Automap4 Operator Guide

Automap4 Operator Guide

This guide explains the automap4 main window, menus, and the standard workflow for S-curve discovery, PRBS identification, closed-loop supervisory control, recording, and reporting. The screenshots use the light theme because it prints and embeds better in documentation.

Top Toolbar / Status Bar

Figure 1: automap4 main window (light theme).

Primary control buttons

Status readouts

Menu Reference

File

Grades

The Grades menu lists all grade_*.npz files found in the configured grade directory. Selecting one loads it immediately. A grade carries the measured S-curve, the mapping matrix, the controller tuning (Q/R/S/T, aggressiveness, integral, max delta_u), and the actuator-level settings. The grade is the tuning carrier.

Control

View

Help

Preferences Dialog

Figure 2: Preferences dialog (Machine Configuration).

Preferences are split into Machine Configuration, General Settings, and Controller Tuning. Machine identity values (headbox width, actuator width, machine speed, paper length, max databoxes, max actuators, grade directory, ZMQ bind host/port) are site identity and are not changed by the preset buttons.

Machine Configuration

General Settings

Controller Tuning

Preset Buttons

S-Curve Discovery

The S-curve is the actuator-to-sheet alignment: which databox each actuator's response lands on. It is measured, never assumed. automap4 ships no default S-curve; PRBS is blocked until a measured curve exists.

Figure 3: S-Curve Discovery dialog.

  1. Open Control → S-Curve Discovery...
  2. The dialog suggests a bump set. Edit it by dragging, Shift+dragging, or double-clicking zones in the bar.
  3. On a noisy scanner use 7–8 % amplitude; on a clean scanner 5 % is usually enough. Press Run.
  4. If no transport-delay override is set, the session first measures the delay with a mid-sheet probe bump. Answer the skip-tau question if it appears.
  5. Watch the run: blue = queued, yellow = bumping, green = measured, orange = edge-clipped, red = no response. The lower chart shows the provisional 96-zone mapped profile.
  6. When the fit is verified, press Accept S-curve. This seeds the mapper, unlocks PRBS, and offers to save a seed grade.

PRBS Identification

PRBS refines the discovered S-curve and learns the true response shapes. Run 30–40 scans with a verified discovery seed; you do not need the full 511-scan sequence.

Figure 4: PRBS Identification dialog after the control gate opens.

  1. Make sure Preferences carries the discovery preset (or click Reset to Discovery Defaults again).
  2. Open Control → Toggle PRBS (Ctrl+Space). Set amplitude to 3 %, set the Control threshold (default 90 %), and click Start PRBS.
  3. Watch the Mapping Progress bar and the per-zone quality bars: grey = not learned, orange = learning, green = fully learned.
  4. When the progress bar turns green (above the Control threshold), you can click Stop PRBS or Stop PRBS & enable control.
  5. Post-PRBS tuning auto-applies. Save the grade with File → Save Grade or when prompted.

Closed-Loop Control

Figure 5: Main window in CONTROL mode.

  1. Load the grade saved from PRBS (Grades menu or File → Load Grade).
  2. Clear any manual zones. The mode badge should show CD control is allowed and Map is above the threshold.
  3. Click the CD control button (or press Ctrl+C). The first time, a dialog asks whether to record; choose or accept control_log.csv.
  4. The mode badge changes to CONTROL. The actuator chart shows setpoints moving as the controller flattens the profile.
  5. Watch the centre 2-sigma trend. If it is too jumpy, lower alpha; if too sluggish, raise alpha. Do not change R/S/T by hand unless auto-tune is OFF.
  6. To stop control, click the CD control button again or press Ctrl+C. When control stops, automap4 offers to save any mapping refinements made during control.

Mapping & Control Info

Figure 6: Mapping & Control Info dialog (Ctrl+I).

Mapping Info shows the S-curve provenance, mapping quality, learned counts, response width and gain, transport delay, and the current controller tuning. Use it after PRBS and after enabling control to sanity-check that width, tau, and the derived R/S/T are plausible.

Monitor Windows

Figure 7: Mapping Heatmap.

View → Mapping Heatmap visualises the learned mapping matrix. A solid diagonal band from top-left to bottom-right means the alignment between actuators and scanner positions is correct.

Figure 8: Mapped Actuator Profile window.

View → Mapped Actuator Profile shows the two 96-zone BWT profiles (simple zone average and PRBS-weighted) that automap4 sends back to the machine. Use it to compare the two mappings and as a cross-check for external controllers.

Figure 9: S-Curve Mapping window (F1 help available).

View → S-Curve Mapping shows the measured S-curve, the bumped zones ± measured sigma, and the fit readout.

Figure 10: PRBS Mapping window (F1 help available).

View → PRBS Mapping shows each actuator's PRBS-learned footprint coverage versus the discovery curve clamp band.

Figure 11: Wander Trend window.

Figure 12: 2-Sigma Trend window.

View → Wander Trend and View → 2-Sigma Trend track sheet shift and profile variability over time. Use them to judge whether the controller is holding the sheet steady or whether drift is physical (wander/MD disturbance).

Recording and Field Reports

Click Record to start logging. The default filename depends on the current mode:

The CSV contains one summary row per scan. The NPZ contains the full profiles, actuator positions, setpoints, edges, machine speed, timestamps, and the learned mapping matrix.

  1. Select View → Generate Field Report (PDF)...
  2. Choose the recorded CSV file.
  3. Choose where to save the PDF. The default name matches the CSV basename.
  4. automap4 builds a PDF containing the mapping heatmap, quality distribution, 2-sigma trend, wander trend, and tuning summary.

Setting up machinedriver.py

machinedriver.py is the bridge between the real paper machine (via STEC/Multiverse) and automap4 (via ZeroMQ). In the lab, modelstdcom.py replaces the real machine's Multiverse; machinedriver.py itself does not change.

  1. Launch machinedriver.py on a PC that can reach both the STEC server and automap4.
  2. In the Connection group enter the ZMQ host/port where automap4 is listening (default 5555) and the STEC host/port of the Multiverse broker.
  3. Open the Subscription Mapping table and map each field to the correct STEC subscription name (actuator_positions, actuator_setpoints, profile, left_edge, right_edge, machine_speed, scanner_direction). Set actuator_positions/profile/edges/speed/scanner_direction to to_zmq and actuator_setpoints to from_zmq.
  4. Click Save. machinedriver will auto-connect to automap4 and start forwarding scans.
  5. Use the Profile Orientation checkboxes if the scanner orientation needs reversal.

About automap4

Figure 13: About automap4 dialog.

Help → About automap4 shows the package version, git branch, commit, build date, origin, and exact checkout instructions to reproduce the running version. The same build metadata is available in machinedriver and modelstdcom.