Why does sample rate matter, and what goes wrong if it's too low?

The sample rate determines the fastest signal feature an instrument can faithfully capture, and if it's too low you get aliasing and missed events. By the Nyquist principle you must sample well above the highest frequency component of interest, in practice several times the fastest edge or bit rate, not just twice, to reconstruct edges and catch detail, because at the boundary you can't resolve the waveform's shape or timing. If you undersample, two things go wrong: aliasing, where a fast signal appears as a spurious lower-frequency signal that looks real but isn't (the same folding problem as in ADC sampling), leading you to misread timing; and, critically for debugging, you miss narrow glitches and runts entirely, a brief spike or a too-short clock pulse that falls between samples simply isn't recorded, so the very anomaly you're hunting is invisible. Catching glitches requires both a high sample rate (to have a sample land on the narrow event) and sufficient memory depth (to keep a long enough capture at that rate to include the rare event), which is why instruments advertise both sample rate and memory depth and why there's a tradeoff between time window and resolution. On a logic analyzer there's also the threshold and setup/hold consideration, and on a scope the analog bandwidth (front-end) is a separate limit from the sample rate, a low-bandwidth scope rounds off fast edges even if it samples quickly. So you choose an instrument and settings whose sample rate and bandwidth comfortably exceed your signal, and for glitch-hunting you ensure enough sample rate and memory to actually capture the rare narrow event.