2.18 Parallel (Shunt) Terminations

In high speed circuits, parallel transmission line terminations can be useful. This termination is placed at the end of the transmission line as shown in Figure 2.15.

Figure 2.15 A series (source) terminated transmission line.

Since there is no reflection at the ends of the lines, line current continues to flow as long as there is a signal. Because of the extra power consumption, this technique is usually limited to low voltage logic. In general, this logic design approach has a higher noise margin than the series termination case.

In series terminated circuits, the clock signals are usually timed so that logic signals have time to reflect and return to the source. This timing guarantees that all logic connections made along the transmission path will receive the doubled voltage before the clock signal arrives.

On long lines with all the loads at the end of the lines, parallel terminations can be effective. The clock signal can arrive anytime after the logic signal reaches the termination. On a long line, compensation for line loss can be made by increasing the value of the terminating resistor. The reflection that results adds to the signal, thus, reducing the chances of error. Parallel termination has the drawback that energy must be supplied to the line as long as the logic signals are present. If the transmissions are infrequent and the logic levels are at zero, most of the time, this can be a very effective approach.

When an external cable brings a balanced logic signal to a circuit board, the cable must be terminated in its characteristic impedance. If the cable impedance is 100 ohm and if the individual trace impedances are 50 ohm then the cable impedance is matched at the connection. In this approach, the traces must be routed independently (spaced apart) and connected to the proper logic pins. Each trace is then properly terminated (typically 50 ohm). If the cable conductors are connected to a 100-ohm transmission line on the board, the cable would in effect be double terminated. This problem is discussed at the end of Section 3.5.

The parallel termination of stripline poses a special problem. Assume that the stripline is between a ground and power plane. The transmission line energy can be thought of as traveling in two paths. The first path is between the strip line and the ground plane, and the second path is between the transmission line and the power plane. The correct approach for terminating symmetric stripline is to terminate each energy path separately. If the characteristic impedance is 50 ohm then the correct parallel termination is one 100 ohm resistor to the ground plane and one to the power plane. These connections would be at the power and ground pins of the receiving logic. With this approach, it is easy to see where the return current flows.

If one 50-ohm terminating resistor is used, there are several problems. The 50-ohm resistor is a mismatch for just one of the transmission waves. This means that there will be a reflection at the resistor on this half of the stripline transmission. The second energy path sees no terminating resistor. Some of the energy at this mismatch will propagate and reflect into the ground/power plane. The energy will also transmit and reflect at any nearby decoupling capacitors. It is easy to see that a single parallel terminating resistor can cause many unexpected reflections. Problems may not occur if the rise times are greater than 1 ns.

If the stripline is run between two ground planes then two parallel terminating resistors are still required. If one resistor is used, the reflections will involve connecting vias instead of decoupling capacitors.