PCB Layout for CMOS Sensors

This imaging sensor requires some particular layout rules

When I was younger and I got my first cell phone with a camera, I couldn’t stop taking pictures. It wasn’t until later that I learned about the different technologies that could be used to capture images and videos. If you are designing a product that will include a camera, then you’ll need to decide between a CCD sensor and a CMOS sensor for your product. Newer cameras almost exclusively use CMOS sensors for handheld cameras, and some major manufacturers have even discontinued their CCD sensor product lines. CMOS sensors are likely to be the sensor of choice in a variety of new products.

CCD vs. CMOS Sensors

The choice between a mono and color sensor for your particular application may be obvious, but selecting a CCD vs. CMOS sensor may not. The two types of imaging sensors are quite different and each is ideal for different applications.

CCD imaging sensors generally provide light detection with lower noise and higher sensitivity. When used for imaging, they generally provide higher resolution. CCDs use a particular manufacturing process at a lower scale than CMOS-based devices, thus they tend to carry a higher cost. These qualities make CCD sensors ideal for low-volume, high precision imaging devices.

Contrast this with CMOS sensors, which can be produced using the same process as CMOS-based ICs. These devices cost less and tend to use less power, but they tend to produce images with lower resolution and more noise. However, the way in which the CMOS sensors convert an optical signal to an electrical signal is much faster than the case of CCD sensors, thus CMOS sensors tend to provide high frame rate. This makes CMOS-sensors ideal for consumer electronics applications. Note that there is some debate on the image speed and quality produced with CMOS sensors compared to CCDs; the two products are not perfect substitutes for each other.

Another type of CMOS sensor is the back-side illuminated (BSI) CMOS sensor. By switching the order of the photodiode and copper conductors in a pixel stack, the photodiode in each pixel receives more light, allowing it to operate in low-light conditions. These sensors are currently used in the newest generations of smartphones.

The differences between CCD and CMOS sensors arise from the way in which excited charge carriers in each pixel are read by the imaging sensor. CCDs read from the entire sensor all at once (called global exposure), and each pixel is read individually using shift registers. The output from each pixel is converted to a digital number and transmitted to a processor as serial data. CMOS sensors use addressing to expose each pixel individually. An amplifier is integrated within each pixel, and each row is connected to a small ADC. These small components switch very quickly, thus CMOS sensors tend to operate faster than CCDs with the same number of pixels.

CMOS sensor for a DLSR camera

PCB Layout for CMOS Sensors

The integrated nature of CMOS sensors and modules reduces the  number of components you need to include on your board. Although these sensors can provide low noise images, a sensor itself has high sensitivity to noise. Ripple and fluctuations in power supply voltage, EMI from other components, and noise coupling from the substrate can all induce noise in the output from a CMOS sensor, leading to grainy or noisy images. Significant ripple and transient undershoot or overshoot can also degrade image quality.

These devices are inherently mixed-signal devices; they need to convert an analog signal generated at each pixel to a digital number, which is then transmitted as serial data to an external signal processing IC. For proper mixed-signal routing, you’ll need to consider how and where the ground return flows, connecting split planes with jumpers or zero Ohm resisters, and avoiding routing traces through splits in power planes. 

Some CMOS sensors will use a differential signaling standard like LVDS, and the basic routing rules with impedance matching for the particular signaling standard need to be followed when routing the converted signal away from the sensor. CMOS sensors generally require at least two voltage sources to power the digital and analog circuits in the chip. The analog circuits are typically powered to ~3 V, depending on the manufacturer. The digital circuits may require a range of different voltages, depending on the signaling standard used for the digital outputs.

The main challenge in noise reduction is shielding. These sensors are typically surface mounted or through-hole mounted. Some CMOS sensors come as a die that mounts to a BGA package, but these packages generally include a small number of balls compared to BGAs for something like an FPGA. Other image sensors are integrated with their signal processing block as a single system-on-chip (SoC). A stackup that includes at least four layers is normally recommended for these sensors or SoCs.

This allows you to spread a ground layer underneath the surface layer that contains the CMOS sensor, although you will need to separate the analog and digital sections over different portions of the ground plane. Note that, because of noise sensitivity, you should not split the ground plane to ensure sufficient shielding. Be sure to place other high speed/high-frequency circuits away from the imaging sensor.

Depending on the way a CMOS sensor is placed in a PCB, you may consider routing your input and output connections over a flex ribbon. If you do this, then you need to include grounded traces on the flex ribbon to provide reliable return paths. This reduces loop inductance associated with the circuit, which reduces EMI susceptibility throughout the circuit. Note, however, that some manufacturers will explicitly recommend that you not use a flex PCB with their CMOS sensors.

CMOS sensor connected to a flex ribbon

Finally, any capacitors that are used in the design for bypass/decoupling should have low effective series resistance (ESR) and effective series inductance (ESL) in order to accommodate the desired frame rate and switching speed. The signal processing IC generally runs at high speed, and the capacitors need to function as designed at high speed. Low ESR and ESL values ensure that the capacitor actually provides capacitive impedance over a broader range of frequencies.

Electrical vs. Mechanical CMOS Sensor Layout

Obviously, PCB layout for a CMOS sensor is an important part of integrating these components into a new product, but you will need to balance the electrical layout and the mechanical enclosure for your product. The mechanical layout needs to accommodate additional optics (lenses, filters, and/or fiber couplers) alongside the PCB that supports the CMOS sensor and associated components.

Depending on the functionality of the product, the optics may need to move in order to adjust focus. The CMOS sensor and the PCB generally need to remain stationary, unless the enclosure folds or has some other complex geometry. With more compact devices, such as smartphones or smart home products, you won’t need to worry about moving lenses or the sensor itself thanks to the use of phase detection or contrast detection autofocus algorithms. In terms of product design, this requires close collaboration between the mechanical and electrical design teams.

Working with the design and analysis tools in the right PCB layout and design software can help you design your next PCB to include a CMOS sensor. Allegro PCB Designer and Cadence’s full suite of design tools are designed with the layout and MCAD tools you need to ensure your next product can produce high quality images.

If you’re looking to learn more about how Cadence has the solution for you, talk to us and our team of experts.