ECSS Electrical Engineering — EMC Control Plan Contents (space-systems/ecss/e20-emc-control-plan-contents)
Use when the task is the content of the electromagnetic-compatibility control plan described by the ECSS-E-ST-20C Annex A data-requirement -- who owns compatibility control, which design measures are being applied, what electromagnetic budgets those measures have to buy, and when each piece of that work lands against the programme reviews. The check is on the plan document itself, not on the delivery date of the plan and not on the verification activities it points to.
Domain quick reference
- Six content blocks make up the plan: organisation-and-responsibility, control-measure-register, electromagnetic-budget-table, verification-approach-reference, schedule-and-milestone-list and deviation-handling-route. A block that is present as a heading but carries no content counts as undeclared -- an empty section is not a content block, and reading it as one lets an unplanned area pass the review.
- The organisation block has to name three distinct things: the engineer or authority that owns compatibility control, the reporting path by which findings reach project management, and the interface to the board that arbitrates changes. A delegate named without a scope is a finding, because an unscoped delegation cannot be checked against the work breakdown.
- The control-measure register pairs an interference mechanism with the measure that controls it. Mechanisms split into conducted-emission, conducted-susceptibility, radiated-emission, radiated-susceptibility, electrostatic-discharge and lightning-induced-transient; measures into a grounding-scheme, a bonding-scheme, a shielding-scheme, cable-category-segregation, a filtering-scheme and a galvanic-isolation-scheme. A measure only buys margin on the coupling paths it physically acts on: filtering and galvanic isolation act on the conducted paths, shielding, bonding and cable-category-segregation act on the radiated paths, bonding and grounding carry the discharge path.
- Levels in the budget table are in dB, so several emitters seen by one victim combine by a power sum (10 log10 of the summed linear powers), not by arithmetic addition. Two equal emitters sit about 3 dB above one of them, never twice the level. The margin a victim holds is its susceptibility level minus that aggregate, compared against the margin the programme requires.
- Schedule entries anchor to a named review -- preliminary-design-review, critical-design-review, qualification-review, acceptance-review or flight-readiness-review -- and carry the lead in days by which the milestone precedes it. A negative lead means the work lands after the review it is meant to feed.
Workflow
- Resolve every declared block name to its canonical token and reject a block the data-requirement does not recognise; then list the required blocks the plan leaves undeclared or empty.
- Read the organisation block and record a finding for each of the three mandatory roles it does not name, and for every delegate declared without a scope.
- Walk the control-measure register. Resolve the mechanism and the measure of each entry, record an unnamed owner, and reject any pairing whose measure does not act on that mechanism's coupling path -- a rejected pairing does not count as coverage.
- After the register walk, list every mechanism in scope that no surviving entry controls. Narrow the scope list explicitly if the programme has argued a mechanism away; never let an absent entry imply an absent mechanism.
- For each budget row, power-sum the emitter levels, subtract the aggregate from the victim's susceptibility level, and compare the margin against the required one. Absorb float representation error at an exactly met requirement rather than lowering the required margin.
- Report the worst margin and its victim so the table has a single number the review can carry, alongside the full breach list.
- Check each milestone anchors to a known review and leads it by at least the minimum lead, and that the reviews the plan must serve all carry at least one milestone.
- The plan is acceptable only when no block, role, mechanism, budget row or milestone finding remains open.
Pitfalls
- Adding emitter levels in dB. Levels are logarithmic; two 40 dB emitters aggregate to about 43 dB, and summing them to 80 dB turns a compliant budget into a fictional breach that costs real shielding mass.
- Counting an empty heading as a declared block. The plan then reads as complete while an entire control area -- most often the deviation-handling-route -- has no owner at all.
- Accepting a register entry that pairs a mechanism with a measure that does not act on it, such as a filtering-scheme offered against a radiated path. The entry looks like coverage in a table count but buys nothing, and it hides the mechanism that is genuinely uncontrolled.
- Treating the absence of a register entry as evidence that the mechanism does not apply. Scope is declared, not inferred; a mechanism argued away belongs in the scope list with its argument, not silently missing.
- Confusing the control plan with the verification plan. Annex A content is the control side -- organisation, measures, budgets, schedule -- and it points at the verification approach rather than restating its activities.
- Letting a milestone carry a negative lead because the anchor slipped. A milestone that lands after the review it feeds leaves that review with no compatibility evidence, which is a schedule finding, not a rounding detail.
Behavior contract (gate 3)
The block-presence rule, the organisation-role check, the mechanism-to-measure effectiveness rule, the dB power-sum aggregation, the budget-margin comparison and the milestone-lead check are exercised by the gate 3 contract test: scripts/test_e20_emc_control_plan_contents.py against scripts/e20_emc_control_plan_contents_logic.py (stdlib unittest, offline). Run: python3 scripts/test_e20_emc_control_plan_contents.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
