Auto-align: level the sample on the DC bias levels from the camera window

The operator frames a good spot, confirms the two DC levels the detector
reads there, and the rig then measures its own tilt: step 1.5 mm either side
on X and then on Y, and tilt the platform until those levels come back.  The
correction that fixes an offset point is the correction that levels the whole
travel — height error and tilt effect are both proportional to the offset —
so the procedure ends by applying it and leaving it applied.

Both directions are measured from the same starting tilt and averaged, which
makes their disagreement a flatness read-out rather than something averaged
away silently.

core/auto_align.py holds the geometry and the search, Qt-free.  The three
T-axes' azimuths are the whole geometry: T1 lies along +X so it alone tilts
along X, and T0/T2 move as an equal-and-opposite pair to tilt along Y without
touching X (tilt_response derives that, and the tests pin it — an axis map
that drifts would still converge, on the wrong axis).  The search is a secant
null on the split-detector difference: probe once to learn what a microstep
is worth, sign included, then step at the null.  It refuses to servo on a
scope that has not re-triggered, escalates a probe that reads as no response
before calling an axis dead, and stops at a per-axis travel limit.

gui/align_bridge.py runs it on a worker thread; stopping is a threading.Event
rather than a queued command, because the worker is inside a long handler for
the whole run.  The camera window carries the button and the progress window,
and locks the scan panel and the jog pads while a run owns the stage.

Adds immediate MEAN measurements and an acquisition count to the scope
driver, and read_bias_mv to core/scope_inspect — the one scalar the
inspection state was missing.

KNOWN_ISSUES.md records what only the rig can settle: the probe step, the
travel limit, the hold current, and whether the piston the X phase applies
alongside its tilt matters.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Thomas Ales
2026-09-04 14:00:36 -05:00
parent 083cbdaa34
commit 6e8c1cb7a2
13 changed files with 1868 additions and 19 deletions
+107
View File
@@ -258,6 +258,9 @@ class FakeT3R:
self._t = trace
self.is_open = is_open
self._motion_completes = motion_completes
# Microsteps commanded per channel, so a test can read the tilt the
# platform ended up at rather than replaying the move trace.
self.positions = {ch: 0 for ch in range(4)}
def set_microstep(self, ch, micro):
self._t.record("t3r_set_microstep", ch, micro)
@@ -273,9 +276,113 @@ class FakeT3R:
return round(self.MOTOR_FULL_STEPS_PER_REV * microsteps * ratio
* angle_deg / 360.0)
def move(self, ch, steps, velocity, accel):
self._t.record("t3r_move", ch, steps)
self.positions[ch] += steps
def rotate_stage(self, angle_deg, microsteps, velocity, accel):
self._t.record("t3r_rotate", round(angle_deg, 6))
def wait_motion_done(self, ch, timeout):
self._t.record("t3r_wait_motion_done", ch)
return self._motion_completes
class FakeAlignRig:
"""A tilted sample on the tilt platform, as the DC levels would read it.
The detection beam is fixed and the stage carries the sample under it, so
the height error under the beam is the sample's slope times how far the
stage has moved off the reference point. The T-axes tilt the sample the
other way: their three heights define a plane, and its slope adds to the
sample's. Nulling the split-detector difference therefore means cancelling
the sample slope — which is exactly what an aligner has to work out.
The plane is fitted by least squares here, rather than reusing
core.auto_align's closed form, so the two are independent statements of
the same geometry.
``curvature_mv_per_mm2`` bends the surface: a curved sample needs opposite
corrections at +1.5 mm and -1.5 mm, which is the disagreement the
procedure is supposed to report instead of averaging away.
"""
# Actuator azimuths on the platform, in degrees from stage +X.
AZIMUTH_DEG = {0: 120.0, 1: 0.0, 2: 240.0}
def __init__(self, stage, t3r, ref_mm=(50.0, 40.0),
x_slope_mv_per_mm=40.0, y_slope_mv_per_mm=-25.0,
tilt_gain_mv_per_mm=0.2, base_mv=400.0,
curvature_mv_per_mm2=0.0, jitter_mv=0.0):
self._stage = stage
self._t3r = t3r
self.ref_mm = ref_mm
self.x_slope_mv_per_mm = x_slope_mv_per_mm
self.y_slope_mv_per_mm = y_slope_mv_per_mm
self.tilt_gain_mv_per_mm = tilt_gain_mv_per_mm
self.base_mv = base_mv
self.curvature_mv_per_mm2 = curvature_mv_per_mm2
self.jitter_mv = jitter_mv
self._reads = 0
# -- geometry -----------------------------------------------------------
def platform_tilt(self):
"""(x_tilt, y_tilt) of the plane through the three actuator heights."""
import numpy as np
rows, heights = [], []
for ch, azimuth in self.AZIMUTH_DEG.items():
theta = np.radians(azimuth)
rows.append([1.0, np.cos(theta), np.sin(theta)])
heights.append(float(self._t3r.positions[ch]))
_, x_tilt, y_tilt = np.linalg.lstsq(np.array(rows), np.array(heights),
rcond=None)[0]
return float(x_tilt), float(y_tilt)
def slopes_mv_per_mm(self):
"""The residual sample slope the beam sees, after the platform tilt."""
x_tilt, y_tilt = self.platform_tilt()
return (self.x_slope_mv_per_mm + self.tilt_gain_mv_per_mm * x_tilt,
self.y_slope_mv_per_mm + self.tilt_gain_mv_per_mm * y_tilt)
def difference_mv(self):
x_off = self._stage.positions[0] - self.ref_mm[0]
y_off = self._stage.positions[1] - self.ref_mm[1]
slope_x, slope_y = self.slopes_mv_per_mm()
return (slope_x * x_off + slope_y * y_off
+ self.curvature_mv_per_mm2 * (x_off ** 2 + y_off ** 2))
# -- what the scope reports ---------------------------------------------
def level_v(self, channel):
"""CH3 and CH4 as volts: the difference straddling a constant sum.
The sum is fixed because tilt steers the beam across the detector
rather than changing how much light comes back — so a nulled
difference does put both levels back where they were.
"""
self._reads += 1
# A deterministic alternating wobble, so a test can check the median
# of several reads is what keeps the loop stable.
jitter = self.jitter_mv * (1 if self._reads % 2 else -1)
half = 0.5 * self.difference_mv()
mv = self.base_mv + (half if channel == 3 else -half) + jitter
return mv / 1000.0
class FakeAlignScope(FakeScope):
"""FakeScope that also answers the DC measurements auto-align reads."""
def __init__(self, trace: Trace, rig: FakeAlignRig, samples_per_frame=8,
acquisitions_advance=True):
super().__init__(trace, samples_per_frame=samples_per_frame)
self._rig = rig
self._acq = 0
self._advance = acquisitions_advance
def measure_immediate(self, channel, measurement_type="MEAN"):
self._t.record("measure_immediate", channel, measurement_type)
return self._rig.level_v(channel)
def get_acquisition_count(self):
if self._advance:
self._acq += 1
return self._acq