXP i.MX 8M Mini featuring up to 4x Cortex-A53 cores clocked at 1.6GHz and 1x Cortex-M4 core
GPU: GC NanoUltra 3D supporting GC320 2D OpenGL ES 2.0
Memory: 1/2/4 GB LPDDR4 RAM
Storage: 1x eMMC supporting up to 64 GB of storage and 1x microSD card slot
Connectivity: Wi-Fi IEEE 802.11b/g/n and Bluetooth 4.1
Interfaces: 3x USB 2.0, HDMI 2.0, MIPI DSI (Optional)
Power: DC between 5 to 30V
Software support: Yocto Linux and Android
Dimensions: 146×102 mm
The ICOP NX8MM-35 comes equipped with Gigabit Ethernet ports, enabling fast and reliable network communication. This makes it suitable for applications that require high-speed data transfer or networking capabilities. The SBC supports up to 4GB of LPDDR4 RAM, ensuring smooth performance even when running resource-intensive applications. Furthermore, it offers multiple storage options, including eMMC and MicroSD, providing flexibility and enough storage space for various applications.
In terms of software support, the ICOP NX8MM-35 is compatible with various operating systems, including Yocto Linux and Android, allowing developers to choose the platform that best suits their needs. The SBC also offers extensive I/O interfaces, including HDMI, USB 2.0, and GPIOs, facilitating easy integration with other devices and peripherals.
For more information on the ICOP NX8MM-35 single-board computer, head to the official Wiki page.
login
