diff --git a/CHANGELOG.md b/CHANGELOG.md
index 92d3d27..7e9191e 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,11 @@ Driver versions follow [Semantic Versioning](https://semver.org/spec/v2.0.0.html
 
 ## [Unreleased]
 
+### Fixed
+- **sungrow** 1.5.6 — **the zero-power refusal is restored. 1.5.4 was wrong, on a premise that did not survive reading the driver.** The case for waving zero through was that it hands the device back to itself, so refusing it could strand an inverter in a forced state. Neither half holds. `set_battery_idle` writes EMS mode **2** — forced — with command `0xCC` and setpoint 0: it pins the battery at zero under the host's control rather than releasing it. The release is `driver_default_mode`, which FTW reaches through `SendDefault` on shutdown, lease expiry and the watchdog, never through `driver_command`, and which no guard touches. So refusing withholds nothing that is not reachable by a route that cannot refuse — while accepting put mode 2 and a setpoint back on an SG inverter that implements neither, and reported success, because the read-back that would catch it fails on that device too. That is the SG12RT bug through the one action nobody guarded
+- The two sungrow copies disagree about what zero writes — the catalog driver forces idle at mode 2, the FTW v2 target hands the inverter back at mode 0 and clears the lease — and they use different register maps throughout. That belongs to the register-map review `packages/v1/sungrow/PILOT.md` already requires before the v2 target ships; its `verification_status` is `experimental` and the signed package builds from `targets/ftw-observe.lua`, so nothing there reaches hardware today. Recorded because untested drift between these two files is how Pixii's 40288 came back
+- **FTW v2 target** 1.3.4 → 1.3.5 — `driver_command_v2`'s matching zero exemption is removed in the same commit. Raised so a package build cannot claim bytes under a version number that was never assigned to them
+
 ### Fixed
 - **sungrow** 1.5.5 — **Startup asked a family with no battery about its battery limits.** `configure_power_limits` reads 33046, 33047 and 13057 — charge power, discharge power, and the SoC ceiling and floor — and raises the first two when the unit shipped capped. A model that named itself a string inverter answers none of them, so an SG12RT paid three failed reads on every restart to learn nothing. It now skips them. These run once rather than on every poll, so this is not the SG12RT outage returning; the reason to gate them is that they are free to gate. Each write is already conditional on its read having answered, and none of them clears a forced state
 - **The wider version of this change was wrong, and the test that says so ships with it.** The first draft gated `set_self_consumption` on the same family test, covering `driver_init`'s startup reset, the watchdog, `driver_cleanup` and `deinit` — three writes at startup and three on every watchdog tick, aimed at a block the family has no use for. But since 1.5.4 a zero-watt command is accepted whatever the family, and it writes EMS mode 2. So "the device named itself a string inverter" is not proof there is nothing to release: Sungrow shipping a hybrid under a device-type code `classify_device_type` has not been taught lands in exactly that branch, takes the forced write, and is then left with no path that writes mode 0 back. Measured: mode 2 survived the watchdog and cleanup both. `test_the_release_is_still_reachable_on_a_string_inverter` fails if the gate is put back
diff --git a/SUPPORT_STATUS.md b/SUPPORT_STATUS.md
index f24c0a7..c577b4b 100644
--- a/SUPPORT_STATUS.md
+++ b/SUPPORT_STATUS.md
@@ -144,8 +144,8 @@ Catalog source is not proof that a target can install or run a driver.
 | solis_string | 1.1.1 | blixt-l1 | not_assessed | — | — | not_recorded | — | not_assessed | no |
 | sonnen | 2.0.1 | ftw-core | not_assessed | — | — | not_recorded | — | not_assessed | no |
 | sonnen | 2.0.1 | blixt-l1 | not_assessed | — | — | not_recorded | — | not_assessed | no |
-| sungrow | 1.5.5 | ftw-core | observe_only_profile_blocked | 1.3.3 | 1.3.0 | exact_model_firmware_required | — | not_assessed | no |
-| sungrow | 1.5.5 | blixt-l1 | not_assessed | — | — | not_recorded | — | not_assessed | no |
+| sungrow | 1.5.6 | ftw-core | observe_only_profile_blocked | 1.3.3 | 1.3.0 | exact_model_firmware_required | — | not_assessed | no |
+| sungrow | 1.5.6 | blixt-l1 | not_assessed | — | — | not_recorded | — | not_assessed | no |
 | tesla_vehicle | 0.2.0 | ftw-core | not_assessed | — | — | not_recorded | — | not_assessed | no |
 | tesla_vehicle | 0.2.0 | blixt-l1 | not_assessed | — | — | not_recorded | — | not_assessed | no |
 | tibber | 1.1.0 | ftw-core | not_assessed | — | — | not_recorded | — | not_assessed | no |
diff --git a/devices.yaml b/devices.yaml
index b13e7ce..a2d101b 100644
--- a/devices.yaml
+++ b/devices.yaml
@@ -1722,7 +1722,7 @@ manufacturers:
         protocols:
           - protocol: modbus
             driver: "sungrow"
-            version: "1.5.5"
+            version: "1.5.6"
             ders: [battery, meter, pv]
             control: true
         firmware_versions: ""
@@ -1733,7 +1733,7 @@ manufacturers:
         protocols:
           - protocol: modbus
             driver: "sungrow"
-            version: "1.5.5"
+            version: "1.5.6"
             ders: [battery, meter, pv]
             control: true
         firmware_versions: ""
@@ -1744,7 +1744,7 @@ manufacturers:
         protocols:
           - protocol: modbus
             driver: "sungrow"
-            version: "1.5.5"
+            version: "1.5.6"
             ders: [battery, meter, pv]
             control: true
         firmware_versions: ""
@@ -1755,7 +1755,7 @@ manufacturers:
         protocols:
           - protocol: modbus
             driver: "sungrow"
-            version: "1.5.5"
+            version: "1.5.6"
             ders: [battery, meter, pv]
             control: true
         firmware_versions: ""
@@ -1766,7 +1766,7 @@ manufacturers:
         protocols:
           - protocol: modbus
             driver: "sungrow"
-            version: "1.5.5"
+            version: "1.5.6"
             ders: [battery, meter, pv]
             control: true
         firmware_versions: ""
diff --git a/drivers/lua/sungrow.lua b/drivers/lua/sungrow.lua
index c33036d..c049054 100644
--- a/drivers/lua/sungrow.lua
+++ b/drivers/lua/sungrow.lua
@@ -9,7 +9,7 @@ DRIVER = {
   id           = "sungrow-shx",
   name         = "Sungrow SH Hybrid Inverter",
   manufacturer = "Sungrow",
-  version      = "1.5.5",
+  version      = "1.5.6",
   protocols    = { "modbus" },
   capabilities = { "meter", "pv", "battery", "pv-curtail" },
   description  = "Sungrow SH-series hybrid inverters with LFP battery, via Modbus TCP.",
@@ -139,13 +139,13 @@ end
 -- Whether the device has named itself a model with no battery registers.
 --
 -- Read this narrowly. It gates startup *reads* of the battery limit block and
--- nothing else. It must never gate a write that clears a forced state: since
--- 1.5.4 a zero-watt command is accepted on any family, so a device this
--- function calls "string" can still be holding EMS mode 2, and
--- set_self_consumption is the only thing that writes mode 0 back. A device
--- classified here is not proof there is nothing to release -- Sungrow shipping
--- a hybrid under a device-type family classify_device_type has not been taught
--- lands in exactly this branch.
+-- nothing else. It must never gate a write that clears a forced state, because
+-- a device this function calls "string" can still be holding EMS mode 2 --
+-- left there by an earlier run, or by a firmware that answers the type
+-- register differently after an update -- and set_self_consumption is the only
+-- thing that writes mode 0 back. Being classified here is not proof there is
+-- nothing to release: a hybrid shipped under a device-type family
+-- classify_device_type has not been taught lands in exactly this branch.
 local function known_to_have_no_battery()
     return model_family == "string"
 end
@@ -747,20 +747,7 @@ function driver_command(action, power_w, cmd)
     if action == "init" then
         return true
     elseif action == "battery" then
-        -- Zero is not a dispatch. It is the host handing the device back to
-        -- itself: forced mode off, setpoint nought. Refusing to write "stop"
-        -- is a different risk from refusing to write "charge" -- a device
-        -- left in a forced state stays there, and the safe default is the
-        -- one path that must never be gated on how much we know about the
-        -- hardware. It also arrives before the first poll, from the
-        -- lifecycle rather than from the planner, which is exactly when
-        -- nothing has been confirmed yet.
-        --
-        -- On a genuine string inverter this writes to a register block the
-        -- model does not implement. That write fails at the Modbus layer and
-        -- costs nothing; the outage it protects against is a battery left
-        -- charging with no way to say stop.
-        if power_w ~= 0 and model_family ~= "hybrid" and not battery_confirmed then
+        if model_family ~= "hybrid" and not battery_confirmed then
             local why = model_family == "string"
                 and "this model has no battery registers"
                 or "no battery register has answered on this device"
diff --git a/drivers/tests/test_sungrow_model_family.py b/drivers/tests/test_sungrow_model_family.py
index 221af8f..8ecf249 100644
--- a/drivers/tests/test_sungrow_model_family.py
+++ b/drivers/tests/test_sungrow_model_family.py
@@ -392,30 +392,28 @@ def test_string_inverter_still_takes_a_curtail_command() -> None:
 BOTH_DRIVERS = [pytest.param(DRIVER, id="catalog"), pytest.param(FTW_V2, id="ftw-v2")]
 
 
-def battery_command_on(driver: Path, power_w: int = 1000,
-                       polls: int = 4) -> str:
+def battery_command_on(driver: Path) -> str:
     """Send one battery setpoint through whichever ABI this file speaks."""
     if driver == FTW_V2:
         call = '''
 local result = driver_command_v2({
     command = "battery.set_power",
     runtime_action = "battery",
-    inputs = {power_w = POWER_W},
+    inputs = {power_w = 1000},
 })
 print("ACCEPTED " .. tostring(result.status ~= "rejected"))
 print("CODE " .. tostring(result.code or "none"))
 '''
     else:
         call = '''
-local accepted, refusal = driver_command("battery", POWER_W, {})
+local accepted, refusal = driver_command("battery", 1000, {})
 print("ACCEPTED " .. tostring(accepted == true))
 print("CODE " .. tostring(type(refusal) == "table" and refusal.code or "none"))
 '''
-    call = call.replace("POWER_W", str(power_w))
     return f'''
 dofile("{driver}")
 driver_init({{}})
-for poll = 1, {polls} do pcall(driver_poll) end
+for poll = 1, 4 do pcall(driver_poll) end
 host._modbus_write_attempts = 0
 {call}
 print("WRITES " .. tostring(host._modbus_write_attempts))
@@ -467,65 +465,231 @@ def test_neither_copy_refuses_a_battery_it_can_confirm(
         f"control off the fleet.")
 
 
-# ---------------------------------------------------------------------------
-# Zero is not a dispatch
-# ---------------------------------------------------------------------------
+# --------------------------------------------------------------------------
+# Zero watts, before the first poll
+#
+# srcfl/ftw#704 stalled here. FTW's TestSungrowZeroBatteryCommandForcesIdle
+# loads the driver, sends {"action":"battery","power_w":0} with no poll first,
+# and expects the forced-idle writes. The guard above refuses it: before the
+# first poll nothing has named a family and no battery register has answered.
+#
+# The refusal stands, and these tests carry the reason so it is not re-argued.
+#
+# The case for waving zero through was that it is not a dispatch but the
+# safety operation that hands a device back to itself, so refusing it can
+# strand an inverter in a forced state. Neither half survives contact with
+# this driver on this host:
+#
+#   * Zero does not hand the inverter back. driver_command("battery", 0)
+#     writes EMS mode 2 -- forced -- and pins the battery at 0 W under FTW's
+#     control. The write that returns an inverter to its own self-consumption
+#     is mode 0, and it lives in driver_default_mode.
+#   * The release is a separate entrypoint, and it is not gated. FTW reaches
+#     it through Registry.SendDefault on shutdown, lease expiry, the telemetry
+#     watchdog and the stale-site-meter standdown, never through
+#     driver_command. driver_init writes self-consumption too, before the run
+#     loop can accept any command -- so the window where a battery command is
+#     refused is a window where the inverter is already back on its own. On
+#     the control v2 path FTW does not even rely on that: it calls
+#     driver_default_mode_v2 itself once driver_init returns.
 #
-# A zero-power battery command is the host handing the device back to itself:
-# forced mode off, setpoint nought. It arrives from the lifecycle rather than
-# the planner, so it can land before the first poll -- when nothing has been
-# confirmed and the guard above would otherwise refuse it.
+# Refusing therefore withholds nothing that is not reachable by a route which
+# cannot refuse. Accepting would put mode 2 and a setpoint back on an SG
+# inverter that implements neither, and report success, because the read-back
+# that would catch it fails on that device too. That is the SG12RT bug again,
+# arriving through the one action nobody thought to guard.
 #
-# Refusing to write "stop" is a different risk from refusing to write
-# "charge". A device left in a forced state stays there, and the safe default
-# is the one path that must never be gated on how much is known about the
-# hardware. On a string inverter the write fails at the Modbus layer and costs
-# nothing; the outage it prevents is a battery with no way to be told to stop.
+# The residue is honest and small: if the init write fails on a real hybrid,
+# battery commands are refused until the first poll confirms the battery --
+# at most one poll interval, with driver_default_mode available throughout.
 #
-# FTW's TestSungrowZeroBatteryCommandForcesIdle sends exactly this, straight
-# after load. These cases are that test's rule, held here so the two cannot
-# drift.
+# The two copies agree on the guard, and the tests below hold both to it. They
+# do NOT agree on what zero writes: the catalog driver forces idle (mode 2),
+# the v2 target hands the inverter back (mode 0, device_state "default", which
+# clears FTW's lease). The v2 target uses a different register map throughout
+# -- 13050 and 13051 as per-direction limits rather than force command and
+# setpoint -- so this is one question inside a larger one. It belongs to the
+# register-map review packages/v1/sungrow/PILOT.md already requires before
+# that target ships; its verification_status is "experimental" and the signed
+# package builds from targets/ftw-observe.lua, so nothing here reaches
+# hardware today. Written down because untested drift between these two files
+# is how Pixii's 40288 came back.
+# --------------------------------------------------------------------------
+
+NOTHING_ANSWERED_YET = '''
+-- No registers configured, so every read answers 0 -- both the mock's default
+-- and Sungrow's documented "not present" code at 4999. No family named, no
+-- battery register answered: the state a driver sits in between driver_init
+-- and its first poll, and the state FTW's Go test leaves it in.
+'''
+
+
+def ems_mode_written() -> str:
+    """Report the last value written to the EMS mode register, 13049.
+
+    2 is forced control, 0 hands the inverter back to self-consumption. One
+    line tells apart a command that takes the device over from one that
+    releases it.
+    """
+    return '''
+local mode = "none"
+for _, call in ipairs(host._calls) do
+    if call.func == "modbus_write" and call.args[1] == 13049 then
+        mode = tostring(call.args[2])
+    end
+end
+print("EMS_MODE_WRITTEN " .. mode)
+local writes = {}
+for _, call in ipairs(host._calls) do
+    if call.func == "modbus_write" then
+        writes[#writes + 1] = tostring(call.args[1]) .. "=" .. tostring(call.args[2])
+    end
+end
+print("WRITE_LOG " .. table.concat(writes, ","))
+'''
+
+
+def zero_command_before_any_poll(driver: Path) -> str:
+    """Init, then one zero-watt battery command, with no poll in between.
+
+    The earliest a command can reach a driver on FTW. Registry.Add runs
+    driver_init and then starts the run loop, whose select serves the command
+    channel before the first poll timer fires -- and that timer is a full poll
+    interval out, longer with host.set_warmup_s.
+    """
+    if driver == FTW_V2:
+        call = '''
+local result = driver_command_v2({
+    command = "battery.set_power",
+    runtime_action = "battery",
+    inputs = {power_w = 0},
+})
+print("ACCEPTED " .. tostring(result.status ~= "rejected"))
+print("CODE " .. tostring(result.code or "none"))
+'''
+    else:
+        call = '''
+local accepted, refusal = driver_command("battery", 0, {})
+print("ACCEPTED " .. tostring(accepted == true))
+print("CODE " .. tostring(type(refusal) == "table" and refusal.code or "none"))
+'''
+    return f'''
+dofile("{driver}")
+driver_init({{}})
+host._modbus_write_attempts = 0
+{call}
+print("WRITES " .. tostring(host._modbus_write_attempts))
+'''
+
 
 @pytest.mark.parametrize("driver", BOTH_DRIVERS)
-@pytest.mark.parametrize("device,label", [
-    pytest.param(STRING_INVERTER,
-                 "a model that names itself a string inverter",
-                 id="named-string"),
-    pytest.param(UNIDENTIFIED_STRING,
-                 "a model that names nothing and answers no battery register",
-                 id="unidentified-string"),
-    pytest.param(HEALTHY_HYBRID, "a hybrid", id="hybrid"),
-])
-def test_zero_is_never_refused(driver: Path, device: str, label: str) -> None:
-    out = run_lua(device + battery_command_on(driver, power_w=0))
+def test_zero_watts_is_still_a_battery_command(driver: Path) -> None:
+    """A setpoint of zero is a setpoint. It proves nothing about the battery.
+
+    Nothing here says this device has one: no family named, no battery
+    register answered. A guard that trusts the number rather than the evidence
+    is not a guard -- an SG string inverter takes a zero as readily as it
+    takes a thousand, and answers success to both.
+    """
+    out = run_lua(NOTHING_ANSWERED_YET + zero_command_before_any_poll(driver))
+
+    assert out["ACCEPTED"] == "false", (
+        f"{driver.name} accepted a zero-watt battery command on a device that "
+        f"has confirmed nothing. Zero is not exempt: it writes the same three "
+        f"registers to the same models, and reports the same false success.")
+    assert out["WRITES"] == "0", (
+        f"{driver.name} refused and still wrote {out['WRITES']} registers")
+    assert out["CODE"] == "no_battery", (
+        f"{driver.name} refused with code {out['CODE']!r}, not 'no_battery'")
+
+
+def test_zero_watts_takes_the_inverter_over_rather_than_handing_it_back() -> None:
+    """The claim that decides it.
+
+    If zero released the device, refusing it would withhold a safety action.
+    It does not: it writes mode 2 and holds the battery at 0 W under FTW's
+    control. An inverter is more controlled after this command than before.
+    """
+    body = HEALTHY_HYBRID + f'''
+dofile("{DRIVER}")
+driver_init({{}})
+for poll = 1, 3 do pcall(driver_poll) end
+host._calls = {{}}
+local accepted = driver_command("battery", 0, {{}})
+print("ACCEPTED " .. tostring(accepted == true))
+''' + ems_mode_written()
+    out = run_lua(body)
 
-    # Not refused by the guard is the rule. Whether the write then succeeds is
-    # a separate matter: on a device that does not implement 13049 the
-    # read-back fails and the driver says so, which is the honest outcome. The
-    # thing that must never happen is the command being turned away before it
-    # is tried.
-    assert out["CODE"] != "no_battery", (
-        f"{driver.name} refused a zero-power battery command on {label}. Zero "
-        f"is the safe default -- the host handing the device back to itself. "
-        f"A driver that will not accept 'stop' can leave a battery in a "
-        f"forced state with no way out.")
+    assert out["ACCEPTED"] == "true", (
+        "a confirmed hybrid must still take a zero-watt setpoint")
+    assert out["WRITE_LOG"] == "13049=2,13050=204,13051=0", (
+        f"zero wrote {out['WRITE_LOG']}. FTW's Go test expects these three "
+        f"and no others, so the two repositories agree on what zero does.")
+    assert out["EMS_MODE_WRITTEN"] == "2", (
+        f"zero left EMS mode at {out['EMS_MODE_WRITTEN']}. Mode 2 is forced "
+        f"control; mode 0 is the release. Calling this a safety operation "
+        f"reads it backwards.")
+
+
+def release_before_any_poll(driver: Path) -> str:
+    """Init, then the release, with no poll in between."""
+    if driver == FTW_V2:
+        call = '''
+local result = driver_default_mode_v2({reason = "host_shutdown"})
+print("RELEASED " .. tostring(result.status ~= "rejected"))
+'''
+    else:
+        call = '''
+print("RELEASED " .. tostring(driver_default_mode() == true))
+'''
+    return f'''
+dofile("{driver}")
+driver_init({{}})
+host._calls = {{}}
+{call}
+''' + ems_mode_written()
 
 
 @pytest.mark.parametrize("driver", BOTH_DRIVERS)
-def test_zero_lands_before_the_first_poll(driver: Path) -> None:
-    """The ordering FTW's Go suite exercises: load, then command, no poll.
+def test_the_release_is_never_gated(driver: Path) -> None:
+    """Why refusing a battery command is safe, and the thing that must not move.
 
-    Nothing has answered yet, so neither the model family nor the battery is
-    confirmed. This is precisely when the guard must not fire.
+    Refusing costs nothing only while handing the inverter back stays
+    reachable without evidence of a battery. Gate driver_default_mode the way
+    driver_command is gated and the whole argument collapses: a device would
+    then be able to hold a forced state that nothing is allowed to clear.
     """
-    out = run_lua(UNIDENTIFIED_STRING
-                  + battery_command_on(driver, power_w=0, polls=0))
+    out = run_lua(NOTHING_ANSWERED_YET + release_before_any_poll(driver))
+
+    assert out["RELEASED"] == "true", (
+        f"{driver.name} refused to hand the inverter back on a device that "
+        f"has confirmed nothing. Shutdown, lease expiry, the watchdog and the "
+        f"stale-meter standdown all arrive here, and none of them can wait "
+        f"for a poll.")
+    assert out["EMS_MODE_WRITTEN"] == "0", (
+        f"the release left EMS mode at {out['EMS_MODE_WRITTEN']}, not 0. "
+        f"Anything else keeps the inverter under FTW's control.")
+
 
-    assert out["CODE"] != "no_battery", (
-        f"{driver.name} refused a zero-power command issued before the first "
-        f"poll. Detection has had no chance to run at that point, and a "
-        f"safety idle cannot wait for it.")
+def test_init_hands_the_inverter_back_before_a_command_can_arrive() -> None:
+    """The other reason refusing costs nothing.
+
+    A container that dies mid-force leaves the inverter holding the last
+    command. driver_init clears it -- and it runs before FTW's run loop exists
+    to accept a command at all. So there is no window where a battery command
+    is refused and the inverter is stranded in a forced state.
+    """
+    body = NOTHING_ANSWERED_YET + f'''
+dofile("{DRIVER}")
+driver_init({{}})
+''' + ems_mode_written()
+    out = run_lua(body)
 
+    assert out["EMS_MODE_WRITTEN"] == "0", (
+        f"driver_init left EMS mode at {out['EMS_MODE_WRITTEN']}. It has to "
+        f"reach self-consumption unconditionally: the family is often still "
+        f"unknown here, and an inverter still holding yesterday's forced "
+        f"charge is the case this write exists for.")
 
 # --------------------------------------------------------------------------
 # Startup traffic, and the line it must not cross
@@ -543,11 +707,12 @@ def test_zero_lands_before_the_first_poll(driver: Path) -> None:
 #
 # That last clause is the whole boundary, and the last test here holds it. An
 # earlier draft of this change gated set_self_consumption on the same family
-# test. Since 1.5.4 a zero-watt command is accepted on every family, so a
-# device classified "string" -- a Sungrow hybrid shipped under a device-type
-# code classify_device_type has not been taught lands there -- can be holding
-# EMS mode 2, and set_self_consumption is the only thing that writes mode 0
-# back. Gating it left that device forced with nothing able to release it.
+# test. A device classified "string" -- a Sungrow hybrid shipped under a
+# device-type code classify_device_type has not been taught lands there -- can
+# still be holding EMS mode 2, left by an earlier run or by a firmware that
+# answers the type register differently after an update. set_self_consumption
+# is the only thing that writes mode 0 back, so gating it left that device
+# forced with nothing able to release it.
 # --------------------------------------------------------------------------
 
 BATTERY_LIMIT_BLOCK = [13057, 33046, 33047]
@@ -603,8 +768,8 @@ def test_string_inverter_startup_reads_no_battery_limit_it_cannot_have() -> None
         f"registers. Those are the ones nothing needs.")
 
     # Two failed reads remain, both at 13049, and both are meant to. One is
-    # set_self_consumption's readback, which must run on every family since
-    # 1.5.4 -- see test_the_release_is_still_reachable_on_a_string_inverter.
+    # set_self_consumption's readback, which must run on every family --
+    # see test_the_release_is_still_reachable_on_a_string_inverter.
     # The other is the startup EMS-state log, which is exactly where a device
     # misclassified as "string" while holding mode 2 would show up. Gating
     # either to make this number nicer would cost more than it saves.
@@ -633,11 +798,11 @@ def test_hybrid_still_gets_its_power_limits_raised() -> None:
 def test_the_release_is_still_reachable_on_a_string_inverter() -> None:
     """The line the startup gate must not cross.
 
-    Since 1.5.4 a zero-watt command is accepted whatever the family, and it
-    writes EMS mode 2. So "the device named itself a string inverter" is not
-    proof there is nothing to release: Sungrow shipping a hybrid under an
-    unrecognised device-type code lands in exactly that branch, takes the
-    forced write, and then needs mode 0 written back.
+    "The device named itself a string inverter" is not proof there is nothing
+    to release. A Sungrow hybrid shipped under an unrecognised device-type
+    code lands in exactly that branch, and it can already be holding EMS mode
+    2 from an earlier run or an earlier firmware. Mode 0 still has to be
+    written back.
 
     Gate set_self_consumption on the family and this fails -- the device stays
     at mode 2 through the watchdog and through cleanup, with nothing left that
@@ -652,8 +817,11 @@ def test_the_release_is_still_reachable_on_a_string_inverter() -> None:
 dofile("{DRIVER}")
 driver_init({{}})
 for poll = 1, 3 do pcall(driver_poll) end
-driver_command("battery", 0, {{}})
-print("FORCED " .. tostring(host._modbus_registers.holding[13049]))
+-- Forced by an earlier run rather than by a command: the guard refuses a
+-- battery setpoint on this family, which is the point -- the device is
+-- holding mode 2 and nothing on the command path can clear it.
+host._modbus_registers.holding[13049] = {{2}}
+print("FORCED " .. tostring(host._modbus_registers.holding[13049][1]))
 driver_default_mode()
 print("AFTER_WATCHDOG " .. tostring(host._modbus_registers.holding[13049]))
 driver_cleanup()
@@ -662,7 +830,7 @@ def test_the_release_is_still_reachable_on_a_string_inverter() -> None:
     out = run_lua(body)
 
     assert out["FORCED"] == "2", (
-        "a zero-watt command is meant to reach every family since 1.5.4")
+        "the fixture failed to leave the device in a forced state")
     assert out["AFTER_WATCHDOG"] == "0", (
         f"the watchdog left EMS mode at {out['AFTER_WATCHDOG']}, not 0. The "
         f"device is holding a forced state and the release is the only write "
diff --git a/index.yaml b/index.yaml
index 0712026..acaa556 100644
--- a/index.yaml
+++ b/index.yaml
@@ -561,13 +561,13 @@ drivers:
     size_bytes: 6158
     sha256: "92163e4356b895e16b0205e849aa42a2dcea0d799f8cc6bcbce60b5e2cd92285"
   - name: "sungrow"
-    version: "1.5.5"
+    version: "1.5.6"
     tier: core
     protocol: modbus
     ders: [battery, meter, pv]
     control: true
-    size_bytes: 45670
-    sha256: "9667377a29f26d4325e3df450b93d9a16b802f778b54bb15a894eab26b8be306"
+    size_bytes: 44846
+    sha256: "8f350fc209f7231139c23bc336695b6e00267fb4b3618e26c754c77fd3ac7e7d"
   - name: "tesla_vehicle"
     version: "0.2.0"
     tier: core
diff --git a/manifests/sungrow.yaml b/manifests/sungrow.yaml
index db7e29e..99da9d3 100644
--- a/manifests/sungrow.yaml
+++ b/manifests/sungrow.yaml
@@ -1,5 +1,5 @@
 name: "sungrow"
-version: "1.5.5"
+version: "1.5.6"
 tier: core
 author: "Sourceful Labs AB"
 protocol: modbus
@@ -42,9 +42,9 @@ tested_devices:
     notes: "Commercial/industrial three-phase string inverter. Needs 1.5.3: 1.2.1 stopped polling the SH battery registers this family does not answer, and 1.5.3 refuses a battery dispatch instead of writing them. A zero-watt command is still written on every family from 1.5.4 -- it is the release, and on a model without the registers it fails harmlessly."
     min_driver_version: "1.5.3"
 min_host_version: "2.0.0"
-size_bytes: 45670
+size_bytes: 44846
 dkb_id: "sungrow_sh_rt"
-sha256: "9667377a29f26d4325e3df450b93d9a16b802f778b54bb15a894eab26b8be306"
+sha256: "8f350fc209f7231139c23bc336695b6e00267fb4b3618e26c754c77fd3ac7e7d"
 signature: ""
 
 bytecode_sha256: ""
diff --git a/packages/v1/sungrow/targets/ftw.lua b/packages/v1/sungrow/targets/ftw.lua
index 39e1777..e2cca02 100644
--- a/packages/v1/sungrow/targets/ftw.lua
+++ b/packages/v1/sungrow/targets/ftw.lua
@@ -12,7 +12,7 @@ DRIVER = {
     id = "sungrow",
     name = "Sungrow hybrid and string inverter",
     manufacturer = "Sungrow",
-    version = "1.3.4",
+    version = "1.3.5",
     protocols = { "modbus" },
     capabilities = { "pv", "battery", "meter" },
     description = "Sungrow hybrid and string inverters via Modbus.",
@@ -413,14 +413,7 @@ function driver_command_v2(command)
         -- implement. Checking only for "string" left the gap: an SG inverter
         -- whose device-type register never answered fell through and got the
         -- writes. Matches drivers/lua/sungrow.lua, deliberately.
-        --
-        -- Zero is exempt. It is not a dispatch but the host handing the
-        -- device back to itself, it arrives from the lifecycle rather than
-        -- the planner -- so before anything has been confirmed -- and a
-        -- device left in a forced state with no way to say stop is the worse
-        -- outage. On a string inverter the write fails at the Modbus layer
-        -- and costs nothing.
-        if inputs.power_w ~= 0 and model_family ~= "hybrid" and not battery_confirmed then
+        if model_family ~= "hybrid" and not battery_confirmed then
             return {
                 status = "rejected",
                 code = "no_battery",
diff --git a/support-status.json b/support-status.json
index 44a16b6..90368a2 100644
--- a/support-status.json
+++ b/support-status.json
@@ -1962,7 +1962,7 @@
     },
     {
       "catalog_source": true,
-      "catalog_version": "1.5.5",
+      "catalog_version": "1.5.6",
       "driver_id": "sungrow",
       "package_id": "com.sourceful.driver.sungrow",
       "targets": {