Guide · controlled isolation

A direct known-good control is the shortest useful test.

Rate fallback and intermittent disconnects can look alike in applications but require different evidence.

When the USB4 class is below target

First distinguish a negotiated link class from application throughput. Confirm the exact host port, device, and cable declarations. Remove any dock, hub, monitor, or adapter and use a short known-good cable. On supported Windows systems, USB4 settings can list attached USB4 hubs and devices and show capabilities.

If direct reaches target and the intermediary path does not, the intermediary layer is implicated. If direct is also below target, return to the host, cable, device, power, and firmware baseline.

When devices disconnect

  1. Reproduce a defined trigger: heavy transfer, adding a device, display load, or idle/wake.
  2. Record whether one peripheral or every downstream function drops.
  3. Compare direct and hub paths.
  4. Add external power, reduce simultaneous load, and change to a known-good cable—one variable per run.
  5. For idle/wake failures, include operating-system power state and resume behavior in the evidence.

USB selective suspend can suspend individual ports without affecting others and is enabled by default in Windows. That makes idle timing useful evidence, not proof that selective suspend is the root cause.

Sources

Isolate a USB4 fallback · Rank dropout evidence →