1
\$\begingroup\$

My circuit uses LAN8742A ethernet phy. The system has a metal case and gets 220V AC to power supply board. After converting 5V DC it goes to my board that has Ethernet and mcu. I screwed case and power supply pcbs earth point to mains earth. The problem is this: When i strike 2kV with ESD machine to metal case power supply generates spike appx 48V on output :/ And it goes to my circuit. I have standart linear regulator circuit for 5V to 3.3V. Spike value decrases a bit little but neverthless it locks/breaks ethernet phy. Ethernet leds goes crazy or hangs at wrong position. After i power off AC and power on again everthing starts to work normally. I need an advice for solving this. My circuit scheamatics are: power ethernet

\$\endgroup\$
1
\$\begingroup\$

It seems that you are experiencing soft error due to ESD injection. It means that your MCU internal registers have been altered by the ESD injection. Most of the available chips in the market are already protected for 2kV ESD spikes. Large number of new chips are also have internal protection for 4kV ESD spikes. Theoretically, with 2kV ESD you must not have any problem on your circuit. The 48V spikes those are generated during ESD injection are also OK and even natural reaction of ESD electrical charge distribution over the internal capacitor of the ESD gun and the intrinsic capacitor of your circuit. By the way according to your schematic I have following recommendations for you. Based on what you have written, part of ESD current finding its way towards the GND pins, through your MCU.

  1. read "System level ESD co-design" that is published by John Wiley and Sons.
  2. add a dedicated GND section to your board dedicated for the connectors and ESD protection devices and connect it to the main ground plane of the card using small opening (just for supporting the return paths of the circuit).
  3. Add an ESD protection TVS or capacitor (10nF 0603) close to your power rail connector (between 5V, 0V)
  4. An ESD protection device is necessary for all the system I/O pins (including RJ45 Ethernet connector pins(using especial solid state bidirectional ESD protection device)) and their connection with ground must be separated by moat or whatever from the card main ground plane and have very short paths towards the GND pin of the connector.
  5. The metallic case must not have connection/adjucency with not earthed parts of the PCB.
  6. The correct Earth must have less than 3V AC with Neutral at the AC outlet otherwise your earth could not do its duty well.
  7. Use "guard ring" traces those have direct connection with the main ground pin of the connector and surround the board screw holes with it and arm all of its borders with it.
  8. Some changes in the internal firmware of the MCU are also a good practice. For example the firmware developer could follow "scrubbing" patterns. she/he could use watchdog timer and even you could add external watchdog timer to your schematics.

Good luck.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ i seperated my custom pcb from metal case but nothing changed. Because the ESD enteres from power input. When i hit the common earth point with 2kV i can see a spike at power supply output. So i added a mini pcb between power supply output to pcb power input. There were a bi directional TVS(PTVS5V0S1UR) and series resistor. But TVS didnt do its job :/ I changed resistor values and try with inductors which are different values (1uF-100uF) nothing changed:/ I changed TVS uni directional model(SMBJ5,0CA) to bi directional model nothing changed :/ ı cant understand why TVSs dont not clamp the ESD \$\endgroup\$ – İsmail Fatih ILTAR Nov 23 '18 at 9:19
  • 1
    \$\begingroup\$ Finally i totaly seperated power. i energized pcb with battery and when i hit the metal case my board was effected. i carried my pcb to far away from metal case thant it worked. So When the ESD hits, the metal case turns into antenna. Than As i understand my ethernet chip was effected by this RF ESD or something like that. \$\endgroup\$ – İsmail Fatih ILTAR Nov 23 '18 at 13:40
  • \$\begingroup\$ Metallic enclosures are susceptible for ESD but In your case the 2kV ESD is abnormally effective on your circuit! There must be some connection between the case and internal sensitive parts behind the protection devices. any connection of the enclosure with the MCU heat sink could be the root. The concept is this: "Save your sensitive parts from the discharge path/paths" if the ground of the TVSs are shared with the processor ground without any RF attenuation (moats or what ever) you will have problem. for ESD toleration you could think that it is a 300MHz~500MHz signal and design based on it. \$\endgroup\$ – BD_CE Nov 23 '18 at 19:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.