With radar getting adopted even in consumer products, Systems Plus Consulting got curious about how it was being implemented. The technology and costing analysis firm decided to tear down a radar chip, but which one?
With radar getting adopted even in consumer products, Systems Plus Consulting got curious about how it was being implemented. The technology and costing analysis firm decided to tear down a radar chip, but which one? The company selected Vayyar’s first-generation RF system-on-chip for several reasons.
The was primarily interested in how Vayyar designed such a highly integrated single-chip RF SoC. System Plus analysts were also intrigued by the ability of Vayyar’s SoC to create high-resolution 4D images.
Radar used to be a slow-growing market catering to mature applications such as the military. Things have changed — a lot. Radar, especially “imaging radar,” is hotter than the Kardashians. Automotive and consumer applications are firing system designers’ imaginations and driving radar growth into double-digit numbers.
While radar remains critical to military bases and aircraft carriers, they are massively invading homes, family cars and even smartphones. For automotive applications, Tier Ones and car OEMs are working on imaging radars for both advanced driver-assistance systems (ADAS) and in-cabin occupancy detection.
Radar technology suppliers are also keeping an eye on smartphone opportunities. Central to the collaboration between Infineon and Google announced last year, for example, was the enabling of gesture control in Google Pixel 4 smartphones, using Infineon’s radar technology. Although it is far from clear if every phone will feature radar technology any time soon, the field of radar apps is rapidly expanding.
Vayyar designed its RF SoC into its own Walabot line of products. The Walabot Home fall-detection system enables family members and caregivers to keep tabs on aged relatives who might be prone to falling. Being radar based, it can perceive people through walls and curtains. One of the benefits is that the person being monitored does not require a wearable device.
Systems Plus Consulting (a Yole Développement group company) was intrigued by the Vayyar chip, which analyzes a multitude of signals sent and received from integrated transceivers and processing them by a high-speed DSP on the SoC. System Plus revealed that Vayyar designed a small board housing both the RF SoC and MCU. That setup allows Vayyar’s RF SoC to work with any external application processor picked by a system vendor.
The first-generation of Vayyar’s chip — the subject of this teardown — was based on a 3-10 GHz RF SoC. Given frequency range restrictions imposed by different countries, Vayyar has already designed follow-ons, including one that works at 57-64 GHz, offering larger maximum bandwidth and a higher range of resolution. Another is based on the 77-81 GHz range.
The RF SoC comes with “an on-chip DSP with a large amount of SRAM memory in the transceiver die,” according to System Plus. The RF SoC provides the data to an MCU placed on the Walabot board. The MCU “only transforms the SRAM data into USB data stream type.” This step is critical since it makes Vayyar’s RF SoC neutral and flexible enough to work with any external CPU or application processor chosen by systems designers. It doesn’t matter if it’s Qualcomm’s Snapdragon or someone else’s app processor. It can then execute complex imaging algorithms if needed.
Beyond Walabot Home system
Beyond the Walabot Home system, Vayyar is making inroads in the automotive market. Vayyar sealed a deal with Valeo, a leading Tier One company, two years ago. At that time, Valeo announced plans to use Vayyar radar sensors for monitoring infants’ breathing and triggering an alert in case of emergency, especially if infant has been left inside a vehicle alone.
Last November, Vayyar raised $109 million in a series D round of funding led by Koch Disruptive Technologies (KDT), an investment subsidiary of U.S. multinational Koch Industries. Nabbing Koch as a strategic investor was a big deal, since Koch and its various subsidiaries can potentially open the door for Vayyar’s imaging sensors into many more market segments.
Earlier this year, Vayyar announced a partnership with Japan-based automotive parts company Aisin Seiki, seeking to jointly develop “4D high-resolution short-range exterior sensors for vehicles” for such applications as blind-spot detection.
Walabot Home Setup
Walabot Home is an intelligent system that keeps track of people’s movement and understands if they have fallen down and need assistance. The small and thin device uses a sensor system that processes low power radio waves similar to Wi-Fi signals to determine a person’s location.
The RF SoC at the heart of the system combines 3GHz-81GHz transmitters and receivers to create high-resolution 4D images by analyzing a multitude of signals sent and received — without an external CPU. Thanks to the integration of transceivers and a high data processing speed DSP, Vayyar technology is able to create accurate situation scenarios.
Through its Walabot Home System, Vayyar technology can display the size, position, and movement of people and objects, allowing a complete real-time classification of the environment without using a camera. The lack of an imaging camera means there is one less avenue for potentially violating privacy.
Walabot HOME is a fall alert system which monitors people, continuously preventing possible dangers and accidents. This smart home device can sense if a person has accidentally fallen, while at home, and needs help.
The system features an UWB (Ultra-Wide Band) band system that detects people and their position in space. UWB transmits and receives signals through the use of radio-frequency energy pulses of extremely short duration (from a few tens of picoseconds to a few nanoseconds). It is, in practice, a wireless protocol that allows reaching a band of the order of gigabits per second with electrical power in the antenna of tenths of watts. (The FCC defines a signal as UWB when its band is >500 MHz or has a fractional band >20%.) The advantage of this technique lies in the fact that the brevity of the pulse makes the UWB less sensitive to interference due to the reflection of the wave itself.
“A single RF SoC by Vayyar is used to transmit/receive the 3.3 – 10 GHz RF signal. The system uses two boards: one for RF transceiver (yellow contour in figure 1), SRAM data collecting, and USB type data stream. One (green contour in figure 1) for data processing and BT/Wifi connectivity,” said Stéphane Elisabeth, costing analysis expert at System Plus Consulting.
The two boards as shown in figure 2 are connected with a Flex PCB structure to offer more flexibility. Thanks to the flexible nature of the connection, there is a considerable reduction in space, weight and costs compared to a similar solution on a rigid base.
The thermal dissipation management (TIM, thermal interface material) is carried out in two ways: “one on top of the application processor and another one directly on the heat sink, said Elisabeth. He added, “We assumed that the heat sink is made with an aluminum alloy type A380.”
A380 is one of the most commonly specified aluminum alloys. A380 alloy has excellent fluidity, pressure tightness, and resistance to hot cracking. Die casting with aluminum alloy 380, in particular, produces high-quality, cost-effective parts and products that are durable.
The RF SoC has 48 In/Outpath from the die to the balls under the package. “On the 48 RF I/O, only 42 is used to connect the antennas. The RF SoC uses an integrated DSP with a large amount of SRAM memory in the transceiver die and provides the data to the MCU. The MCU only transforms the SRAM data into a USB data stream type. The on-chip DSP exempts the design from any external CPU to execute complex imaging algorithms,” said Elisabeth.
The RF SoC uses 6-layer PCB substrate and is soldered on a 10-layer PCB substrate (figure 4). It is packaged using lidless FCBGA. The MMIC is made of two quadrature oscillators in order to generate an intermediate frequency signal on-chip to be directly processed by the Analog-to-Digital Converter (ADC).
In addition to the SoC processor VYYR2401-A3 RF, in the block diagram in Figure 2 and 3 we can see an MSM8909 processor that allows emergency mobile communication, and a Codec/Audio by Qualcomm device for speaker and microphone management. The Qualcomm Snapdragon 210 MSM8909 is an entry-level SoC for Android-based tablets and smartphones with four ARM Cortex-A7 CPU cores (quad-core) at 1.1 GHz. The SoC integrates Bluetooth 4.1 + BLE, 802.11n (2.4 GHz) WiFi, and a Cat 4 4G-LTE-Modem (LTE FDD, LTE TDD, WCDMA (DC-HSDPA, HSUPA), CDMA1x, EV-DO Rev. B, TD-SCDMA and GSM/EDGE) with a max. speed of 150 Mbps.
The Cypress CYUSB2014 controller sends the data analyzed by SoC RF to a USB protocol. It is a SuperSpeed peripheral controller, providing integrated and flexible features. It has a fully configurable, parallel, general programmable interface called GPIF II, which can be connected to any processor, ASIC, or FPGA. GPIF II is an enhanced version of the GPIF in FX2LP, Cypress’s flagship USB 2.0 product.
The radar signal management board consists of 21 antennas which ensure high resolution. “The dimension of the antenna implies a large RF board to support the 21 antennas,” said Elisabeth. As the system works for 3 to 10 GHz,the dimension of the antenna is large (λ/4 = ~15 mm). He added “the number of antenna is connected to have a better resolution. But as the frequency here is not just high frequency but low, so the size is very large.”
“The high working frequency of this system is about is around 9.6 gigahertz. So it’s based on the measurement of the bow tie,” Elisabeth added.
Bow-Tie design is widely used in applications for imaging, radars, Wi-Fi access points and pulse antennas due to its characteristics of low profile, large bandwidth, low losses and high radiation efficiency (figure 6).
A bow tie design is a wire approximation (planar-type variations) of a biconic dipole topology. The design of the antenna is ideal on a size and cost basis, offering simple geometry and robustness. Bow-tie offers good control of its input impedance, and it is easy to make.
Cost and Analysis
System Plus evaluated the costs of the entire system, highlighting how the RF SoC only took 10% of the cost of the system (figure 7).
“As the system is large, almost 30% of the cost comes from PCB (RF board, Wifi/BT antenna,..) and interconnects. Memory (RAM, Flash) and processor (Qualcomm Snapdragon 210) take almost 20% of the cost. 30% of the cost is related to discrete components such sensors, PMIC, connectivity front-end. Another good 10 % come from the display,” said Elisabeth.
Vayyar has developed other versions which are already on the market with operating frequencies of the order of 60-80 GHz. VYYR7201-A0 works at 57-64 GHz and VYYR7202-A1 works at 77-81 GHz. The first is implemented in the Vayyar V60G-home and is suitable for gesture recognition applications, but also for In-room people detection and of infants left in car. It has 46 linear polarized PCB embedded antennas. The other one has 40 linear polarized antennas and is implemented in the Vayyar V80G. It is suitable for In/outside car application and for the detection of potential intrusions.