| ??? 07/12/06 04:15 Modified: 07/12/06 04:22 Read: times |
#120055 - Some answers... Responding to: ???'s previous message |
David said:
-Since this TVS is bidirectional, I presume it will guard against both positive and negative ESD events? Yes, this thingy is a SMD varistor working in both directions. David said:
-Do I need the ferrite bead shown in the application circuit on the right? If I understand correctly, this is only to help protect the IO pin from the clamping voltage of the TVS? ie, the TVS clamps the voltage at, say, 18v. Obviously, this is too high for a standard IO pin to handle. So, the bead helps reduce this current/voltage spike, while the TVS handles the large spike? If that's correct, I will plan on using one. Yes, your assumption is correct. But it's not so much the protection of I/O pins, which should now easily be able to withstand the rest ESD energy, but to provide a high impedance for ESD currents wanting to flow to and from your 4-layer board! The ferrite beads, either the discrete ones or the multi wire common mode chokes isolate the 4-layer board from the ESD currents running through the TVS. David said:
-Since it has built in +-15kV protection, I presume I don't need any additional protection on the TX and RX lines? (I don't mind adding additional protection if necessary). Here you are not quite right. The built-in ESD protection is mostly against human body model ESD events only. So, additional protection is highly recommended. David said:
-What do they mean by 10/1000us or 8/20us surges? (Everyone quotes them, but no one defines them?) Has nothing to do with ESD events, but with surges, as they can be observed on lightnings etc. As they last considerably longer than ESD events, a considerably higher energy must be absorbed by the parts. These 10/1000µsec and 8/20µsec specifications tells, how much energy the part can absorb, but has nothing to do with ESDs. Take care, David, as Steve already mentioned, your TVS must withstand up to 100A, which is much higher than the peak current of human body model ESDs!! Eventually you must put some TVS in parallel or choose a higher rated variant. When paralleling, feeding all the wires running to the connectors through a common mode choke can slow down the rise time a bit and help to divide the partial currents equally to the individual TVS. All the TVS must be connected from each wire to chassis ground of your enclosure, right where the wires enter or leave the chassis! So, either you spend the TVS an extra board or you prepare a certain reserved area on your 4-layer board for the TVS parts only. Route the shortest traces possible for the TVS section. Take care even each millimeter counts!! All the connections must run on shortest paths to the enclosure (chassis ground), via wide wide traces (a local ground plane, also called chassis plane, prefered)! The connections from these TVS to the actual signal lines on 4-layer board should be made then by ferrite beads. But you can also feed several of the lines through a common mode choke. All the cables should enter or leave your enclosure at only one point!!! This prevents the ESD currents from flowing over the whole enclosure, and, of special interest, from partially flowing over your 4-layer board!! Connect the signal ground of 4-layer board to the enclosure (chassis ground) where all the TVS are connected to. But choose a quiet location for it. And, make the connection at only one point. Otherwise the ESD currents can again partially flow across your 4-layer board. How the ground lines are treated depends a bit on your application. Is your device a stand alone thingy? Or is it mounted in a second, bigger enclosure, as a module? It's easy to design an ESD robust device: Just think about how the ESD currents are flowing (mostly on a path to protection earth) and how these currents can be prevented from running over your 4-layer board. No ESD current must ever flow across your 4-layer board!!! Not even partially!!! Kai |
