ECSS Solar Cells -- Diode Human Body ESD (space-systems/ecss/e2008-diode-human-body-esd)
Use when the task is clause 9.6.16 of ECSS-E-ST-20-08C -- whether a protection diode survives the discharge a charged person delivers when a finger or a hand tool reaches one of its leads. A panel bay is full of people and insulating surfaces, so this is not an exotic event: it is the ordinary one, and the campaign that measures it has to look like the person rather than like the bench.
Domain quick reference
- The discharge network is the whole claim. A capacitance charged to the step voltage, emptied through a series resistance into the device, is what makes the pulse a person. A network outside its bands delivers a different event carrying this event's name, and every number derived downstream then describes that other event instead.
- Three quantities follow from the network and the step: the peak current is the step divided by the series resistance, the decay constant is the resistance times the capacitance, and the stored energy is what the junction has to absorb. None of them is read off an instrument; all three are derived, so a network check comes first.
- Stress is a ladder, not a single shot. Each level sits a fixed ratio above the last, and the level the part last held is the withstand. A ladder that climbs too gently ends below the level of interest and reports a survival that was never tested for.
- A level is several pulses, in both polarities. One pulse measures luck. One polarity leaves a conduction path of the diode entirely unstressed, which for a protection diode is the path that matters.
- Pulses need a recovery interval between them. Fired back to back they stack a thermal excursion the model does not contain.
- Survival is not the same as still conducting. Discharge damage shows up first as reverse leakage that grew and, later, as a forward voltage that moved. Both are taken before and after the series and both carry a limit.
- A withstand level is grouped into a band so parts can be compared. The band is a reporting convenience; the pass or fail is against the policy floor, not against the band edge.
Workflow
- Validate the stress policy first: withstand floor, pulses per level, ladder levels and ratio, recovery interval, network capacitance and resistance bands, and the two drift limits. A band whose lower edge sits above its upper edge is refused rather than used, and a ladder ratio of one is refused because it never rises.
- Check the declared network against its bands before anything else. Out of band, the campaign is reported as network deficient and the survival number is not defended.
- Build the ladder from the start voltage, ratio and level count, and derive the peak current, decay constant, transferred charge and stored energy at the surviving level.
- Take the reverse leakage and forward voltage drift across the series as fractions of their starting values, and hold each against its limit. A drift landing exactly on a limit passes; the comparison tolerance absorbs representation error and the limit does not move.
- Check the plan shape: pulses per level, polarity coverage, ladder depth and ratio, recovery interval. Report every finding, not the first.
- Close on one verdict, the network outranking the rest because it decides whether the campaign measured this event at all: network deficient, withstand deficient, stress plan deficient, or survival accepted.
Pitfalls
- Reporting a withstand voltage from a network nobody checked. The series resistance sets the peak current directly, so a resistance a factor low turns a modest step into a stress the model never intended, and the reported level belongs to no model at all.
- Treating the capacitance as the only network parameter. Capacitance sets the energy, resistance sets the peak current and together they set the decay; two of the three are wrong if either is out of band.
- Firing one pulse per level. Discharge damage is partly statistical, and a single pulse that missed the weak path returns a pass that the next unit will not repeat.
- Stressing one polarity. A protection diode conducts one way and blocks the other, and the blocking path is the one an unplanned discharge finds. A single-polarity campaign leaves it unmeasured.
- Climbing the ladder too gently to save time. A shallow ratio with few levels tops out below the floor, and the campaign then reports the top of its own ladder as the part's withstand.
- Declaring survival because the part still conducts. A junction with leakage several times its starting value is damaged; it simply has not finished failing yet, and the drift limits are what catch it.
- Comparing a derived drift or energy against a limit by bare arithmetic. These come out of products and divisions that land a few units in the last place either side of a limit on different hosts, so the comparison absorbs that error while the limit itself is never relaxed.
Behavior contract (gate 3)
The policy validation, discharge network band check, peak current, decay constant, transferred charge and stored energy derivations, the geometric step ladder, the withstand band grouping, the leakage and forward voltage drift fractions, the polarity coverage and plan shape checks, and the survival verdict are exercised by the gate 3 contract test: scripts/test_e2008_diode_human_body_esd.py against scripts/e2008_diode_human_body_esd_logic.py (stdlib unittest, offline). Run: python3 scripts/test_e2008_diode_human_body_esd.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
