Monday, June 17, 2013

My Experience With "Noise": Part IVd

We were discussing about surge voltage transient and how to make our designs more and more immune to this kind of disturbances...we are not done yet.

The picture below shows a circuit how a TVS doide is connected one one input port, which shows the parasitic inductors L1, L2, L5 & L6 seen by the physical copper traces on the board when a high frequency transient V1 is applied to the input. Please note that the equivalent load at the input port is represented by a capacitor and resistor in parallel: C1 & R1. Also note that the signal ground and frame ground are usually separate and usually connected together by a capacitor (usually having high voltage rating such as 2KV) and high value resistance (~10M ohm); that is what is represented by C2 and R2 in the Figure 1 below. Please note that V1 represents a surge transient source and that is why it is referenced to "FRAME GND" not the signal ground. Also it is to be noted that the TVS device (represented by D1 in Figure 1) or any other suitable transient suppressor device shall always be connected between the line it intends to protect and always the "FRAME GND" not the signal ground (assuming these two ground connections are not the same). 

Figure 1: Surge suppressor & parasitic inductance
Each piece of copper trace on the PCB imposes some inductance to a "high frequency" transient current, caused by ESD, EFT or Surge. Since the objective of the transient suppressing device D1 here is to divert the transient energy back to the source from right at the port entry, the protection device shall ideally placed right between the port connector pins: signal under protection and frame ground. But it is not always practically possible to do the same. So care must be taken to make the traces to TVS as short and wide as possible. The intention here is to reduce the values of the parasitic inductors L1, L5 and L6 to "practical" zeros. The trace represented by L2 could have a longer length and theoretically it should be better to have a bit of inductance on that trace (L2) to offer impedance to any transient noise tries to propagate on its path to the device under protection (but if it is permitted for its correct functioning).

We will discuss this further when we will take a look at the some layout design techniques. Next topic I would like to pick would be "conducted RF".    

No comments: