What methods should be used to tie DC power supply ground to chassis ground to prevent 3-phase AC Line to chassis ground surge testing from causing items powered by the DC supply to fault during a surge event? A fault could be caused by the surge causing a ground bounce which causes invalid data or signal noise.
Hi @ravi.blore , the fact that you are already thinking of managing ground planes is a sign that you are thinking about EMC at the good time!
Some advices:
Plan the stack-up carefully. Spend all the time you need to plan it. Use at least one solid ground plane (without slots), so traces are routed over it.
Use one ground, do not split them. The main goal is to provide a very low impedance return path to signals. Having a solid ground plane does that.
If you have a high number of layers, think about tight coupled pairs of layers.
Differential signals such as RS-422, 485 need a constant impedance, with a value that varies depending on the type of signal (120 ohm for CAN bus, for example)
Heads up: differential signals do not have RX and TX signals, but a Positive (P) and Negative (N) signals. Both pairs are used for RX and TX, with a specific protocol to determine who needs to transmit and receive.
Normally, we need just one termination resistor, placed on the furthest end, the place of the board that is the furthest from the driver. It is also important to keep a controlled impedance in the whole path to avoid reflections.
Regarding the shielding: when using a differential pair, we protect the signals against common-mode noise, so they are more protected than non-differential signals. However, you can still have couplings along the way, so it is when we need to incorporate shielding. However, it is usually done in very noisy environments
It is hard to define rules of thumb for Hardware design and EMC, since the conditions change from design to design.
However, what you can do is define some general principles:
Keep connections short and wide: any kind of conductive connection (PCB trace, plane, wire) has an impedance. An ideal situation is to have zero impedance. In the real world, we need to keep this impedance as low as possible.
Current goes AND comes back: always check where the energy comes from and where the energy goes. Check the current loops and keep them as small as possible
Sorry to say, not yet! We will need to wait a little before seeing fully automated solutions driven by AI. The main challenge with AI is to get good data, and the EMC data world is quite distributed and unstructured.
Today, what you can find at Denpaflux is a combination of AI tools and human expertise, where the work made by experts is made much more efficiently
Find your way. Do not accept the premise “EMC has been always hard, so your are going to have a hard time with it”. Learn how to work on it at different stages of the design and you will master it. Working on it at the design phase will save you a lot of time and frustration!
One surprise that people find (and I found in my first tests) is that EMC is not just electronics, but also other aspects such as Mechanics. We use the mechanical case to ground electronic devices, and this connection should have a very very low impedance. If the mechanical case is covered by a paint layer or has a mechanical treatment that increases its impedance, it can provoke high-loop currents and fail radiated emissions. That is why it is common to see people with drills in the EMC laboratories.
When making EMC reviews, it is important to look all the parts of the system, not just the PCBs. At Denpaflux, we check all files (block diagrams, schematic, PCB files and mechanical drawings) to ensure that there are no surprises in the lab!