1Introduction
With the development of technology and smart devices, machines are becoming increasingly intelligent. To make machines smarter, we have added “eyes” to them, allowing them to observe the world we live in and make decisions based on what they see. The “eyes” of smart machines are the cameras we are familiar with. However, the application of camera modules can bring some radiation issues, particularly the issue of clock signals.
2Introduction to Camera Module
Figure 1 Common products with cameras
Figure 1 shows some common machines with camera modules. Although their appearances vary, the camera module parts inside them are quite similar.
Figure 2 Camera architecture
It can be observed that common solutions typically use a DSP + ISP + CMOS sensor framework. We can see that the camera module contains a main clock (MCLK) and a pixel clock (PCLK), which are the main focus points for rectifying the excessive radiation from the camera.
3Case Analysis
Figure 3. ADAS camera
This is an ADAS camera, which only has camera functionality. I will use this case to share the radiation rectification of the camera module.
The initial test data is as follows:
Figure 4. Initial data
From the data, we can see that this camera’s radiation exceeds the limit, with several single noise exceeding the limit and one data packet being large. The narrow-band single signals here are caused by the clock data on our machine. Additionally, we found that there is not just one clock frequency exceeding the limit, but multiple clock frequencies.
Next, we look at the PCB and find two obvious issues:
1. The PCLK and MCLK clock lines are not properly grounded and isolated from the data lines, causing clock noise to couple into the data lines and radiate out.
2. The data lines and clock lines have not undergone any filtering measures (RC filtering), leading to severe data exceeding the limit.
Therefore, we shielded the layout and added 1nF and 10nF capacitors for filtering at the power supply.
Figure 5. Partial PCB image
This product has only two clocks: one for the camera’s PCLK and MCLK, and the other for the SPI FLASH CLK.
They are all provided with a reference clock by the DSP (27MHz crystal oscillator). Therefore, we can utilize our company’s flagship product, the spread spectrum IC, to spread the main clock of 27MHz.
Figure 6. Spread spectrum usage diagram
Figure 7. Data after spread spectrum
Final rectification plan:
1. Add spread spectrum SSDCI1108AF-08-CT to the 27MHz main clock.
2. Shield the wiring.
3. Add 1nF and 10nF capacitors for filtering at the power supply.
4Conclusion
The application of camera modules in an increasing number of products has led to many clock issues. At the same time, clock problems are a significant challenge in EMC. Through this case study, we can see that using spread spectrum ICs to directly address clock issues is a powerful tool, but it is only a tool and cannot completely eliminate clock problems. Therefore, attention should be paid to grounding and isolating clock traces during the initial design to avoid clock noise coupling into other lines. Additionally, we can reserve RC filtering on data and clock traces for future rectification. I hope the rectification ideas from the above case can provide some help to engineers, and I welcome any suggestions for improvement.
About Tactics
Dedicated to providing one-stop EMC solutions
●Technical services ●Experimental testing
●Corporate training ●Component sales
Utilizing new products to stay at the forefront of the times!
Welcome to consult!
Tel: 0755-82908734
Mobile: 13510869661
13417580974
E-mail: [email protected]
Corporate WeChat assistant
QQ group
Leave a Comment
Your email address will not be published. Required fields are marked *