3
\$\begingroup\$

In my current design I have one RJ45, one RS232 and two RS485 ports. I have TVS diodes CDSOT23-SM712 for serial channels and ESD7104MUTAG for Ethernet port. Now I have connected the ESD GND of the TVS to a separate GND island (referred to as CHGND) and then AC coupled that to System GND via 1nF (2kV) in parallel with 10MΩ (both 1206 package). My idea is to not directly couple noise from ESD return to system ground. This is also suggested in Microchip App note AN2054.

My system is powered by 12V DC power supply coming from a mains Flyback or external 12V DC power supply. I don't have a Chassis GND or Earth. The enclosure is ABS.

Now my review engineer is against that, pointing out that once a ESD strike comes in, it lifts the CHGND high (or low) until the bleeding resistor discharges that. This makes the line we are trying to protect experience the high voltage relative to GND. (I simulated this for 8kV HBM pulse and the voltage rise was about 770V) Also not working for repetitive strikes.

What do you guys think? Is it better to directly connect ESD return to system GND or AC couple?

Thanks in advance.

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
3
  • \$\begingroup\$ At high dV/dt (e.g. when lightning strikes, when ESD is inflicted on the cable, etc.) the capacitor will be present as a low impedance. Therefore, your so called "CHGND" should already be 'directly connected' to system GND (just momentarily). The problem with your simulation is that the "low impedance" isn't quite low enough. This can be fixed by increasing the capacitance. Try that and see what happens? \$\endgroup\$
    – DSWG
    Nov 1, 2018 at 0:32
  • \$\begingroup\$ TVS diodes n-terminal(cathode) should be signal GND pour and that pour should b e connected to Chassis GND only using High speed capacitors. This signal GND pour should be connected to System GND near to high speed capacitors. This is to make sure ESD event will be diverted to Chassis, but for normal operation Chassis GND is isolated from Signal GND so, that No common mode currents are possible. \$\endgroup\$
    – user19579
    Nov 2, 2018 at 9:12
  • \$\begingroup\$ Thanks all for your valuable comments. I decided to connect the return grounds of all TVSs to a moat and connect it to system GND at the 12V feed point of the comms board. Then add a CMC, a TVS and some bulk capacitance to 12V feed in point. \$\endgroup\$
    – Kaush
    Nov 2, 2018 at 20:14

1 Answer 1

Reset to default
3
\$\begingroup\$

You have an ABS enclosure and some serial ports those need to be protected against ESD.

I think your review engineer is right. There is no need to the capacitor C1 and resistor R1.

First of all I recommend every body to reread "System level ESD co-design" That is Published by John Wiley and Sons. ESD is acting like a 300+ MHz signal. your ESD protection circuit and its PCB must be able to pass this signal through itself. It is true and also a good practice that the return path of the ESD towards the main ground pin of the power supply connector must be some how separated (physically or by moat). If the protector device inject the ESD into a ground plane, it could make phenomena line soft errors.

Therefore, It is a good practice to have a ground polygons under all of the connectors (all the connector body must be over the ground polygon and the polygon must be bigger than connector occupied space on the board) and connect the protection devices to that polygon and through it to the main ground pin of the power supply. The injected ESD to this polygon by protection device will drained into the main ground pin of the board (with shortest possible way) and will not bounce back to the sensitive parts of the circuit. Obviously, the protection devices must be close to connectors too. your ground plane/planes (if you have multi layer board) also connected to the main ground pin of the power supply connector but the ESD will not make trouble for them because drainage capability of your main ground pin.

In case of ESD injection onto plastic/metallic connectors the ground polygon that is under them in the assembled layer, will absorb and guide it towards the main ground pin.

in case of ESD injection onto ABS enclosure, you must be worry about any ABS pipes or gussets those are connected to or close to your PCB. These unwanted proximity with enclosure could guide the ESD behind your protection lines and make bad damages for you. All such proximity with enclosure must be protected by direct guard ring/trace towards the protection polygon that has direct adjacency with connectors.

In case of ESD injection into high frequency signal cables, all of related pins have been protected by proper TVS and related Y-shaped connection on the PCB. the Y connection will guide the ESD towards main ground pin through the ground polygon and the protection device has very few attenuation on the main regular signal.

In case of ESD injection into low frequency signal cables, all related pins normally protected with ESD capacitors or special protection chips and on the PCB the thick Y connection is recommended to use at the entrance of the protection device and narrow trace 90 degree connection recommended for after protection device towards protected circuit to maximize the impedance discontinuity and reflection of unwanted high frequency waves those are injected towards sensitive parts. obviously ESD capacitors with footprint 0603 or lower could be destroyed by passing through 15kV ESD. Please look at the following presentation by Bosch. http://www.emcsociety.org/2011%20Events/IEEESEMMLCCESD.pdf

Normally I use the combination of a Spark gap or intrinsically spark gap (the grounded polygon adjacent to connectors at their assembled layer (top or bottom)) and the protection device that is TVS, protection chip or ESD capacitor.

\$\endgroup\$
8
  • \$\begingroup\$ Hi BD_CE, Thanks heaps for your detailed answer. That cleared out many doubts in my mind. One other issue I have is that my mains power section is on a different board and the communication interfaces are on another board. They are connected by a two wire 12V DC supply link. Now I have a moat on the comms board around connectors, to which I have connected the GND of all TVSs. I would prefer to take it to power board GND through a hex metal standoff, but its practically not possible. I can connect the moat to the GND of the 12V harness but I am not sure if this is a good idea. \$\endgroup\$
    – Kaush
    Nov 2, 2018 at 4:35
  • \$\begingroup\$ You are welcome. I recommend reading GROUNDS FOR GROUNDING A Circuit-to-System Handbook by John Wiley and Sons to all designers. To mitigate the unwanted emission and RF susceptibility, the 12V DC supply link must be twisted pair. There is no need to use standoff and the wire link is enough. more over standoff is not so reliable and could transfer unwanted mechanical tension to the board. I did not understand the following sentence: " I can connect the moat to the GND of the 12V harness but I am not sure if this is a good idea." what is the gain of connecting Void trace to ground? \$\endgroup\$
    – BD_CE
    Nov 2, 2018 at 5:12
  • \$\begingroup\$ Hi BD_CE, Thanks again. The moat has all the GND connections of the TVSs. So it is the dumping ground of all the ESD energy. My idea is to connect that to the power ground right at the power output, which is on the power board, away from all other signals and GND planes. However if we dont use a metal standoff kind of structure to carry this ESD energy to the power board, then I have to dump it at the 12V input of the comms board. Then the ESD energy will go through the 12V harness down to power board. This would couple the ESD to the 12V rail, isnt it? have another TVS across 12V/GND? \$\endgroup\$
    – Kaush
    Nov 2, 2018 at 11:26
  • \$\begingroup\$ Now I got the message. Do not be worry about coupling on the 12 V DC power line.Its drainage feature will prevent bouncing back the ESD towards the board sensitive parts. There is not much difference between DC power lines. For +15kV ESD injection the GND line is slightly better path and for -15kV ESD injection the 12V line is slightly better one. Also DC lines are well protected. At your comms board DC power rail entrance it is recommended to have a common mode filter then differential mode one then ESD protection device then in-rush current protection device and then big capacitor. \$\endgroup\$
    – BD_CE
    Nov 2, 2018 at 13:15
  • \$\begingroup\$ So the idea is to have a CMC, CBF, TVS and bulk cap (12V filter block) to the 12V feed point to comms board and connect the moat for ESD dumping right at the 12V connector GND point, between the filter block and the 12V connector. This way the inductance in the filter block will limit the ESD propagating to the GND plane of the comms board and most of the energy will be diverted towards the power board via the 12V feeder cable. \$\endgroup\$
    – Kaush
    Nov 2, 2018 at 20:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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