diff --git a/apps/predbat/fetch.py b/apps/predbat/fetch.py
index 9f03c20c6..b63057ab0 100644
--- a/apps/predbat/fetch.py
+++ b/apps/predbat/fetch.py
@@ -1489,7 +1489,7 @@ def find_charge_window(self, rates, minute, threshold_rate, find_high, alt_rates
             else:
                 if rate_low_start >= 0 and rate_low_end >= minute:
                     rate_low_end = minute
-                break
+                    break
             minute += 5
             alt_rate_last = alt_rate
 
diff --git a/apps/predbat/tests/test_find_charge_window.py b/apps/predbat/tests/test_find_charge_window.py
new file mode 100644
index 000000000..bc8814dcc
--- /dev/null
+++ b/apps/predbat/tests/test_find_charge_window.py
@@ -0,0 +1,318 @@
+# -----------------------------------------------------------------------------
+# Predbat Home Battery System
+# Copyright Trefor Southwell 2026 - All Rights Reserved
+# This application maybe used for personal use only and not for commercial use
+# -----------------------------------------------------------------------------
+# fmt: off
+# pylint: disable=consider-using-f-string
+# pylint: disable=line-too-long
+# pylint: disable=attribute-defined-outside-init
+
+
+def _assert_window(name, start, end, average, expected_start, expected_end, expected_average=None):
+    """Helper: assert window fields and return 0/1 failed."""
+    failed = 0
+    if start != expected_start:
+        print("ERROR: {}: expected start={}, got start={}".format(name, expected_start, start))
+        failed = 1
+    if end != expected_end:
+        print("ERROR: {}: expected end={}, got end={}".format(name, expected_end, end))
+        failed = 1
+    if expected_average is not None and average != expected_average:
+        print("ERROR: {}: expected average={}, got average={}".format(name, expected_average, average))
+        failed = 1
+    return failed
+
+
+def test_find_charge_window(my_predbat):
+    """
+    Comprehensive tests for find_charge_window covering all code paths:
+
+      Path A - mixed rate exceeds combine_rate_threshold → window closed early
+      Path B - combine_export_slots=False hits export_slot_split → split
+      Path C - combine_charge_slots=False hits charge_slot_split → split
+      Path D - manual_all_times slot split at plan_interval_minutes
+      Path E - alt_rates alternate_rate_boundary terminates export window;
+               also 24-hour export cap
+      Path F - window start (rate_low_start set for first time)
+      Path G - window continuation and correct average calculation
+      Path H - rate outside threshold while window in progress → closes window
+      Path I - gap in rates while window in progress → closes window (regression
+               for previous break-too-early bug)
+      Misc  - no qualifying rates, find_high=False, find_high=True, zero rate
+               excluded for find_high, scan from non-zero minute, scan stops at
+               forecast boundary
+    """
+    failed = 0
+    low_rate = 5.0
+    high_rate = 20.0
+    thresh_lo = 10.0   # threshold for charge (find_high=False)
+    thresh_hi = 15.0   # threshold for export (find_high=True)
+
+    # Preserve ALL settings we may temporarily change, and set known baselines
+    # so the test is independent of whatever previous tests may have modified.
+    old_forecast_minutes = my_predbat.forecast_minutes
+    old_minutes_now = my_predbat.minutes_now
+    old_combine_charge = my_predbat.combine_charge_slots
+    old_combine_export = my_predbat.combine_export_slots
+    old_combine_thresh = my_predbat.combine_rate_threshold
+    old_manual_all_times = my_predbat.manual_all_times
+    old_charge_slot_split = my_predbat.charge_slot_split
+    old_export_slot_split = my_predbat.export_slot_split
+    old_plan_interval = my_predbat.plan_interval_minutes
+
+    # Fix known-good baseline values
+    my_predbat.forecast_minutes = 24 * 60
+    my_predbat.minutes_now = 12 * 60
+    my_predbat.combine_charge_slots = True
+    my_predbat.combine_export_slots = True
+
+    scan_end = my_predbat.forecast_minutes + my_predbat.minutes_now + 12 * 60
+
+    # -----------------------------------------------------------------------
+    # Test: no rates → no window found
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: empty rates dict returns no window")
+    s, e, avg = my_predbat.find_charge_window({}, 0, thresh_lo, find_high=False)
+    if s != -1 or e != -1:
+        print("ERROR: empty rates: expected (-1,-1), got ({},{})".format(s, e))
+        failed = 1
+
+    # -----------------------------------------------------------------------
+    # Test: simple charge window (find_high=False) — Path F + G
+    # Rates start at 0, low for 120 min, high after → window [0, 120]
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: simple charge window (find_high=False)")
+    rates = {}
+    for m in range(0, 120, 5):
+        rates[m] = low_rate
+    for m in range(120, scan_end, 5):
+        rates[m] = high_rate
+
+    s, e, avg = my_predbat.find_charge_window(rates, 0, thresh_lo, find_high=False)
+    failed |= _assert_window("simple charge", s, e, avg, 0, 120, low_rate)
+
+    # -----------------------------------------------------------------------
+    # Test: simple export window (find_high=True) — Path F + G
+    # Rates start at 0, high for 120 min, low after → window [0, 120]
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: simple export window (find_high=True)")
+    rates2 = {}
+    for m in range(0, 120, 5):
+        rates2[m] = high_rate
+    for m in range(120, scan_end, 5):
+        rates2[m] = low_rate
+
+    s, e, avg = my_predbat.find_charge_window(rates2, 0, thresh_hi, find_high=True)
+    failed |= _assert_window("simple export", s, e, avg, 0, 120, high_rate)
+
+    # -----------------------------------------------------------------------
+    # Test: find_high=True with zero rate — zero should NOT qualify
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: find_high with zero rate excluded")
+    rates_zero = {0: 0.0, 5: 0.0, 10: 0.0}
+    s, e, _ = my_predbat.find_charge_window(rates_zero, 0, thresh_hi, find_high=True)
+    if s != -1:
+        print("ERROR: zero rate for find_high should not qualify, got start={}".format(s))
+        failed = 1
+
+    # -----------------------------------------------------------------------
+    # Test: rate goes ABOVE threshold while window in progress — Path H
+    # Low rates 0..59, high rate at 60.. → window closes at 60
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: rate above threshold closes window (Path H)")
+    rates_h = {}
+    for m in range(0, 60, 5):
+        rates_h[m] = low_rate
+    for m in range(60, scan_end, 5):
+        rates_h[m] = high_rate
+
+    s, e, avg = my_predbat.find_charge_window(rates_h, 0, thresh_lo, find_high=False)
+    failed |= _assert_window("path H charge close", s, e, avg, 0, 60, low_rate)
+
+    # -----------------------------------------------------------------------
+    # Test: Path A — combine_rate_threshold: second rate too different
+    # Low rate 5 for 0..55, then rate 8 at 60 (diff = 3 > threshold 1.0)
+    # → window closes at 60 due to combine_rate_threshold
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: combine_rate_threshold closes mixed window (Path A)")
+    my_predbat.combine_rate_threshold = 2.0   # allow up to 2p difference
+    rates_a = {}
+    for m in range(0, 60, 5):
+        rates_a[m] = 5.0
+    for m in range(60, scan_end, 5):
+        rates_a[m] = 8.5  # diff = 3.5 > 2.0
+
+    s, e, _ = my_predbat.find_charge_window(rates_a, 0, thresh_lo, find_high=False)
+    failed |= _assert_window("path A combine_rate_threshold", s, e, _, 0, 60)
+    my_predbat.combine_rate_threshold = old_combine_thresh
+
+    # -----------------------------------------------------------------------
+    # Test: Path C — combine_charge_slots=False splits charge at charge_slot_split
+    # Low rates from 0 to 120, charge_slot_split=30 → window [0, 30]
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: combine_charge_slots=False splits slot (Path C)")
+    my_predbat.combine_charge_slots = False
+    my_predbat.charge_slot_split = 30
+    rates_c = {m: low_rate for m in range(0, 120, 5)}
+
+    s, e, _ = my_predbat.find_charge_window(rates_c, 0, thresh_lo, find_high=False)
+    failed |= _assert_window("path C charge slot split", s, e, _, 0, 30)
+    my_predbat.combine_charge_slots = True
+
+    # -----------------------------------------------------------------------
+    # Test: Path B — combine_export_slots=False splits export at export_slot_split
+    # High rates from 0 to 120, export_slot_split=30 → window [0, 30]
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: combine_export_slots=False splits slot (Path B)")
+    my_predbat.combine_export_slots = False
+    my_predbat.export_slot_split = 30
+    rates_b = {m: high_rate for m in range(0, 120, 5)}
+
+    s, e, _ = my_predbat.find_charge_window(rates_b, 0, thresh_hi, find_high=True)
+    failed |= _assert_window("path B export slot split", s, e, _, 0, 30)
+    my_predbat.combine_export_slots = True
+
+    # -----------------------------------------------------------------------
+    # Test: Path D — manual_all_times slot split at plan_interval_minutes
+    # rate_low_start=0 is in manual_all_times; low rates from 0 to 120;
+    # plan_interval_minutes=30 → window closes at minute 30
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: manual_all_times split at plan_interval_minutes (Path D)")
+    my_predbat.plan_interval_minutes = 30
+    my_predbat.charge_slot_split = 30
+    my_predbat.manual_all_times = [0]   # start minute is a manual time
+    rates_d = {m: low_rate for m in range(0, 120, 5)}
+
+    s, e, _ = my_predbat.find_charge_window(rates_d, 0, thresh_lo, find_high=False)
+    failed |= _assert_window("path D manual split", s, e, _, 0, 30)
+    my_predbat.manual_all_times = []
+
+    # -----------------------------------------------------------------------
+    # Test: Path E — alternate_rate_boundary splits export window
+    # find_high=True; alt_rates changes significantly between minutes 25→30;
+    # high export rates throughout; plan_interval_minutes=30;
+    # → export window should end at 30 (boundary reached after plan_interval)
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: alternate_rate_boundary splits export window (Path E)")
+    my_predbat.plan_interval_minutes = 30
+    my_predbat.export_slot_split = 30
+    # alt_rates min=0, max=20 → alt_rate_threshold=2.0; jump of 10 qualifies
+    alt_rates_e = {}
+    for m in range(0, 30, 5):
+        alt_rates_e[m] = 0.0      # low alt rate
+    for m in range(30, scan_end, 5):
+        alt_rates_e[m] = 20.0     # high alt rate (diff=20 >= threshold=2.0)
+
+    rates_e = {m: high_rate for m in range(0, scan_end, 5)}  # all qualify
+
+    # Window starts at 0; alternate_rate_boundary is set at minute 30;
+    # at minute 30, (30 - 0) = 30 >= plan_interval_minutes=30 → break
+    s, e, _ = my_predbat.find_charge_window(rates_e, 0, thresh_hi, find_high=True, alt_rates=alt_rates_e)
+    failed |= _assert_window("path E alt rate boundary", s, e, _, 0, 30)
+
+    # -----------------------------------------------------------------------
+    # Test: 24-hour export cap (Path E first condition)
+    # find_high=True with high rates spanning >24h; window must cap at 24*60
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: 24-hour export cap (Path E)")
+    rates_24h = {m: high_rate for m in range(0, scan_end, 5)}
+    s, e, _ = my_predbat.find_charge_window(rates_24h, 0, thresh_hi, find_high=True)
+    if e - s != 24 * 60:
+        print("ERROR: 24h cap: expected window length={}, got length={}".format(24 * 60, e - s))
+        failed = 1
+
+    # -----------------------------------------------------------------------
+    # Test: correct average over varying rates (Path G)
+    # Three rate bands: 4, 6, 8 each for 1 slot (5 min) → average = (4+6+8)/3 = 6
+    # Raise combine_rate_threshold so all three rates are combined into one window.
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: correct average calculation (Path G)")
+    my_predbat.combine_rate_threshold = 10.0   # allow wide mix for this test
+    rates_avg = {0: 4.0, 5: 6.0, 10: 8.0}
+    for m in range(15, scan_end, 5):
+        rates_avg[m] = high_rate   # end the window
+
+    s, e, avg = my_predbat.find_charge_window(rates_avg, 0, thresh_lo, find_high=False)
+    expected_avg = round((4.0 + 6.0 + 8.0) / 3, 2)
+    failed |= _assert_window("path G average", s, e, avg, 0, 15, expected_avg)
+    my_predbat.combine_rate_threshold = old_combine_thresh
+
+    # -----------------------------------------------------------------------
+    # Test: scan from non-zero starting minute
+    # Low rates start at minute 60 but scanning begins from minute 0 (gap
+    # before window), versus starting scan at minute 60 directly
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: scan from non-zero start minute")
+    rates_nz = {}
+    for m in range(60, 180, 5):
+        rates_nz[m] = low_rate
+    for m in range(180, scan_end, 5):
+        rates_nz[m] = high_rate
+
+    s, e, _ = my_predbat.find_charge_window(rates_nz, 60, thresh_lo, find_high=False)
+    failed |= _assert_window("non-zero scan start", s, e, _, 60, 180)
+
+    # -----------------------------------------------------------------------
+    # Test (regression — gap handling): gap BEFORE the qualifying window
+    # Rates absent for 0..55, low 60..175, high 180+
+    # Scan must continue past missing entries and find window at 60
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: gap before qualifying window (regression)")
+    rates_gap_before = {}
+    for m in range(60, 180, 5):
+        rates_gap_before[m] = low_rate
+    for m in range(180, scan_end, 5):
+        rates_gap_before[m] = high_rate
+
+    s, e, avg = my_predbat.find_charge_window(rates_gap_before, 0, thresh_lo, find_high=False)
+    failed |= _assert_window("gap before window", s, e, avg, 60, 180, low_rate)
+
+    # -----------------------------------------------------------------------
+    # Test (regression — gap handling): gap WITHIN an active window (Path I)
+    # Low 0..115, gap 120..175, high 180+
+    # Open window must close at first missing entry = 120
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: gap within active window closes it (Path I)")
+    rates_gap_within = {}
+    for m in range(0, 120, 5):
+        rates_gap_within[m] = low_rate
+    for m in range(180, scan_end, 5):
+        rates_gap_within[m] = high_rate  # gap 120..175
+
+    s, e, _ = my_predbat.find_charge_window(rates_gap_within, 0, thresh_lo, find_high=False)
+    failed |= _assert_window("gap within window", s, e, _, 0, 120)
+
+    # -----------------------------------------------------------------------
+    # Test (regression): rate_scan_window finds window when initial gap precedes it
+    # -----------------------------------------------------------------------
+    print("Test find_charge_window: rate_scan_window finds window past initial gap")
+    gap_start = my_predbat.minutes_now + 60
+    window_start = gap_start + 60
+    window_end = window_start + 120
+
+    rates_rsw = {}
+    for m in range(window_start, window_end, 5):
+        rates_rsw[m] = low_rate
+    for m in range(window_end, scan_end, 5):
+        rates_rsw[m] = high_rate
+
+    found_rates, _, _ = my_predbat.rate_scan_window(rates_rsw, 5, thresh_lo, find_high=False)
+    if not found_rates:
+        print("ERROR: rate_scan_window found no windows when gap precedes qualifying rates")
+        failed = 1
+    elif found_rates[0]["start"] != window_start:
+        print("ERROR: rate_scan_window expected first window start={}, got {}".format(window_start, found_rates[0]["start"]))
+        failed = 1
+
+    # Restore all settings
+    my_predbat.forecast_minutes = old_forecast_minutes
+    my_predbat.minutes_now = old_minutes_now
+    my_predbat.combine_charge_slots = old_combine_charge
+    my_predbat.combine_export_slots = old_combine_export
+    my_predbat.combine_rate_threshold = old_combine_thresh
+    my_predbat.manual_all_times = old_manual_all_times
+    my_predbat.charge_slot_split = old_charge_slot_split
+    my_predbat.export_slot_split = old_export_slot_split
+    my_predbat.plan_interval_minutes = old_plan_interval
+    return failed
diff --git a/apps/predbat/unit_test.py b/apps/predbat/unit_test.py
index 9ee09067f..32d5db64c 100644
--- a/apps/predbat/unit_test.py
+++ b/apps/predbat/unit_test.py
@@ -95,6 +95,7 @@
 from tests.test_integer_config import test_integer_config_entities, test_expose_config_preserves_integer
 from tests.test_plan_json_rate_adjust import run_test_plan_json_rate_adjust
 from tests.test_rate_replicate_missing_slots import test_rate_replicate
+from tests.test_find_charge_window import test_find_charge_window
 from tests.test_carbon import test_carbon
 from tests.test_download import test_download
 from tests.test_ohme import test_ohme
@@ -213,6 +214,7 @@ def main():
         ("multi_car_iog", run_multi_car_iog_tests, "Multi-car IOG tests", False),
         ("rate_add_io_slots", run_rate_add_io_slots_tests, "Rate add IO slots tests", False),
         ("rate_replicate", test_rate_replicate, "Rate replicate comprehensive tests (missing slots, IO, offsets, gas)", False),
+        ("find_charge_window", test_find_charge_window, "Find charge window gap handling tests", False),
         ("find_charge_rate", test_find_charge_rate, "Find charge rate tests", False),
         ("find_charge_rate_string_temp", test_find_charge_rate_string_temperature, "Find charge rate string temperature", False),
         ("find_charge_rate_string_curve", test_find_charge_rate_string_charge_curve, "Find charge rate string charge curve", False),