MADISON, Wis. — ON Semiconductor doesn’t usually come off as a powerhouse in image sensor technology, largely because the $2.8 billion company has a broad product portfolio spread across different markets.
Moreover, ON Semiconductor’s image sensors aren’t designed into consumer products such as smartphones or tablets -- the two biggest market drivers in the recent growth of CMOS image sensors. The company’s focus on the high-margin industrial market for machine vision tends to keep it under the radar.
But this is a general perception that’s about to change.
ON Semiconductor just last week signed a definitive agreement to acquire Truesense Imaging Inc., which for 40 years served as Eastman Kodak’s imaging division.
Yole Développement’s list of key players in machine vision begins with Truesense and continues on to Aptina, ON Semiconductor, PhotonFocus, e2v, CMOSIS, Xfab, Lfoundry, TowerJazz, Sony, and Hamamatsu. Given this order of importance, ON Semiconductor has clearly carved out space on the most visible part of the map by buying its biggest competitor.
The acquisition, scheduled to close at the end of the current quarter, is expected to strengthen ON Semiconductor’s already established high-speed, high-performance CMOS image sensor business.
ON Semiconductor combined its home-grown imaging and light sensor business with the CMOS image sensor product lines it acquired in 2011 from Cypress Semiconductor. Cypress’s image sensor technology was rooted in IMEC spinoff FillFactory in Belgium, which Cypress bought in 2004.
On to automotive?
Industry observers expect ON Semiconductor to move its image sensor products into the automotive market soon. However, in an interview with EE Times, Edwin Ringoot, strategic marketing manager for the Image Sensor business unit, hesitated to comment on any specifics for the company's plan to move in the automotive sector. He only agreed that carmakers’ demand for multiple cameras is becoming a significant market force.
As Eric Mounier, a senior analyst at Yole, explains, CMOS image sensors for the automotive sector must have specific characteristics. They include: high dynamic range, no motion artifacts, a low price (less than $10), robustness, and low-light sensing.
Brian O'Rourke, senior principal analyst at IHS, agrees. The ability to stand up to difficult environments, in addition to near-IR capability and longer lifecycle, are particularly important for automotive, he says: “A new sensor maker entering the automotive market would need to invest years before seeing a payoff.”
That timetable might well serve ON Semiconductor. The company has specialized in high-performance image sensors for machine vision for years, although it probably needs further modifications to its image sensor product lines to compete effectively in automotive.
Internally defining high-speed sensors as a product capable of capturing images between 250 frames per second and 500 frames per second, ON Semiconductor has the No. 1 share in that specific market, according to Ringroot. Imagine a car driving at high speed on a highway, he says. The ability for that car to detect a small object on the highway would require a high-speed image sensor.
Making it clear that ON Semiconductor does not compete with smartphone image sensor suppliers such as Sony or Omnivision, Ringoot says his company has a stronghold in such markets as specialty CMOS sensors and standard and turnkey custom image and light sensors.
Armed with very fast global shutter 2D sensors and products that meet industrial and space/military standards, ON Semiconductor’s image sensors are typically gunning for applications such as machine vision, biometrics, barcode, medical, digital cinema, space and military, motion analysis, and intelligent traffic system (ITS), says Ringroot.
Against that backdrop, ON Semiconductor this week launched a new family of CMOS image sensors called Python. With resolutions of 300 K, 500 K, and 1.3 megapixels, respectively, the Python 300, 500, and 1300 are designed to address the needs of general-purpose industrial image sensing applications. Later this year, ON Semiconductor is planning to add more Python sensors with higher resolutions.
Next page: Techno requirements for machine vision