- This article is filed under:
- Bar Code/OCR
- Component Verification
- In Process Alignment
- In Process Verification
- Laser Inspection
- Laser Measurement
- Material Analysis
- Material Handling
- Medical Imaging
- Measurement (Non Contact)
- Non-Visible Inspection
- Process Control
- Quality Assurance
- Robotic Guidance
- Spectral Analysis
- System Verification
- System Analysis
- Thermal Imaging
- Video Maintenance
- Visual Inspection
- Visual Testing
- 3D Modeling
- Consumer Goods
- Containers (glass, plastic, metal)
- Fabricated Metal
- Food & Beverage
- Medical Devices
- Medical Imaging
- Miscellaneous Manufacturing
- Scientific Imaging
- Semiconductor Equipment
- Traffic Systems
- Application Specific Machine Vision Systems
- Cables/Cable Assemblies/Connectors
- Embedded Vision Processors/Compact Vision Systems
- Frame Grabbers & Other Imaging Boards
- Image Sensors
- Smart Cameras
- Video Microscopes
- Vision Systems
- System Integration - Vision Systems
- System Integration - Automation
- System Integration - Robotics
- » View All
FLIRs Guide to Integrating Board Level Cameras for Embedded Vision
by FLIR - FLIR Systems, Inc. Posted 10/01/2020
The availability of compact and powerful single board computers is enabling exciting new product designs. This is particularly useful in applications where miniaturization improves cost and/or efficiency. Additionally, vision systems are able to leverage fully-featured board level machine vision cameras to reduce a products overall size even further and provide operational flexibility while supporting custom or non-standard optics. Typical examples include medical diagnostics, metrology, robotics, embedded vision, packaging and print inspection, handheld scanners, benchtop labs and other space constrained systems.
In this article we cover several important aspects to consider when choosing an embedded vision camera; these include feature set, form factor and physical footprint, interface options, lens mounting, software support, thermal management, and electromagnetic compatibility.
Form Factor and Feature Set
When transitioning from cased to board level cameras, system designers should carefully consider their imaging and camera performance requirements. Many small board level cameras only support low resolution sensors, few GPIO lines, and limited on-camera features. Conversely, the board level variants of many full-featured machine vision cameras are simply standard cameras with their cases removed. While these cameras may achieve the required imaging performance, they may not be significantly smaller than standard cased models. Such cameras frequently use standard GPIO and interface connectors which are bulky and not ideal for embedded applications. For example, typical industrial locking connectors alone are about the same size as a Blackfly S board level camera.
FLIR’s Blackfly S board level cameras were designed from the ground up with embedded systems in mind. They deliver the same imaging performance and rich feature-set found on cased Blackfly S models in a remarkably compact 29mm x 29mm x 10mm form factor; while compact GPIO and interface connectors yield additional space savings. Another important benefit to FLIR’s embedded vision camera line-up is the availability of the same form factor on all cameras ranging with sensors from 1/3” to 1.1” - a consistent form factor across multiple camera models makes developing and upgrading systems and future product variants extremely easy.
Board level cameras are an attractive option for customers looking to integrate non-standard optics or to place the image sensor as close to their target as possible. With no fixed lens mount, board level cameras give designers the freedom to select optics other than the standard C, CS or S-Mount lenses commonly used in the machine vision industry. This design is also ideal for biotechnology and laser beam profiling; applications that frequently use no lens at all. Another common application of a board level camera is to enable the lens mount to be integrated into another product part – hence the term ‘embedded vision camera’. Furthermore, moulding a lens mound directly into a product’s housing can further reduce costs by simplifying manufacturing and assembly. To evaluate a board level camera which does not ship with a lens mount, a mount accessory should also be purchased. If cased models with identical sensors and features to board level models are available, these can be used as development platforms.
When it comes to choosing the right lens mount option for a board level camera, one of the most important factors to consider is the size of the sensor used. In general, S-mount lenses are designed to be used with 1/3" sensors or smaller with lower resolutions, typically less than 2MP. CS-mount lenses on the other hand are designed to work with sensors that are between 1/3” to 1/2”. If the sensor is 1/2” or bigger, it’s best to use a C-mount lens.
Cased machine vision cameras rely on the surface area of their cases to dissipate heat generated by the sensor, FPGA and other components. With no case, high-performance board level cameras may have additional design requirements to ensure they are operating within their recommended temperature range. In such cases, providing adequate heatsinking is key. Manufacturers will commonly provide a maximum junction temperate for this highest temperature component. On FLIR Blackfly S cameras, the maximum junction temperature of the FPGA is specified at 105 degrees C (221 Fahrenheit).
System designers must ensure that their thermal management solution meets this guideline. The size of the heatsink, the surface area of the chassis the camera is mounted to, or the type of active cooler required will depend on the sensor, the frame rate, the operating environment and the amount of on-camera image processing being carried out. In order to attach the heatsink on the camera, we recommend use of thermal pastes over thermal pads to minimize board stress on the camera.
Case Design and Rapid Prototyping
In most cases, board level cameras are integrated directly into an embedded vision system/product and a case is not required. However, for applications where the camera will not be integrated into a product, and thus camera internals are left exposed to the elements, a case might be necessary to prevent damage. For rapid prototyping, embedded system designers can easily design and print a case for the camera using 3D printers or use generic plastic cases that can encapsulate the camera and mount the camera in place using spacers and mounting brackets.
Interfaces and Connectors
USB 3.1 Gen 1 is an ideal interface for embedded systems. Its ubiquity guarantees support across a diverse range of hardware - from desktop PCs to ARM based single board computers (SBCs). Direct Memory Access (DMA) keeps latency to a minimum without the need for filter drivers. USB 3.1 Gen 1 also provides power and up to 480 Mbytes/Sec of data throughput over a single cable, simplifying both mechanical and electrical design.
A key goal of embedded system designers often entails miniaturizing existing designs. In these cases, the maximum cable length is much less important than the volume of the cable and connector. Flexible Printed Circuit (FPC) cables can support USB 3.1 Gen 1 over cable lengths of up to 30m. As the name suggests, FPC cables are flexible and can bend and twist to fit inside tightly packed systems. Additionally, high quality locking connectors and shielded FPC cables with locking tabs can ensure a highly secure and reliable connection.
However, one potential drawback of USB 3.1 interface is that it is a high frequency signal which can cause interference on wireless devices up to 5 GHz (e.g. GPS signal). For applications that utilize such wireless frequencies, we offer FLIR board level cameras with GigE interfaces too.
MIPI CSI is another common interface on many embedded boards. However, the complexity of the MIPI protocol and drivers can make development more time consuming compared to USB. Low-Voltage Differential Signalling (LVDS) based interfaces are also available and are designed to interface directly with a host-side FPGA; however, each channel of signal transmission requires two wires – a small but important drawback in certain applications.
When selecting a camera for use in an embedded system, software support is an important consideration which should not be overlooked. An SDK which supports both desktop and embedded systems gives designers the freedom to develop their vison applications on familiar tools and easily deploy them to the embedded platform of their choice. FLIR’s Spinnaker SDK provides support for desktop Windows and Linux on x86, x64, and ARM based systems.
Without the shielding provided by a case, the Electromagnetic Compatibility (EMC) of board level cameras will be different than cased models. All cased machine vision cameras from FLIR are EMC certified; however, board level cameras are not. As these board level cameras are embedded into other products / systems, the final product needs to be certified separately. Irrespective of the application, it is always advisable to follow best practices for electromagnetic interference (EMI) management just like any other electrical component.
Board level cameras are revolutionizing embedded vision systems providing the freedom and flexibility to design innovative products that are compact and versatile. Besides the factors covered in the article, it is also important to consider futureproofing your embedded system using high quality sensors, optics and reliable components. FLIR’s entire line-up of board level cameras are designed ground-up with such applications in mind and come with an industry leading 3-year warranty. Our machine vision experts can help you select the right level of imaging performance and form factor optimized for your embedded system.