Short-Circuit Energy Component

     Short-Circuit current on the direct path from power supply Vdd to the ground exists while p-MOS and n-MOS transistors are both conducting. This is the case during the output switching time, both charging and discharging. 
Consequently, energy dissipation due to the short-circuit energy component is calculated by multiplying the energy of one transition with the output switching activity factor.
       Energy associated with short-circuit component can be measured by integrating the current of the n-mos transistor during the low-to-high output transition, and vice versa by measuring the p-mos current while the n-mos is discharging the output to low value. The current due to the capacitive coupling is isolated and taken into account as well.
       The first solution of the short-circuit energy component dissipated in the inverted was given by [Veendrick 84]. More accurate model was presented by [Hedenstierna and Jeppson 87]. Both models were based on the square-low MOS equations. [Sakurai and Newton 90] introduced alpha-power law MOS model and derived expression for short-circuit energy dissipation of the inverter under the same assumptions as in [Veendrick 84].

        Design considerations for minimizing the short-circuit energy component are given in the following examples.

Verba docent, exampla trahunt

     In order to illustrate the effect of input/output transition times, i.e. input/output slopes, and output load on the short-circuit energy dissipation of the inverter gate under test, the setup shown in Figure 1 is used.

Figure 1. Circuit setup

By varying output and input fanouts, h and q, respectively, the measured short-circuit energy component is illustrated in Figure 2.
The maximum of the short-circuit energy dissipation occurs when the input fanout is big (q), whereas the fanout at the output is small (h).

Figure 2. Short-Circuit Energy dependence on input/output fanouts (q and h, respectively)
 

With a certain amount of latitude remaining, the rule for minimization of associated short-circuit energy component would be right-hand side of the polyhedron. The concluding remarks of the analysis given in [Veendrick 84] state that if the output and input signal have equal transition times, the short-circuit will be only a fraction (<20%) of the total dissipation and to minimize the short-circuit energy dissipation an inverter should be designed in a way that the input transition times are less than or about equal to the output transition times.
The proposed optimization corresponds to the setup case shown in Figure 1 and having  the equal slopes at  the input and output when q value is set to be equal to h. The results are summarized in the following graphs, Figures 3, 4.

Figure 3. Short-Circuit Energy Dissipation for equal input/output fanouts (h=q)

Figure 4. Percentage of Esc in total energy dissipation, Etot, versus h (h=q).

References: