The sample rate in Example 6.15 was selected to be T_s = 0.05 \text{ sec}. We can see from Fig. 4.22 that the effect of the sampling is to hold the application of the control over one sample period, thus the actual delay varies between zero and one full sample period. Therefore, on the average, the effect of the sampling is to inject a time delay of T_s/2 = 0.05/2 = 0.025 = T_d \text{ sec}. From Eq. (6.71), we see that the phase lag due to this sampling at the crossover frequency of 5 \text{ rad/sec }, where we measure the PM, is ∠G_D = −ωT_d = −(5)(0.025) = −0.125 \text{ rad } = −7°. Therefore, the PM will decrease from 45° for the continuous implementation to 38° for the digital implementation. Fig. 6.59(a) shows that the overshoot, M_p, degraded from 1.2 for the continuous case to ≈1.27 for the digital case, which is predicted by Eq. (6.32) and Fig. 6.38.

∠G_D(jω) = −ωT_d           (6.71)

ζ ≅ \frac{PM}{100} .           (6.32)

In order to limit the phase lag to 20° at ω = 5 \text{ rad/sec}, we see from Eq.(6.71) that the maximum tolerable T_d = 20/(5 ∗ 57.3) = 0.07 \text{ sec}, so that the slowest sampling acceptable would be T_s = 0.14 \text{ sec}. Note, however, that this large decrease in the PM would result in the overshoot increasing from ≈20% to ≈40%.