diff --git a/pandas/tests/extension/test_arrow.py b/pandas/tests/extension/test_arrow.py
index a7c243cdfe74f..1bfe0106e1938 100644
--- a/pandas/tests/extension/test_arrow.py
+++ b/pandas/tests/extension/test_arrow.py
@@ -1010,14 +1010,29 @@ def _patch_combine(self, obj, other, op):
         else:
             expected_data = expected
             original_dtype = obj.dtype
-        pa_array = pa.array(expected_data._values).cast(original_dtype.pyarrow_dtype)
-        pd_array = type(expected_data._values)(pa_array)
+
+        pa_expected = pa.array(expected_data._values)
+
+        if pa.types.is_duration(pa_expected.type):
+            # pyarrow sees sequence of datetime/timedelta objects and defaults
+            #  to "us" but the non-pointwise op retains unit
+            unit = original_dtype.pyarrow_dtype.unit
+            if type(other) in [datetime, timedelta] and unit in ["s", "ms"]:
+                # pydatetime/pytimedelta objects have microsecond reso, so we
+                #  take the higher reso of the original and microsecond. Note
+                #  this matches what we would do with DatetimeArray/TimedeltaArray
+                unit = "us"
+            pa_expected = pa_expected.cast(f"duration[{unit}]")
+        else:
+            pa_expected = pa_expected.cast(original_dtype.pyarrow_dtype)
+
+        pd_expected = type(expected_data._values)(pa_expected)
         if was_frame:
             expected = pd.DataFrame(
-                pd_array, index=expected.index, columns=expected.columns
+                pd_expected, index=expected.index, columns=expected.columns
             )
         else:
-            expected = pd.Series(pd_array)
+            expected = pd.Series(pd_expected)
         return expected
 
     def _is_temporal_supported(self, opname, pa_dtype):
@@ -1097,7 +1112,14 @@ def test_arith_series_with_scalar(
         if mark is not None:
             request.node.add_marker(mark)
 
-        if all_arithmetic_operators == "__floordiv__" and pa.types.is_integer(pa_dtype):
+        if (
+            (
+                all_arithmetic_operators == "__floordiv__"
+                and pa.types.is_integer(pa_dtype)
+            )
+            or pa.types.is_duration(pa_dtype)
+            or pa.types.is_timestamp(pa_dtype)
+        ):
             # BaseOpsUtil._combine always returns int64, while ArrowExtensionArray does
             # not upcast
             monkeypatch.setattr(TestBaseArithmeticOps, "_combine", self._patch_combine)
@@ -1121,7 +1143,14 @@ def test_arith_frame_with_scalar(
         if mark is not None:
             request.node.add_marker(mark)
 
-        if all_arithmetic_operators == "__floordiv__" and pa.types.is_integer(pa_dtype):
+        if (
+            (
+                all_arithmetic_operators == "__floordiv__"
+                and pa.types.is_integer(pa_dtype)
+            )
+            or pa.types.is_duration(pa_dtype)
+            or pa.types.is_timestamp(pa_dtype)
+        ):
             # BaseOpsUtil._combine always returns int64, while ArrowExtensionArray does
             # not upcast
             monkeypatch.setattr(TestBaseArithmeticOps, "_combine", self._patch_combine)
@@ -1165,18 +1194,41 @@ def test_arith_series_with_array(
         # since ser.iloc[0] is a python scalar
         other = pd.Series(pd.array([ser.iloc[0]] * len(ser), dtype=data.dtype))
 
-        if pa.types.is_floating(pa_dtype) or (
-            pa.types.is_integer(pa_dtype) and all_arithmetic_operators != "__truediv__"
+        if (
+            pa.types.is_floating(pa_dtype)
+            or (
+                pa.types.is_integer(pa_dtype)
+                and all_arithmetic_operators != "__truediv__"
+            )
+            or pa.types.is_duration(pa_dtype)
+            or pa.types.is_timestamp(pa_dtype)
         ):
             monkeypatch.setattr(TestBaseArithmeticOps, "_combine", self._patch_combine)
         self.check_opname(ser, op_name, other, exc=self.series_array_exc)
 
     def test_add_series_with_extension_array(self, data, request):
         pa_dtype = data.dtype.pyarrow_dtype
-        if not (
-            pa.types.is_integer(pa_dtype)
-            or pa.types.is_floating(pa_dtype)
-            or (not pa_version_under8p0 and pa.types.is_duration(pa_dtype))
+
+        if pa.types.is_temporal(pa_dtype) and not pa.types.is_duration(pa_dtype):
+            # i.e. timestamp, date, time, but not timedelta; these *should*
+            #  raise when trying to add
+            ser = pd.Series(data)
+            if pa_version_under7p0:
+                msg = "Function add_checked has no kernel matching input types"
+            else:
+                msg = "Function 'add_checked' has no kernel matching input types"
+            with pytest.raises(NotImplementedError, match=msg):
+                # TODO: this is a pa.lib.ArrowNotImplementedError, might
+                #  be better to reraise a TypeError; more consistent with
+                #  non-pyarrow cases
+                ser + data
+
+            return
+
+        if (pa_version_under8p0 and pa.types.is_duration(pa_dtype)) or (
+            pa.types.is_binary(pa_dtype)
+            or pa.types.is_string(pa_dtype)
+            or pa.types.is_boolean(pa_dtype)
         ):
             request.node.add_marker(
                 pytest.mark.xfail(