feedback

What it is ¶

feedback(net) yields one row per net whose name reads as a regulator feedback or sense node: a leaf name ending in _FB, _VFB, _FEEDBACK, _VSENSE, _SENSE, _SNS, or a bare FB / VFB. It is name-derived, the datalog face of the feedback exclusion the test-point rule applies.

For hardware engineers ¶

A feedback net is the divider tap that feeds a switching or linear regulator’s control loop. It is a high-impedance sense node, so anything you hang on it changes the voltage the regulator reads and shifts regulation. That is why a scope probe or a test point on a feedback node is a review finding: the probe’s capacitance and loading disturb the loop it is measuring. During a review you query feedback to list the sense nodes, and you subtract it from rail so a probe-point or pull-up rule does not treat a sense tap as an ordinary rail.

For software engineers ¶

Think of a feedback net as a node you may read but must not tap: observing it changes its value, so it is off-limits to the instrumentation a normal rail allows. feedback is a filtered projection over Nets() with the naming predicate, so rows are 1:1 with feedback-named nets, and an empty result means no net name matched the sense-node lexicon.

Go projector ¶

feedbackFacts in check/facts.go walks Model.Nets() and emits a row for each net where isFeedbackName(name) holds. isFeedbackName (in check/rolenames.go) delegates to the active naming lexicon’s IsFeedback, matching the _FB / sense-suffix patterns on the hierarchy leaf, case-insensitive. One row per feedback-named net; empty when no net matches.

Datalog ¶

List every feedback / sense node:

feedback(?n) => ?n

A supply rail that is not a sense node (the rails a probe or pull-up rule may treat as ordinary distribution):

rail(?n), not feedback(?n) => ?n

Schematic ¶