SAFETY-30: grade each inverter against its own power, and the site's total against the meter

THE DEFECT, in one line. `inv.sign_wd.evaluate(stood, self.grid, ...)` compared
the command THIS controller put on ONE actuator against what the WHOLE SITE is
exchanging with the network. At a roster of one those are two views of the same
thing - specs.md §6.2's identity - and that is what makes the check sound and
what earned it the best record in the safety table (NOTES.md: 93 % agreement
while healthy against 4 % during runaway, four observed episodes, zero false
positives). At two inverters they are not, and it fails in BOTH directions:

  FALSE TRIP  inverter A commanded -1500 W while the meter reads +1500 W is
              evidence of nothing when inverter B is discharging 3000 W exactly
              as told. Nothing is faulty and the highest-value check fires.
  MISSED      the converse, and it is worse. A unit commanded +1500 W and
              MEASURED at -1500 W, beside a sibling at +1500 W, leaves P1
              exactly where a healthy site leaves it. The summation destroys the
              evidence and no threshold recovers it.

DEV-02 had already shipped the per-INSTANCE half - one SignWatchdog per
inverter, namespaced latch, a trip on B no longer latching A. What was left was
the COMPARISON, and it is two questions rather than one:

  PER UNIT  this unit's standing command against ITS OWN measured power
            (`Inverter.batt`, holding register 0x0518, per-instance since
            DEV-02). Conclusion: "the actuator I am driving is not doing what I
            told it" - a fact about one device, so the remedy stays the per-unit
            `failsafe` SET that arbiter.CLAIM_SCOPE has classified SCOPE_UNIT
            from the day it was written. That entry read "the evidence is
            site-wide and the remedy is not"; the first half is now simply true.
  SITE      the SUM of the standing commands against the meter, as a new
            `sign_site` SET at P_SAFETY on the site actuator, latched under flat
            `site_sign_tripped` / `site_sign_reason`. This is the 2026-08-20
            fault day and NO per-unit check can see it, because every inverter
            may be faithfully executing its share of a number that is itself
            wrong. It stops every inverter and the banner names the site.

WHAT IT IS NOT, said here because two guards that overlap without saying so is
how a false-alarm generator gets built. It is not SAFETY-31: circulation.py
compares the units AGAINST EACH OTHER and catches them pushing power THROUGH
each other, which nets to zero at the meter and about which no single command is
evidence. This compares each unit against ITS OWN COMMAND. They overlap in
exactly one scenario - a unit reversing while its sibling pulls the other way -
and test_watchdog.py asserts on that fixture that both fire and what each
concludes. It is also not the saturation freeze, which is a MAGNITUDE test
(|command - measured| > 500 W) that shapes the loop and fires on a merely
saturated unit; this is a DIRECTION test that latches and parks one.

THE FOUR NUMBERS, RE-DERIVED AT TWO INVERTERS - one carried with a justification
and one refused. 1890 closed-loop runs on admin/goodwe's TEST-19 two-master rig
in which NOTHING is faulty, swept over the whole range config.yaml accepts for
`gain` and `slew_w`, across load reversals with write staggers and flapping
loads, and RUN AT BOTH ROSTERS so the N=1 arm is the control:

  unit_sign_disagree_w  250   `saturation_w / 2`. The check gates on both sides,
                              opposed, so the smallest gap it can see is 500 W -
                              exactly where compute()'s own anti-windup test
                              already says the inverter is not following. It
                              cannot fire on a divergence the control law calls
                              ordinary tracking. Worst legitimate opposition at
                              250 W: 10.0 s; the curve is nearly flat, because a
                              unit's own power chases its own command.
  unit_sign_disagree_s   85   8.5x that 10.0 s, 2.8x the firmware's own 30 s
                              link veto (inverter.LINK_STALE_S), under half the
                              180 s site hold.
  site_sign_disagree_w  300   CARRIED. Its argument is the meter's noise floor
                              and the 15 W deadband; the roster touches neither,
                              and the campaign agrees - the gate opens in 994 of
                              1050 runs at two inverters and 955 of 1050 at one.
  site_sign_disagree_s  325   NOT CARRIED, and this is the number the card
                              exists to not assume. The worst legitimate
                              crossing at 300 W is 110 s with one inverter and
                              215 s with TWO, because the command must travel a
                              rail that is the roster sum at a `slew_w` that is
                              not - reachable at gain 0.05, slew_w 50 and a
                              2.2 kW load reversal, all schema-legal. 180 s
                              would have shipped a false trip. 325 s is 1.51x
                              the measured worst, inside the 900 s billing
                              quarter and inside every episode of the fault day.

AT ONE INVERTER NOTHING CHANGES, and "nothing" is asserted rather than assumed.
The two questions are the same sentence there, so only the detector that shipped
is armed: same comparison against the meter, same 300 W / 180 s, same reason
string byte for byte, same namespaced latch and flat rollback mirror, same
`failsafe` claim, same banner, same single status row - and the site detector is
NEVER CALLED, counted on the object over twelve cycles.

VERIFIED BY BREAKING IT. Thirteen mutations on fresh per-mutation trees with no
__pycache__, mounted :ro, with the battery refusing to report at all unless the
baseline is green. Twelve red on the first pass. M4 - the site half armed at N=1
as well - SURVIVED, because the check read `disagree_since is None` and the unit
detector trips first, parks the inverter and leaves the site standing at 0 W:
the assertion ran, the sentence was true, and the data could not reach the
failure it named. Replaced with a call counter on the object; M4 is red now.
Recorded as a survivor in DECISIONS/SAFETY-30-per-unit-sign.md §5, with the
first battery's own RED BASELINE - a shared mutation directory read mid-copy -
discarded rather than reported.

RECORDED, NOT FIXED: the site comparison is defeated by a square-wave load whose
half period is about one control cycle, which holds it opposed for ~1000 s. The
N=1 control arm is what makes that interpretable - it is identical at the roster
this add-on ships, so it is a property of comparing a lagged command against a
meter and has nothing to do with the second inverter. It wants an alternation
test rather than a number, and a card of its own.

SAFETY-24's lo-only asymmetry proof re-run unchanged: 1296 combinations, 0
negative floors, `hi` +inf on every claim, every claim on the SITE actuator.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BwRLw7eKYj2byVdwB6kbjU

goodwe_controller/DOCS.md

@@ -287,6 +287,10 @@ behind it would trip the firmware watchdog on a system that is working fine.
 | `auto_start` | false | Start controlling on boot (only after commissioning) |
 | `sign_disagree_w` | 300 | Both the command and the meter must exceed this before a sign difference counts. Must stay **below `max_w`** or the watchdog can never fire, and the add-on refuses to start |
 | `sign_disagree_s` | 180 | How long they must stay opposed, continuously, before the failsafe trips |
+| `unit_sign_disagree_w` | 250 | **Two or more inverters only.** Both a unit's own share and its own measured power must exceed this before their sign difference counts. It is `saturation_w / 2`, so the smallest divergence it can see is exactly the gap at which the loop already calls the inverter "not following". Must stay **below `max_w`** or it can never fire, and the add-on refuses to start |
+| `unit_sign_disagree_s` | 85 | **Two or more inverters only.** How long one unit's command and its own power must stay opposed before that unit is parked |
+| `site_sign_disagree_w` | 300 | **Two or more inverters only.** Threshold for the site-wide check - the sum of the commands against the meter. The same 300 W as `sign_disagree_w`, for the same reason: near zero the sign is noise |
+| `site_sign_disagree_s` | 325 | **Two or more inverters only.** How long the site's total command and the meter must stay opposed before every inverter is parked. Longer than `sign_disagree_s` because the command has to cross zero over a rail that is the sum of the roster, at a `slew_w` that is not - measured at 110 s with one inverter and 215 s with two |
 
 #### The sign-disagreement failsafe
 
@@ -298,6 +302,43 @@ there. The banner reads `FAILSAFE`, the commissioning list says which two
 numbers disagreed and for how long, and the `Controller status` entity reports
 `failsafe`.
 
+#### ⚠️ At two or more inverters this is TWO checks, not one (SAFETY-30)
+
+The paragraph above describes what a **one-inverter** install does, and it is
+unchanged: there is one meter, one actuator, and comparing the command against
+the meter is comparing a thing with a view of itself.
+
+With a second inverter that stops being true. The meter answers for **both**
+units, so a command put on one of them can be opposed to it with nothing wrong
+(inverter A charging while inverter B discharges more), and - the worse
+direction - a unit genuinely driving the wrong way can be **hidden**, because its
+sibling's watts arrive at the meter and the sum looks healthy. So the check
+splits in two, and they are different faults:
+
+- **Per inverter** — that unit's own share against **its own measured power**
+  (the `batt_entity` reading, register 0x0518). This catches *"this inverter is
+  not doing what it was told"*. Only that inverter is parked; the site keeps
+  running on the one that works. Thresholds: `unit_sign_disagree_w` /
+  `unit_sign_disagree_s`.
+- **Site-wide** — the **sum** of what every inverter has standing, against the
+  meter. This catches *"the site as a whole is pushing the wrong way"*, which no
+  per-inverter check can see because every unit may be faithfully executing its
+  share of a number that is itself wrong. **Every** inverter is parked and the
+  banner names the site rather than an inverter. Thresholds:
+  `site_sign_disagree_w` / `site_sign_disagree_s`, latched under flat
+  `site_sign_tripped` in `/data/state.json`.
+
+Both latch, both are cleared by the same *stop, then start*, and both survive a
+restart. **On a one-inverter install neither pair does anything at all** - the
+two questions are the same question there, and only the check described above is
+armed.
+
+This is **not** the circulating-power alarm further down: that one compares the
+inverters *against each other* and catches them pushing power through each other,
+which nets to zero at the meter and which no command is evidence about. It is
+also not the saturation freeze, which is a *magnitude* test that shapes the loop
+rather than a *direction* test that parks an inverter.
+
 **It does not clear itself, and restarting the add-on does not clear it either.**
 Pressing *Start / stop control* twice — stop, then start — is the only thing that
 does. That is deliberate on both counts: