“For security purposes, video doorbell technology is widely used in residential, commercial and industrial facilities to replace traditional high-priced CCTV networks without the need for transmission media such as coaxial cable or Ethernet. This article examines some designs related to video doorbells Problems, and around the video, audio and power technology, to give relevant embedded developers a reference.
For security purposes, video doorbell technology is widely used in residential, commercial and industrial facilities to replace traditional high-priced CCTV networks without the need for transmission media such as coaxial cable or Ethernet. This article examines some designs related to video doorbells Problems, and around the video, audio and power technology, to give relevant embedded developers a reference.
seamless user experience
Traditional video doorbell systems involve keys, microphones, and cameras. These systems are usually hardwired to power and the video is routed to specific monitors. Video doorbells for IoT have a similar architecture but are implemented very differently. Motion sensors detect visitors walking to the door and stream video via the cloud to a smartphone or cloud storage. Communication with visitors takes place via bidirectional IP audio streaming and bidirectional video streaming running in the application. The basic functionality of these doorbells can be integrated with a complete security system that can remotely enable/disable keyless locks, trigger alarms or provide automatic feedback based on specific inputs.
Early versions of video doorbells were often plagued by video and audio issues, such as false ringtones and choppy audio, but key features like cloud backup, motion detection, video streaming, and two-way communication required seamless connectivity to be commercially viable. value. These requirements, combined with previous hard-wired power limitations, present modern video doorbell subsystems with their own set of hardware challenges.
error action event
Thermoelectric (aka passive infrared) sensors commonly used in video doorbells are prone to errors, such as incorrectly responding to glare from vehicles driving during the day, gusts of warm wind, bugs, animals, and a wide variety of other objects. Heat-based activity and in the process trigger annoying false alarm tones and notifications on the user’s phone. This greatly reduces the security of a video doorbell as users will eventually ignore the alarm entirely, or even take the doorbell offline. In addition, the frequent occurrence of false motion detection events by PIR sensors can greatly reduce battery life.
A relatively simple solution is to use two PIR sensors with slightly overlapping coverage to create a larger motion detection area (Figure 1). Since the dual sensors only generate notifications for larger objects, smaller objects such as bed bugs and pets will go undetected. Use PIR sensors with other light sensors and temperature/humidity sensors to avoid false triggering due to rapid changes in temperature or light. This multi-modal sensing approach reduces the possibility of false alarms, while also reducing overall power consumption, thereby extending battery life.
Figure 1 Redundant PIR sensors improve the accuracy of human motion detection,
Because multiple beams must be fired for it to be considered a motion event.
Algorithm-based motion detection can also be implemented using an embedded MCU and some firmware to improve accuracy. There are various ways to achieve visual motion based detection, but one of the most common is to compare the current frame to a reference image and track the differences pixel by pixel. This type of image processing must be smart enough to process motion from windmills and trees as part of the background to avoid false positives, a capability that requires considerable processing power.
Some of these filtering tasks can be transferred to cloud-based algorithms that fine-tune image data for specific customers. But this would require relatively large infrastructure to provide support and a good Wi-Fi connection, and would still be high power consumption. So a battery-powered smart doorbell isn’t a smart choice — at least for now. While being able to rely on external power reduces the doorbell’s location options, it also doesn’t require charging or battery replacement.
Image sensor and processor interface issues
Image processing in a video doorbell requires an image sensor, a digital multimedia processor, and in most cases, some peripherals. When choosing an image sensor, there are several things to consider, the most important of which are resolution, frame rate, pixel size, pixel structure, and shutter time. In addition to the many considerations of individual components, there are often interface issues between image sensors and digital media processors.
Unless you pay special attention, you may find that your devices cannot communicate with each other because of mismatched input/output (I/O) interface formats. Because of the large number of differences in the I/O interfaces (I2C, parallel, general purpose I/O), errors like this are much more common than one might think. To avoid this unpleasant situation, designers must ensure that the I/O interface supported by the image sensor is compatible with the I/O of the digital media processor.
Similar problems can arise when two devices have different operating voltages and logic signal levels. Fortunately, voltage converters can easily resolve this mismatch with bidirectional voltage conversion in the range of 0.6 to 5.5 V, and although they add little cost to the product’s BOM, voltage conversion devices provide designers with A wider range of sensors to compensate for this investment, in the past the sensor and MCU had to use the same matching voltage.
environment prone to noise
The full-duplex communication required for modern video doorbells adds additional complexity, requiring the design to deal with erratic feedback caused by the user adjusting the speaker/microphone gain too high. For example, the person receiving the audio needs a relatively large gain on the speaker to adequately discern what the far end is saying, but the close range of the microphone easily detects the sound and often amplifies it, causing bad echoes (Figure 2). In the past, half-duplex communication mitigated this echo by significantly reducing the microphone’s gain when the signal was received by the speaker.
Figure 2 Two-way audio communication requires careful consideration in terms of reverberating speech and echoes
Systems that actively adjust microphone and speaker gain may correct this for full-duplex communication in environments with relatively low ambient noise levels. Unfortunately, this doesn’t work well in environments with unpredictable sources of ambient noise, such as passing buses or other traffic. There are several digital signal processing (DSP) techniques that can address this problem, including Acoustic Echo Cancellation (AEC) and Adaptive Spectral Noise Reduction (ASNR). AEC creates adaptive filters that effectively cancel echoes by initially identifying the transmitted signal and canceling it when it reappears within a certain time window. ASNR utilizes the frequency domain to remove ambient noise and unwanted noise components from the audio signal, thereby removing background noise and broadband noise. AGC is designed to improve the low frequency speech signal of hands-free communication. Audio algorithms such as these provide a superior audio experience by maintaining microphone and speaker gain without unwanted feedback and echoes or resorting to voice switching.
Get the most out of your speakers
While complex DSP algorithms help enable full-duplex audio communication, they often do not maximize the full capabilities of a system’s audio speakers. Because excess heat in a speaker’s voice coil and exceeding its excursion limit can cause rapid damage and cone blowing, audio engineers often impose hard limits on amplification levels that are well below the speaker’s actual capabilities. A software algorithm used in tandem with the amplifier monitors the temperature and excursion of the speakers in real time, and this feedback allows for finer sound pressure levels and higher audio clarity.
Voice Commands and Speech Recognition
Future video doorbells may enable hands-free control based on voice activation and voice recognition technology. Another layer of complexity is added when these voice user interfaces again receive commands from a series of microphones and DSP algorithms. Despite the relatively large distance from the receiving microphone, these doorbells will likely use beamforming algorithms to separate the desired audio signal from background noise. There are already available microphone boards that implement beamforming algorithms that amplify the speech signal from the direction of the speaker for clear speech and audio from noisy environments.
In a truly functional video doorbell product, it is important that these advanced features do not require additional power and can act on the local microphone input signal. We are looking for a design strategy to make the product simpler, low power and small size.
Power Budget Challenge
A practical video doorbell can be powered in one of the following ways: using a rechargeable battery, allowing it to draw power from the house’s existing low-voltage doorbell wiring, or equipping it with a Power-over-Ethernet (PoE) interface. Each of these power options has pros and cons (Table 1). As mentioned earlier, the flexible placement offered by the battery powered unit makes installation easier, while the doorbell cord has the advantage of low maintenance costs.
Power saving is a major concern for battery powered video doorbells, and many of the above algorithms will require more power-intensive processing. Highly specific SoC designs, such as the Texas Instruments (TI) CC3120/CC3220, enable higher levels of parallel processing (wake/sleep triggers, network connectivity) with fewer off-chip transactions (on-chip RAM and/or flash memory), Thereby reducing the total power consumption of the system. Additionally, MCUs designed for battery operation feature multiple power modes, including shutdown, hibernate, sleep, standby and active modes, which can be used by careful developers to further reduce energy consumption.
A major consideration for any product designed to use a home’s existing doorbell power supply is that there is no standard output voltage for these products in AC power supplies, which were originally designed to power bells using voltages between 8 V and 24 V AC. Designed. To minimize product degradation, it is important to pay careful attention to parameters such as output voltage accuracy, voltage ripple, system efficiency at full load, and heat dissipation. This is especially true for particularly sensitive components, such as CMOS image sensors often used in video doorbells. These components are particularly sensitive to noise sources such as power supply fluctuations, electromagnetic interference and temperature changes.
For optimal performance, a video doorbell needs a power supply that can accept a variety of low-voltage converters and produce clean, well-regulated voltages for its various subsystems (sensors, I/O, audio, memory, UI, etc.) Direct current, also has to be miniaturized to fit into a compact enclosure. As shown in Figure 3, this typically involves multiple buck converters, preferably synchronous buck converters that provide high efficiency under heavy loads. In such designs that require a wide voltage range or a large number of discrete power supplies, a single buck regulator can be used to power multiple linear regulators.
Figure 3: Schematic diagram of video doorbell power supply architecture
System efficiency at full and light loads is required for battery powered applications, as well as for products operating in closed packages with little or no ventilation. For video doorbells, features such as user interface, wireless communication monitoring, and motion detection must be carefully implemented to maximize power efficiency. The same attention must be paid to standby currents, such as the quiescent and shutdown currents of the power supply, as they can significantly affect battery life. Low quiescent current can greatly extend battery life as the video doorbell spends most of its time in sleep/hibernate mode. In addition, the synchronous converter has a seamless transition from its PWM mode to power saving mode, allowing it to remain relatively efficient at both full and light loads.
Video doorbells are one of several IoT products that have strict size constraints (and sometimes power constraints) and must balance processor-intensive algorithms with limited power resources. These limitations lead to some unique design challenges that are now made possible by technological advances. Naturally, these challenges will continue to become increasingly complex as artificial intelligence in the form of voice, voice and facial recognition becomes a must-have feature in residential security systems.
By Srinivasan Iyer, Systems Engineer in the Building Automation Division of Texas Instruments (TI), specializing in video surveillance, HVAC, elevators and escalators, among other industries.