With the emergence of increasingly compact components and high-intensity lighting, thermal management and heat dissipation are critical factors that are frequently overlooked during the initial stages of designing a machine vision system. If key components like lighting, cameras, lenses, and vision processors retain too much heat, it may affect the system's longevity and reliability, along with the inspection quality, consistency, and robustness.
Lighting
The intensity and uniformity of machine vision lighting are crucial factors that contribute to the accurate functioning of the system. Heat generated by the LED illuminators can cause a significantly impact if not properly managed. Elevated temperatures can cause a reduction in the light output, as excessive heat degrades the LED's basic structure and its efficiency. As the junction temperature (the temperature of the active region within the LED) increases, the LED's light output decreases. Since most LED illuminators contain many strings of LEDs, heat distribution is non-uniform, which in turn can lead to uneven light distribution.
High temperatures can also cause color shifts in LED lighting, significantly impacting color inspection in machine vision applications. In applications where the system is detecting shade breaks, it will impact the system's accuracy. This is one of the reasons why having a white reference visible in the field of view can help ensure color measurement repeatability.
Halogen lighting is making a little bit of a comeback in machine vision, where it is used in Short Wave Infrared Wavelength (SWIR) applications because they are full spectrum light sources, whereas LEDs tend to be a narrow band. An increase in the filament's thermal temperature will shift the spectrum of the emitted light and make the filament much more shock sensitive. Machine vibration or bumping a hot halogen light will kill it quickly.
Machine vision lighting manufacturers do a great job of mitigating uneven illumination, but the following thermal management strategies can improve longevity:
- Less than 100% Duty Cycle: Only turning on the lights when needed or strobing the lights reduces the waste heat from the LEDs. By strobing the lights rather than operating them continuously, the LEDs can operate at a higher intensity in shorter bursts (typically at very low duty cycles of less than 5%). The "off-time" between strobe pulses allows the LEDs time to dissipate heat, effectively transferring the generated heat load within the light head. Rapid turning on/off halogen lights is not recommended because of the stress it causes on the filament. If needing a full spectrum light source, an arc-flash type bulb is better suited.
- Beefy Mounting Brackets: A 6 mm (1/4 inch) aluminum plate can be an effective heat sink when mounted across the entire length of an LED light. A 3mm aluminum plate between the light and heavy brackets creates an improved thermal transfer.
- Plate Fin Heatsinks: Plate fin heatsinks work by dissipating heat through conduction, convection, and radiation. They are designed to maximize the surface area that comes in contact with the cooling medium, typically air, to effectively transfer and disperse the heat away from the heat-generating source.
- Thermoelectric Cooler Plates: thermoelectric coolers (TECs) or Peltier coolers are solid-state devices that can provide cooling or heating by converting electrical energy into thermal energy using the Peltier effect.
- Custom Enclosures: In extreme heat applications, lights (and cameras) can be placed inside of the enclosure that circulates air or coolant through a radiator.
High-Intensity Pattern Projecting Spot Light SL256 - Advanced Illumination
Optics
Your industrial lens rarely has a narrower operating temperature range than your camera, but optical films and polarizers can be of concern. Polarizers are often used in machine vision to eliminate glare; polarizers' downside is that they degrade over time. With the advent of very high-output LEDs, especially lens-concentrating versions, the standard acetate plastic film linear polarizer material is subject to thermal and energy absorption breakdown. Higher temp plastics and wire grids sandwiched between the glass can be used, but they can be difficult to cut to fit or properly provide full coverage over a light.
For line scan applications, thermal management of the air between the lens and the object to be imaged must be considered. The quality of images can be affected by changes in the air's properties, such as temperature and humidity. These changes can cause the air to bend light differently, leading to minor blurry or distorted images. One common cause of these changes is heat produced by the camera, its lights, other nearby equipment, or the product itself (i.e., large sheets of aluminum being hot rolled). Heat can cause air to move in a turbulent way, creating areas with different light-bending properties. This can make the image look warped or distorted. The amount of distortion depends on many factors and is hard to predict.
Linescan cameras capture images one line at a time as the part (or camera) is moved along the camera's field of view to build the full image. Hot air turbulence can cause a small shift in the position of the image in the direction of flow. The shift can easily be measured by imaging a static line pattern.
Sketch of the measuring scheme for the image shift (source)
Fortunately, the image warping effect can be easily suppressed by using a fan to create a consistent airflow in the optical path between the came and the object.
Cameras
The temperature significantly influences the image quality of camera sensors (particularly in SWIR). As the temperature increases, dark noise and dark current levels rise, leading to a degradation in image quality. Depending on the sensor type and other factors, this effect becomes more pronounced when the sensor's temperature exceeds 40-50°C. An unstable or excessively high temperature can compromise the system's overall performance, potentially damaging electronic components or causing system failure.
Outside of the ambient temperature, heat generation within a camera can occur from multiple sources; the sensor, FPGA, and the interface. Camera FPGAs are used for pixel correction, synchronizing acquisition between multiple cameras, and debayering a color sensor. Power-over-ethernet offers convenience and reduced complexity in the number of cables and connections required. Still, the compliance voltage on such systems is 48-52V, and the excess voltage bleeds off as heat. An increase in camera temperature can also lead to throttling of its frame rate. During commissioning and at quarterly preventative maintenance visits, it is important that a surface temperature measurement is taken to ensure the camera's longevity.
A quality industrial machine vision camera has some mass to aid in thermal management; the bargain-priced so-called industrial camera tends to be significantly noisier. Similar to the strategies mentioned above, with lighting, the same techniques can be used with camera sensors, beefy brackets, and plate fin heatsinks. In more severe environments, TECs or thermal-cooled custom enclosures.
Material Selection for a camera holder (source)
When a camera is mounted to a telecentric lens, it is not uncommon for the mounting bracket to be attached to the lens and not the camera. Attaching a plate fin heatsink can drop its operating temperature by 10 C!
Embedded Processors & Industrial PCs
Both Embedded Processors (i.e., smart cameras) and Industrial PC are mostly designed to be fanless. However, a system commissioned in the winter can behave differently than in the summer if the industrial environment is not air-conditioned. Prolonged exposure to excessive heat can cause premature aging and failure of electronic components, such as processors, memory modules, and power supplies. The biggest impact on a Production environment is throttling.
When an industrial PC's components, such as the CPU or GPU, become too hot, thermal throttling can cause these components to reduce clock speeds to prevent damage. This reduction in clock speed directly affects the system's processing capabilities and makes the processing time in machine vision algorithms, such as object recognition, measurement, or defect detection.
Passive cooling can be insufficient in thermal management for heavy machine vision algorithm loads; active strategies include:
- Add Internal Fans: Most box-style industrial PCs have an option for a fan kit and are well under $100, an excellent investment. Industrial-rated fans will have a 3-5 year lifespan. Be sure to add it as an action item to the preventative maintenance schedule.
- Mount a fan on top of the heatsink: The thermal efficiency of the fin thermal heatsink on the box PC can be dramatically improved by mounting a fan on top of the unit. The increase in airflow speeds up the dissipation of heat.
- 3D print an air horn for the GPU: Adding an internal air horn that focuses the flow from one of the internal fans onto the heatsink can help prevent the GPU's throttling. Consumer-grade GPUs are not designed to run continuously in a production environment. Using a server-grade GPU is recommended for use at The Edge.
- Cooling the control cabinet: Upfront costs for using a Vortex cooler is low, but some manufacturing facilities ban them because they are less energy efficient (a hotly debated topic). Or because the plant's compressed air is not clean enough to run into a control cabinet. Air conditioning units are expensive. As an alternative, the price of cabinet heat exchangers using TECs has decreased dramatically. All of these items do require regular preventative maintenance.
Advantech Mic-770 with a custom 4VT fan mount
Conclusion
Effective thermal management is critical in ensuring a machine vision system's longevity, reliability, and performance. As these systems continue to evolve with more compact components and high-intensity lighting, addressing the heat dissipation challenges in lighting, optics, cameras, and embedded processors or industrial PCs becomes increasingly important. By implementing a combination of passive and active cooling strategies, such as duty cycle management, heatsinks, thermoelectric cooler plates, custom enclosures, and fans, system designers can mitigate the effects of heat on their machine vision applications.
Moreover, monitoring and maintaining these systems regularly is essential to ensure optimal operating conditions and prevent issues like thermal throttling or degradation of components. By investing in proper thermal management solutions and adhering to a preventative maintenance schedule, manufacturers can avoid costly downtime, extend the life of their machine vision systems, and ultimately improve the quality and consistency of their production processes.
