feat(aerotech): settle_time 0.3 s default, trigger calibration per settle time

At the beamline 0.15 s settle left a forward / backward offset of +18 .. +28 um
at 20 mm/s that changed between repeats (oscillation after the ramp); with 0.3 s
it is -2 .. -5 um and reproducible, at ~0.22 s more per line. settle_time and
minimal_run_up stay options. Calibration entries now carry settle_time (and
optionally minimal_run_up); a scan only uses the entries of its own settle time
and run-up mode, so a table measured with one settling is never applied to
another. The printed entry includes settle_time.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017NrH3poJsYBCTVq2Y9fuHZ

csaxs_bec/devices/aerotech/README.md

@@ -59,7 +59,7 @@ Documentation: A3200 6.04.004 help (`A3200.chm`, topic "ASCII Command Interface"
 
 ```python
 scans.aerotech_cont_grid(dev.aerox, 0, 1, 0.01, dev.aeroy, 0, 1, 0.1, exp_time=0.01,
-                         trigger_source="ddg", settle_time=0.15)
+                         trigger_source="ddg", settle_time=0.3)
 ```
 
 - fast axis X moves continuously at `fast_step_size / exp_time`, the step axis Y steps between
@@ -95,8 +95,10 @@ scans.aerotech_cont_grid(dev.aerox, 0, 1, 0.01, dev.aeroy, 0, 1, 0.1, exp_time=0
   output (not tested). The PandA has the same choice as device config `trigger_edge`
   (default `"rising"`, opto wiring).
 - run-up / run-out = twice the ramp distance `v^2 / (2 * acceleration)` + `v * settle_time`
-  (default 0.15 s, measured 2026-10-28: the following error needs ~100-200 ms after the
-  ramp at 1 mm/s / 10 mm/s^2). In a snake the run-out of a line is the run-up of the next.
+  (default 0.3 s since 2026-11-03, was 0.15 s; measured 2026-10-28: the following error
+  needs ~100-200 ms after the ramp at 1 mm/s / 10 mm/s^2; at 20 mm/s 0.15 s left a
+  forward / backward offset that changed between repeats, see the trigger calibration in
+  `docs/developer/aerotech_stages.md`). In a snake the run-out of a line is the run-up of the next.
 - at the turnarounds X does not wait for In Position (`wait_in_position` signal off during
   the scan); X reaches 1 um within ~60 ms but 10 nm only after 0.5-2 s
 - positions are always captured: device `aerocapture` (must be enabled) latches the

csaxs_bec/scans/scans_v4/aerotech_cont_grid.py

@@ -60,17 +60,22 @@ centres of the fast axis (deviation from the nominal cell centres per direction,
 backward - forward offset and its slope along the line) and of the line times.
 
 Trigger calibration: in a snake the backward exposure centres are offset from the forward
-ones, by an amount that depends on the scan velocity (DDG1 at the beamline, 2026-11-03:
-backward - forward +2 um at 2 mm/s, +19 um at 10 mm/s, +26 um at 20 mm/s, -9 um at
-40 mm/s; independent of exp_time within the scatter). With trigger_calibration=True the
-scan shifts the trigger positions by the calibrated offset h (half of backward - forward):
+ones, by an amount that depends on the scan velocity and the settling (DDG1 at the
+beamline, 2026-11-03, backward - forward, independent of exp_time within the scatter):
+with settle_time 0.15 s +2 / +19 / +26 / -9 um at 2 / 10 / 20 / 40 mm/s, at 20 mm/s
+varying by 10 um between repeats (the oscillation after the ramp has not decayed); with
+0.3 s -3 / -8 / -3 / -1 um, at 20 mm/s within 3 um. With trigger_calibration=True the scan
+shifts the trigger positions by the calibrated offset h (half of backward - forward):
 trigger_low by +h, trigger_high by -h, so that the centres of both directions are on the
 same cells. h comes from the table userParameter "trigger_calibration" of the fast axis
-(device config, a list of {velocity, offset [um]}): linear between the calibrated
-velocities, proportional to the velocity below the lowest one, no correction above the
-highest one. The statistics at the end show the applied offset and the table entry this
-scan measured (applied + half of the remaining backward - forward), for new or better
-calibration points; the applied offset is stored in the scan info (trigger_offset).
+(device config, a list of {velocity, settle_time, offset [um]}, optionally
+minimal_run_up: true): only entries of the scan's settle_time and run-up mode, linear
+between their velocities, proportional to the velocity below the lowest one, no
+correction above the highest one or without entries. The statistics at the end show the
+applied offset and the table entry this scan measured (applied + half of the remaining
+backward - forward); the trigger shift moves backward - forward by about 1.15 x 2h, so
+take new entries from scans with the calibration applied. The applied offset is stored in
+the scan info (trigger_offset).
 
 Run-up / run-out of each line: twice the ramp distance v^2 / (2 * acceleration) + v *
 settle_time beyond the trigger positions. The whole range including run-up / run-out is
@@ -85,9 +90,11 @@ and the step axis starts; the next line starts when both have arrived.
 minimal_run_up=True reduces the run-up / run-out to the ramp distance alone (no margin, no
 settle_time; at least MIN_RUN_UP = 0.01 so the axis never turns at a PSO window edge), for data analysis that works with unevenly spaced (captured) positions.
 
-settle_time stays 0.15 s also at high speed: at 20 mm/s / 500 mm/s^2 shorter values
-(0.08 / 0.04 / 0.02 / 0 s) gave backward - forward offsets of -24 / +10 / +4.5 / -21 um,
-a ~10 Hz oscillation after the ramp that decays within ~0.15 s (measured 2026-10-29).
+settle_time is 0.3 s by default: at 20 mm/s / 500 mm/s^2 shorter values (0.08 / 0.04 /
+0.02 / 0 s) gave backward - forward offsets of -24 / +10 / +4.5 / -21 um, a ~10 Hz
+oscillation after the ramp (measured 2026-10-29); at the beamline (2026-11-03) 0.15 s
+still left +18 .. +28 um at 20 mm/s, different in every repeat, 0.3 s -2 .. -5 um. 0.3 s
+costs ~0.22-0.28 s per line compared with 0.15 s.
 
 Scan procedure:
     - prepare_scan
@@ -137,14 +144,23 @@ DDG_EXT_FALLING_EDGE = 2
 DDG_SINGLE_SHOT = 5
 
 
-def interpolate_trigger_offset(table: list[dict], velocity: float) -> float | None:
+def interpolate_trigger_offset(
+    table: list[dict], velocity: float, settle_time: float, minimal_run_up: bool = False
+) -> float | None:
     """
     Trigger offset h [um] for a scan velocity from a calibration table of
-    {"velocity": ..., "offset": ...} entries: linear between the calibrated velocities,
-    proportional to the velocity below the lowest one (no offset at rest), None above the
-    highest one or without a table.
+    {"velocity": ..., "settle_time": ..., "offset": ...} entries (optionally
+    "minimal_run_up": True). Only the entries of the given settle_time and run-up mode are
+    used: linear between their velocities, proportional to the velocity below the lowest
+    one (no offset at rest), None above the highest one or without entries.
     """
-    points = sorted((float(e["velocity"]), float(e["offset"])) for e in table)
+    points = sorted(
+        (float(e["velocity"]), float(e["offset"]))
+        for e in table
+        if bool(e.get("minimal_run_up", False)) == bool(minimal_run_up)
+        and e.get("settle_time") is not None
+        and np.isclose(float(e["settle_time"]), settle_time)
+    )
     if not points:
         return None
     velocities, offsets = zip(*points)
@@ -233,7 +249,7 @@ class AerotechContGrid(ScanBase):
         trigger_edge: str = "rising",
         trigger_device: str | None = None,
         shutter_delay: float | None = None,
-        settle_time: float = 0.15,
+        settle_time: float = 0.3,
         minimal_run_up: bool = False,
         capture_device: str = "aerocapture",
         trigger_calibration: bool = True,
@@ -259,10 +275,10 @@ class AerotechContGrid(ScanBase):
             trigger_edge (str): DDG1 external trigger edge, "rising" (opto output, J7-6/7) or "falling" (open collector / differential output). Default is "rising".
             trigger_device (str | None): Name of the PandaLineTrigger device for trigger_source="panda"; ignored otherwise. Default is None ("pandatrig").
             shutter_delay (float | None): Delay from the line trigger to the first exposure (time for the shutter to open) in s. Default: the value of the trigger hardware (DDG1 / PandA shutter-to-open delay, 8 ms), 0 without trigger hardware. With "ddg" a given value is set on DDG1 for this scan and restored afterwards.
-            settle_time (float): Run-up at constant velocity before the trigger position, in s. Default is 0.15.
+            settle_time (float): Run-up at constant velocity before the trigger position, in s; the trigger calibration is per settle_time. Default is 0.3.
             minimal_run_up (bool): Run-up / run-out only the ramp distance v^2 / (2 * acceleration), at least 0.01, without the 2x margin and settle_time: faster lines, but the exposures start right after the ramp while the fast axis may still oscillate, so the captured positions are less evenly spaced. Default is False.
             capture_device (str): Name of the A3200PositionCapture device (always used). Default is "aerocapture".
-            trigger_calibration (bool): Shift the trigger positions by the calibrated offset of the fast axis for the scan velocity, so that forward and backward exposure centres are on the same cells (userParameter "trigger_calibration"). False for training scans of new calibration points. Default is True.
+            trigger_calibration (bool): Shift the trigger positions by the calibrated offset of the fast axis for the scan velocity, settle_time and run-up mode, so that forward and backward exposure centres are on the same cells (userParameter "trigger_calibration"). False for training scans of new calibration points. Default is True.
 
         Returns:
             ScanReport
@@ -665,11 +681,15 @@ class AerotechContGrid(ScanBase):
         table = (
             parameters.get(TRIGGER_CALIBRATION_PARAMETER) if isinstance(parameters, dict) else None
         )
-        offset = interpolate_trigger_offset(table or [], velocity)
+        offset = interpolate_trigger_offset(
+            table or [], velocity, self.settle_time, self.minimal_run_up
+        )
         if offset is None:
             logger.warning(
                 f"aerotech_cont_grid: no trigger calibration of {self.fast_axis.name} for"
-                f" {velocity:g} / s, forward and backward exposures are not aligned."
+                f" {velocity:g} / s, settle_time {self.settle_time:g} s"
+                f"{', minimal run-up' if self.minimal_run_up else ''}: forward and backward"
+                " exposures are not aligned."
             )
         return offset
 
@@ -681,12 +701,14 @@ class AerotechContGrid(ScanBase):
         if applied is not None:
             state = f"offset {applied:+.2f} um applied"
         elif self.trigger_calibration:
-            state = "no calibration for this velocity"
+            state = "no calibration for this velocity / settle_time"
         else:
             state = "off"
+        mode = ", minimal_run_up: true" if self.minimal_run_up else ""
         return (
             f"  trigger calibration [um]: {state}; this scan measured"
-            f" {{velocity: {line['target_velocity']:g}, offset: {measured:.2f}}}"
+            f" {{velocity: {line['target_velocity']:g}, settle_time: {self.settle_time:g},"
+            f" offset: {measured:.2f}{mode}}}"
         )
 
     def _trigger_window(self, line_index: int) -> tuple[float, float]:

docs/developer/aerotech_stages.md

@@ -99,10 +99,12 @@ trigger_high = exposure_end + v * shutter_delay - v * readout_time / 2   (backwa
 ### Run-up, settling and limits
 
 Run-up / run-out per side: `2 * v^2 / (2 * acceleration) + v * settle_time`. In a
-snake the run-out of a line is the run-up of the next. `settle_time` stays 0.15 s: at
-20 mm/s and 500 mm/s^2 there is a ~10 Hz oscillation after the ramp that decays within
-~0.15 s; shorter values (0.08 / 0.04 / 0.02 / 0 s) gave backward minus forward offsets
-of -24 / +10 / +4.5 / -21 um (2026-10-29).
+snake the run-out of a line is the run-up of the next. `settle_time` is 0.3 s by
+default (was 0.15 s until 2026-11-03): at 20 mm/s and 500 mm/s^2 there is a ~10 Hz
+oscillation after the ramp; shorter values (0.08 / 0.04 / 0.02 / 0 s) gave backward
+minus forward offsets of -24 / +10 / +4.5 / -21 um (2026-10-29), and at the beamline
+0.15 s left +18 .. +28 um at 20 mm/s, different in every repeat, 0.3 s -2 .. -5 um
+(2026-11-03). 0.3 s costs ~0.22-0.28 s per line.
 
 `prepare_scan` checks the whole travel range including run-up, and the Y range,
 against the soft limits before anything moves. The message separates the scan range
@@ -110,26 +112,38 @@ from the run-up so a user can see why a scan inside the limits does not fit.
 
 ### Trigger calibration
 
-The backward exposure centres are offset from the forward ones by a velocity
-dependent amount (DDG1 at the beamline, 2026-11-03, scans 4897-4917: backward minus
-forward +0.9 / +2.1 / +9.3 / +18.6 / +25.7 / +3.0 / -8.8 um at 1 / 2 / 4 / 10 / 20 / 30 /
-40 mm/s; 5 vs 20 ms exposure time the same within the scatter; repeats within 0.1 um
-at 1 mm/s and 0.2 um at 40 mm/s). It is not a constant delay: about 0.5-1 ms up to
-20 mm/s, then it falls and changes sign. With the PandA (2026-10-29) it was below 1 um at
-20 mm/s.
+The backward exposure centres are offset from the forward ones by an amount that
+depends on the velocity and the settling (DDG1 at the beamline, 2026-11-03, backward
+minus forward; 5 vs 20 ms exposure time the same within the scatter):
+
+| v [mm/s] | 1 | 2 | 4 | 10 | 20 | 30 | 40 |
+|---|---|---|---|---|---|---|---|
+| settle 0.15 s | +0.9 | +2.1 | +9.3 | +18.6 | +18 .. +28 | +3.0 | -8.8 |
+| settle 0.3 s | | -3.2 | -5.8 | -8.2 | -4.6 / -1.7 | -1.1 / -1.6 | -0.7 |
+
+With 0.15 s it repeats within 0.5 um up to 10 mm/s and at 40 mm/s, but not at 20 mm/s:
+the oscillation after the ramp has not decayed. It is not a constant delay (about
+0.5-1 ms up to 20 mm/s at 0.15 s, then it changes sign). With the PandA (2026-10-29) it
+was below 1 um at 20 mm/s.
 
 `trigger_calibration=True` (default) shifts `trigger_low` by `+h` and `trigger_high`
 by `-h` (later in the direction of motion), with `h` = half of backward minus forward
 from the table `userParameter.trigger_calibration` of the fast axis
-(`interpolate_trigger_offset`: linear between the points, proportional to the velocity
-below the lowest one, `None` above the highest one). The device config is read from
-`_config` of the fast axis (the scan server's devices are `DeviceBaseWithConfig`,
-whose `user_parameter` reads the same). The applied `h` is in the scan info
-(`trigger_offset`, with `scan_velocity`); the statistics print the entry the scan
-measured, `h_applied + (backward - forward) / 2`. Only the mean offset per direction is
-corrected; the variation along the line at high velocity (oscillation after the ramp)
-remains. The calibration is for the configured acceleration (500 mm/s^2) and
-`settle_time` 0.15 s.
+(`interpolate_trigger_offset`): only the entries of the scan's `settle_time` (and
+`minimal_run_up: true` entries only with `minimal_run_up`), linear between their
+velocities, proportional to the velocity below the lowest one, `None` above the
+highest one or without entries. The device config is read from `_config` of the fast
+axis (the scan server's devices are `DeviceBaseWithConfig`, whose `user_parameter`
+reads the same). The applied `h` is in the scan info (`trigger_offset`, with
+`scan_velocity`); the statistics print the entry the scan measured,
+`h_applied + (backward - forward) / 2`.
+
+The shift acts slightly more than 1:1: backward minus forward changed by about
+1.15 x 2h at 8 and 10 mm/s. New entries are therefore taken from scans with the
+calibration applied, or as the zero crossing of two calibrated scans. Only the mean
+offset per direction is corrected; the variation along the line (rms 3-6 um per
+direction from 10 mm/s on, peak-to-peak 13-16 um) does not change with the settle time
+or the trigger shift. The calibration is for the configured acceleration (500 mm/s^2).
 
 ### Step during the run-out
 

docs/user/saxs/aerotech_stages/aerotech_stages.md

@@ -71,7 +71,7 @@ scans.aerotech_cont_grid(
 | `exp_time` | exposure time per cell in s; the X velocity is `fast_step_size / exp_time` |
 | `relative` | `True`: positions are offsets from the current X / Y position, and both axes return there after the scan |
 | `trigger_source` | `"ddg"` (default, beamline timing) |
-| `settle_time` | constant velocity time before the first exposure of a line, default 0.15 s (leave it) |
+| `settle_time` | constant velocity time before the first exposure of a line, default 0.3 s; 0.15 s is faster but less precise at high velocity (see below) |
 | `shutter_delay` | time from the line trigger to the first exposure, for the shutter to open. Default: the value of DDG1 (8 ms); a given value is set on DDG1 for this scan only |
 | `minimal_run_up` | `True`: run-up only the acceleration distance (at least 10 um), no settling. Faster lines, but the exposures are less evenly spaced; use the captured positions. Default `False` |
 | `trigger_calibration` | `True` (default): forward and backward lines are aligned with the calibration of `aerox` (see below). `False` only for training scans of new calibration points |
@@ -101,42 +101,54 @@ colours alternate in Y by one line spacing.
 ### Forward and backward alignment (trigger calibration)
 
 Without correction the backward lines are offset from the forward lines along X, by
-an amount that depends on the scan velocity (measured 2026-11-03 at the beamline,
-backward minus forward):
+an amount that depends on the scan velocity and on the settle time (measured
+2026-11-03 at the beamline, backward minus forward in um):
 
-| Velocity [mm/s] | 0.5 | 1 | 2 | 4 | 8 | 10 | 20 | 30 | 40 |
-|---|---|---|---|---|---|---|---|---|---|
-| Offset [um] | -0.6 | +0.9 | +2.1 | +9.3 | +15.8 | +18.6 | +25.7 | +3.0 | -8.8 |
+| Velocity [mm/s] | 2 | 4 | 10 | 20 | 30 | 40 |
+|---|---|---|---|---|---|---|
+| `settle_time=0.3` (default) | -3.2 | -5.8 | -8.2 | -4.6 / -1.7 | -1.1 / -1.6 | -0.7 |
+| `settle_time=0.15` | +2.1 | +9.3 | +18.6 | +18 .. +28 | +3.0 | -8.8 |
 
-It does not depend on the exposure time and is reproducible. The scan therefore moves
-the trigger positions of both directions by half of it, so that forward and backward
-exposures land on the same cells. The calibration is a table in the device config of
-`aerox` (`userParameter` `trigger_calibration`, velocity and offset), interpolated
-between the calibrated velocities. Above 40 mm/s there is no calibration: the scan
-runs uncorrected and says so in its summary.
+It does not depend on the exposure time. With 0.15 s the offset is larger and, from
+20 mm/s on, different in every repeat: X still oscillates after the acceleration when
+the exposures start. With 0.3 s it is small and reproducible; this costs about 0.22 s
+more per line (about 27 % on fast scans, negligible on slow high-resolution scans).
+
+The scan moves the trigger positions of both directions by half of the offset, so that
+forward and backward exposures land on the same cells. The calibration is a table in
+the device config of `aerox` (`userParameter` `trigger_calibration`: velocity, settle
+time, offset), with entries for 0.3 s and 0.15 s. A scan uses the entries of its own
+`settle_time`, interpolated between the calibrated velocities. Without entries for its
+settle time, with `minimal_run_up=True`, or above 40 mm/s the scan runs uncorrected and
+says so in its summary.
 
 The summary of each scan shows the applied offset and the table entry the scan
 measured, for example:
 
 ```text
-trigger calibration [um]: offset +9.29 um applied; this scan measured {velocity: 10, offset: 9.41}
+trigger calibration [um]: offset -4.09 um applied; this scan measured {velocity: 10, settle_time: 0.3, offset: -3.80}
 ```
 
 A calibrated scan should show a backward - forward mean of a few um at most. What
-remains along a line at high velocity (rms per direction 4-7 um from 10 mm/s on) is
-the motion of X and cannot be corrected by the trigger.
+remains along a line at high velocity (rms per direction 3-6 um from 10 mm/s on) is
+the motion of X and cannot be corrected by the trigger or the settle time.
 
-**New calibration point** (other acceleration, settle time, or a velocity outside the
-table): run a short scan with `trigger_calibration=False` at that velocity, for example
-X +-1 mm, 6 lines:
+**Other settings:** `settle_time` and `minimal_run_up` stay available as options. For a
+faster scan use `settle_time=0.15` (calibrated, but at 20-30 mm/s only to about
++-5 um). For any other settle time or `minimal_run_up=True` there is no calibration:
+the lines are offset as in the table above.
+
+**New calibration point** (other settle time, acceleration, or velocity): run a short
+scan at that velocity and settle time, for example X +-1 mm, 6 lines:
 
 ```python
 scans.aerotech_cont_grid(dev.aerox, -1.0, 1.0, 0.1, dev.aeroy, -0.1, 0.1, 0.04, exp_time=0.005, relative=True, trigger_calibration=False)  # 20 mm/s
 ```
 
-and add the printed `{velocity: ..., offset: ...}` to `trigger_calibration` of `aerox`
-in `sastt_aerotech.yaml` (ask the beamline staff to commit it). For a quick test without
-a config change: `dev.aerox.update_user_parameter({...})`.
+and add the printed `{velocity: ..., settle_time: ..., offset: ...}` to
+`trigger_calibration` of `aerox` in `sastt_aerotech.yaml` (ask the beamline staff to
+commit it). Then repeat the scan with the calibration on and take the entry it prints:
+the trigger shift acts slightly more than 1:1, so the second value is the better one.
 
 ### Examples
 
@@ -164,12 +176,12 @@ scans.aerotech_cont_grid(dev.aerox, -1.75, 1.75, 0.01, dev.aeroy, -1.5, 1.5, 0.0
 ### Run-up and soft limits
 
 To pass through the field of view at constant velocity, X travels beyond it on both
-sides: twice the acceleration distance `v^2 / (2 * 500 mm/s^2)` plus `v * 0.15 s` to
-settle. The run-up per side is
+sides: twice the acceleration distance `v^2 / (2 * 500 mm/s^2)` plus `v * settle_time`
+(0.3 s) to settle. The run-up per side is
 
-| Scan velocity | 0.5 mm/s | 2 mm/s | 20 mm/s |
-|---------------|----------|--------|---------|
-| Run-up per side | 0.08 mm | 0.31 mm | 3.8 mm |
+| Scan velocity | 0.5 mm/s | 2 mm/s | 20 mm/s | 40 mm/s |
+|---------------|----------|--------|---------|---------|
+| Run-up per side | 0.15 mm | 0.61 mm | 6.8 mm | 15.2 mm |
 
 Before anything moves, the scan checks the whole range including the run-up against
 the soft limits. If it does not fit, the scan stops with a message that says by how
@@ -177,7 +189,7 @@ much and on which side, for example
 
 ```text
 The scan range of aerox 146.5000 .. 150.0000 is within the limits, but aerox needs
-3.8400 extra travel on each side to accelerate to the scan velocity 20 / s and settle
+6.9580 extra travel on each side to accelerate to the scan velocity 20 / s and settle
 ...
 ```
 

tests/tests_scans/test_aerotech_cont_grid_scan.py

@@ -206,10 +206,21 @@ def test_velocity_above_configured_velocity_aborts(v4_scan_assembler, device_man
 
 
 def test_scan_faster_than_move_velocity_restores_move_velocity(v4_scan_assembler, device_manager):
-    # 20 mm/s scan, move velocity 10 mm/s, max_scan_velocity 50 mm/s
+    # 20 mm/s scan, move velocity 10 mm/s, max_scan_velocity 50 mm/s; settle_time 0.15 s
+    # so that the run-up fits the +-10 limits at 100 mm/s^2
     _add_devices(device_manager, max_scan_velocity=50.0)
     scan = v4_scan_assembler(
-        "aerotech_cont_grid", "aerox", -1.0, 1.0, 0.1, "aeroy", -2.0, 2.0, 2, exp_time=0.005
+        "aerotech_cont_grid",
+        "aerox",
+        -1.0,
+        1.0,
+        0.1,
+        "aeroy",
+        -2.0,
+        2.0,
+        2,
+        exp_time=0.005,
+        settle_time=0.15,
     )
     scan.prepare_scan()
     assert scan._line["target_velocity"] == pytest.approx(20.0)
@@ -222,7 +233,7 @@ def test_scan_faster_than_move_velocity_restores_move_velocity(v4_scan_assembler
 
 
 def test_run_up_outside_limits_aborts(v4_scan_assembler, device_manager):
-    # 20 mm/s at 100 mm/s^2: 2 * 2 mm ramp + 3 mm settle run-up, beyond the +-10 limits
+    # 20 mm/s at 100 mm/s^2: 2 * 2 mm ramp + 6 mm settle run-up, beyond the +-10 limits
     _add_devices(device_manager, max_scan_velocity=50.0)
     scan = v4_scan_assembler(
         "aerotech_cont_grid", "aerox", -4.0, 4.0, 0.1, "aeroy", -2.0, 2.0, 2, exp_time=0.005
@@ -691,9 +702,12 @@ def test_progress_bar_from_the_capture_without_mcs(v4_scan_assembler, device_man
 
 
 CALIBRATION = [
-    {"velocity": 10.0, "offset": 4.0},
-    {"velocity": 2.0, "offset": 1.0},
-    {"velocity": 20.0, "offset": -2.0},
+    {"velocity": 10.0, "settle_time": 0.3, "offset": 4.0},
+    {"velocity": 2.0, "settle_time": 0.3, "offset": 1.0},
+    {"velocity": 20.0, "settle_time": 0.3, "offset": -2.0},
+    # other settling and run-up mode: never mixed into the 0.3 s entries
+    {"velocity": 5.0, "settle_time": 0.15, "offset": 50.0},
+    {"velocity": 5.0, "settle_time": 0.3, "minimal_run_up": True, "offset": 60.0},
 ]
 
 
@@ -711,7 +725,7 @@ CALIBRATION = [
     ],
 )
 def test_interpolate_trigger_offset(velocity, expected):
-    offset = interpolate_trigger_offset(CALIBRATION, velocity)
+    offset = interpolate_trigger_offset(CALIBRATION, velocity, 0.3)
     if expected is None:
         assert offset is None
     else:
@@ -719,7 +733,20 @@ def test_interpolate_trigger_offset(velocity, expected):
 
 
 def test_interpolate_trigger_offset_without_table():
-    assert interpolate_trigger_offset([], 5.0) is None
+    assert interpolate_trigger_offset([], 5.0, 0.3) is None
+
+
+def test_interpolate_trigger_offset_per_settle_time_and_run_up_mode():
+    assert interpolate_trigger_offset(CALIBRATION, 5.0, 0.15) == pytest.approx(50.0)
+    assert interpolate_trigger_offset(CALIBRATION, 5.0, 0.3, minimal_run_up=True) == 60.0
+    assert interpolate_trigger_offset(CALIBRATION, 5.0, 0.2) is None
+    # entries without settle_time are not used
+    assert interpolate_trigger_offset([{"velocity": 5.0, "offset": 1.0}], 5.0, 0.3) is None
+
+
+def test_default_settle_time_is_0p3(v4_scan_assembler, device_manager):
+    scan = _assemble(v4_scan_assembler, device_manager)
+    assert scan.settle_time == 0.3
 
 
 def _trigger_positions(v4_scan_assembler, device_manager, **kwargs):
@@ -771,7 +798,9 @@ def test_calibration_summary_reports_the_measured_entry(v4_scan_assembler, devic
     # applied 2.125 um, still +1 um backward - forward: the entry is 2.625 um
     summary = scan._calibration_summary(1.0)
     assert "offset +2.12 um applied" in summary
-    assert "{velocity: 5, offset: 2.62}" in summary
+    assert "{velocity: 5, settle_time: 0.3, offset: 2.62}" in summary
     scan.trigger_calibration = False
     scan._line["trigger_offset"] = None
-    assert "off; this scan measured {velocity: 5, offset: 0.50}" in scan._calibration_summary(1.0)
+    assert "off; this scan measured {velocity: 5, settle_time: 0.3, offset: 0.50}" in (
+        scan._calibration_summary(1.0)
+    )