The z-domain map is z = e^(sT), so a stable continuous pole s = −1 at T = 0.1 s lands at z = 0.904837 — exactly what the 1/(s+1) row shows. Zero-order hold samples the continuous response at the boundaries and matches the step response exactly; Tustin substitutes s = (2/T)·(z−1)/(z+1), which maps the whole left half-plane inside the unit circle but warps frequency unless you pre-warp; matched pole-zero places each pole at e^(sT) and matches gain. Forward Euler maps s = (z−1)/T and can push a fast pole outside the circle for a too-large T; backward Euler always keeps stable poles stable, at the cost of extra distortion. The sampling advice packages the rule: dt below about 1/(15·ω<sub>max</sub>) is comfortable, and anything above 1/(5·ω<sub>max</sub>) is flagged.
Two design routes then open. Pole placement picks the closed-loop dynamics directly: for 1/(s²+0.6s+1) the gain [4.4, 5.0] moves both eigenvalues to −2 and −3 with zero residual error, because Ackermann's formula is exact for a controllable single-input system. LQR chooses the dynamics for you by minimising a quadratic cost; with equal penalties it returns [0.879333, 0.414214] and a complex pair at −0.74 ± 0.93j — slower poles than placement but with a guaranteed margin and no pole list to invent. Both gains act on the State Space tab's controllable-canonical realisation, A = [[−0.6, −1], [1, 0]], B = [1, 0]ᵀ, whose states are ordered x₁ = ẏ and x₂ = y, so K multiplies [ẏ, y]ᵀ; with the order reversed to [y, ẏ] the same gains read [5.0, 4.4] and [0.414214, 0.879333]. Tuning covers the industrial shortcut: read the ultimate gain and period, apply a rule, then judge by simulation. Ziegler–Nichols is aggressive — 44.9 % overshoot on the third-order plant — while IMC with λ = 1 is smooth at 0 % overshoot but takes 2.66 s to rise instead of 0.43 s.