1. Components placement, Position for main components, the LAN_GND should better include one chassis grounding path such as screw hole so as to offer a shortest path for surge energy flowing
2. The relationship between LAN_GND & GND, for debugging EMI-R & EMI-C issue. The related components may be more than just three if needed. Reserve more pads of components for debugging at different location on the moat.
3. Laptop or AIO type of RJ45 structure and pin definition schematic.
4. The new Surge solution with gas discharge tube based on the Bob-smith circuit. In the case of the space is very marginal, the 75 Ohm can be smaller size since the energy just goes thru the gas discharge tube path. So does the Hi-pot cap. It can be a 2KV or 1 KV Cap.
5. The cost down solution might be use for both cost reduction & space saving purpose. Only if IEEE and 130 meter package losing rate testing were passed, the gas discharge tube & 75 Ohm
can be reduced as the following design schematic. But be careful.
6. One of the most important conclusions of this solution is to use both L-L and L-G protection component at secondary side; also, it should close to transformer as much as possible. Since the L-L became not mandatory any more legally, some design can just reserve the 3304N pads only to save cost.
7. Regarding to the LED trace layout, EMI filter(470 pf Cap) and Surge switch (TVS) should be used the same time. Also, the LED trace should be far from the LAN differential pair as much as possible to avoid the Surge or EMI energy coupling onto. The LED is a weak link when Surge happened, lots of surge failure happened with LED damaged the same time. Also by most EMI debugging experience, the LED trace is very sensitive to EMI noise. That’s why the solution needs to be used on the LED trace.
Song Hang 2010-3-31