EMC Consutancy & Design solutions
Inductive Speed Car measurement
We were contacted by an EMC testing house a few years ago regarding the inability of one of their customers to pass the EMC test.
Prior to doing EMC testing in this lab, this customer had tried three other labs before choosing this one. After the third attempt, the product manager was so sceptical of the high emission level that he decided to repeat the test with another lab (the one that had contacted us) in order to eliminate the possibility of bias. Our investigation at the EMC house took three days. In the end, we were able to locate a few design errors in the original design that needed to be rectified. Specifically, there was a lack of separation between all the different parts of the design, there was an error in the Ethernet surge protection, there was an erroneous filter on the MCU, and there was not enough filtering on the DC input to protect the system.
During the course of the project, we worked on all of these aspects. We removed unnecessary ferrites on the VCC and VCCIO for all of the microcontrollers and reduced the emissions due to the 25 MHz clock. As a result of our redesign, the unit has been re-tested and it has passed its EMC test as a result.
Creating a separation between the sub modules
As a rule of thumb, it is usually a smart idea to create some separations between different parts of your design in order to avoid confusion.
For example, if you have two power supplies, let's say a nominal power supply and a redundant power supply, it makes sense to separate the two. This is because the return currents do not mix between the circuits of the two power supplies. Not just when it comes to the power supply, but also when it comes to every sub-module of your design, this is true.
In order to do this, the first thing we did was to physically separate the logical units from each other.
.png)
Third order harmonic reductions
In the frequency range of 200 MHz-300 MHz, if you look at the radiated emission of a design, you may be able to get some important information regarding the ROS of a signal when you look at the radiated emission. It is generally accepted that if a line is not well terminated, you tend to have some ROS (return of signal) and this tends to create harmonics of the third order.
By "gently" improving the slew rate of the 25 MHz clocks present on the board, we were able to eliminate its occurrence.
reducing the emission from Ethernet
A general rule of thumb is that when emissions below 200 - 300 MHz are present, they tend to use the cables as antennas. In most cases, when there is an issue with emission above 300MHz, it can usually be found at the PCB level.
However, in the original design, we had a failure at 1 GHz. The reason for this failure was Ethernet. The PHY that were used were underpowered and when they were driving the ethernet transformer there was an overshoot, that eventually created some emission through the cables. As a standard practice, we always use low pass filtering (even on ethernet) and so this was resolved during the final EMC testing.
The customer was finally able to sell their product.