Estimating Combined Servo and Galvo Motion Accuracy

William Land II

Business Development Manager

In modern motion control systems, the primary figure of merit is often the global accuracy that may be achieved on the workpiece surface. However, when configuring the motion control architecture for more advanced features, such as Infinite Field of View (IFOV), characterizing each element of the systemic errors can be very challenging. This white paper details how to characterize the constituent error elements and suggests a means of predicting overall system errors for IFOV-enabled motion control platforms.

Testing Overview

The test procedure outlined demonstrates the accurate overlaying of a servo motion assembly’s coordinate frame with a coincident scanner coordinate frame within a laser processing station meant for IFOV operation. The general procedure involves the independent calibration of both the servo system (in this case, an Aerotech ALS50045WB/ALS50030-XY stage assembly) and the scanner system (Aerotech’s AGV-14HP-1064-100T), followed by the alignment of the two calibrated coordinate frames. The ALS5000-XY assembly is calibrated via dual-axis interferometry, while the AGV scanner is calibrated via a mark-and-measure calibration procedure that utilizes a machine vision inspection station with a resolution of 0.86 µm per pixel. The coordinate frames are aligned via a mark-and-measure procedure as well, where the measured necessary coordinate rotation transformation is carried out on the AGV’s axes. Finally, the alignment and congruency of the two coordinate frames is checked via a random combined motion test, which is discussed in detail below.

List of Hardware

Laser Processing Station
- Steel weldment
- Granite base assembly
  - Granite risers and bridge
  - Mounted on vibration isolators
- ALS50045WB/ALS50030-XY stage assembly
  - Precision ground vacuum chuck
  - Ndrive HPe PWM digital amplifier x2
- PRO165-05MM-250-BMS Z stage, mounted vertically on bridge
  - Motor brake
  - Ndrive HPe PWM digital amplifier
- AGV-14HP-1064-100T scanner, mounted on PRO165 carriage
  - Sill S4LFT5100/126 100 mm telecentric F-theta lens
  - Nmark CLS high-performance galvo controller
- Multi-mode, 100 W, IR fiber laser with collimator
- 19” rack-mounted PC
High Precision Inspection Station
- Steel weldment
- Granite base assembly
  - Granite risers and bridge
  - Mounted on passive air isolation system
- Aluminum framed, ventilated enclosure
- ABL15030WB/ABL15030-XY stage assembly
  - Precision ground vacuum chuck
  - Ndrive HLe linear digital amplifiers x2
- PRO165-05MM-150-BMS Z stage, mounted vertically on bridge
  - Motor brake
  - Custom cable management system
  - Ndrive CP10 PWM digital amplifier
- High resolution 5 MP digital machine vision system, mounted on PRO165 carriage
  - 40 mm field flattening lens, 2 mm FOV
  - Shape and shade recognition search algorithm software
  - Water cooled, high stiffness mounting bracket
- PC workstation
Marking Substrates
- Flat to .0001 inches over full marking area
- Marking surface parallel to mounting surface to .0005 inches
- Surface finish held to 60-40 scratch-dig

Grid Marking Parameters

Laser Parameters
- 5.0 W average laser output power
- 50.0 μs pulse duration
Mark Parameters
- 1.0 mm square, cross fiducials
- 1.0 mm/s marking speed
- 8.0 µm pulse pitch
Grid Parameters
- Independent AGV grid
  - 39.0 x 39.0 mm grid size
  - 1.95 mm two dimensional mark pitch
- Independent servo grid
  - 135.0 x 135.0 mm grid size
  - 6.75 mm two dimensional mark pitch
- Random combined motion grid
  - 96.0 x 96.0 mm grid size
  - 4.80 mm two-dimensional mark pitch

Random Combined Motion Grid Marking Procedure

The following series of illustrations demonstrate how the “Random Combined Motion Grid” is created using the combined motion of the servo and scanner coordinate frames. They show the relative positions of the servo axes’ position, the AGV axes’ field of view, and the actual grid to be marked by the random combined motion test program. The gray outline represents the ALS5000-XY assembly. The black square represents the tabletop of the ALS assembly’s upper axis, where the substrate is mounted. The black dots represent the grid to be marked. The grid’s overall size – the dashed blue square – is equal to the size of the servo stages’ total field of view minus the AGV axes’ total field of view, which in this case is 96.0 x 96.0 mm. The red dashed square represents the total field of view of the AGV scanner, which in this case is 39.0 x 39.0 mm. The red dot within the AGV’s field of view represents the location of the laser spot on the substrate.

Procedure

1. The servo and AGV axes begin at their home position, with the substrate centered underneath the AGV’s field of view (See Figure 1).

Figure 1. Servo and AGV axes start at home position with the substrate centered.

2. The servo axes move the first mark location on the substrate underneath the laser spot with the AGV axes at their home position (See Figure 2).

Figure 2. Servo axes move first mark location underneath AGV.

3. The servo axes move a random percent of the size of the AGV’s field of view in a random direction (See Figure 3).

Figure 3. Servo axes move in a random fashion a percentage of the AGV’s field of view.

4. The AGV axes then perform the same random move performed by the servo axes in step 3. If the coordinate frames are perfectly aligned and congruent, the motions of the two sets of axes will null each other leaving the laser spot in the correct location within the grid for a mark to be placed (See Figure 4).

Figure 4. AGV axes null the random motion of the servo axes, and create a mark. The mark will hopefully fall on the correct location within the grid based on the alignment and congruency of the two coordinate frames.

After the creation of the mark, the servo and AGV axes undo the random motion, bringing the mark location back underneath the center of the AGV’s field of view (See Figure 5).

Figure 5. Servo and AGV axes de-null after the random motion and mark creation, bringing mark location #1 back underneath the center of the AGV’s field of view.

6. The servo axes move the second mark location on the substrate underneath the laser spot with the AGV axes at their home position (See Figure 6).

Figure 6. Servo axes move mark location #2 underneath the AGV.

7. The servo axes repeat Step 3, moving a percentage of the AGV’s field of view in a random direction (See Figure 7).

Figure 7. Servo axes move in a random fashion a percentage of the AGV’s field of view.

8. The AGV axes repeat Step 4, where they perform the same random motion as the servo stages in Step 7, nulling the motion and placing the laser spot correctly at mark location #2 (See Figure 8).

Figure 8. AGV axes null the random motion of the servo axes, and create a mark. The mark will hopefully fall on the correct location within the grid based on the alignment and congruency of the two coordinate frames.

9. After the creation of the mark, the servo and AGV axes undo the random motion, bringing the mark location back underneath the center of the AGV’s field of view (See Figure 9).

Figure 9. Servo and AGV axes de-null after the random motion and mark creation, bringing mark location #2 back underneath the center of the AGV’s field of view.

10. The process is repeated until the grid is completed. The resultant grid is then measured with the Inspection Station described above.

Results of Measured Grids

Table 1. Summary of output grid accuracy when manufactured solely by galvo, by servo and through the combination of randomized servo and galvo motion.

Conclusions

An examination of the measurement results shows that overall accuracy and error pattern of the random combined motion grid is largely dictated by the servo axes. This is not unexpected, as the servo axes account for a much larger portion of the overall movement of the random combined motion grid, and they are also the largest contributor of error in the system. The 4*Sigma error distribution for each grid and the vector error are also of particular interest, as they suggest that an RSS (Root Sum of Squares) combination of scanner and servo axis errors can be used to approximate the combined error of the two independent coordinate frames. An exception to this generality is observed in the Y Axis Accuracy, where the independent servo grid’s error was actually larger than that of the combined motion grid’s error. However, a distinct and similar slope in error of both the X and Y axes (shown in the plots of the independent servo grid’s measurement that follows) suggests that this might be nothing more than an artifact of the setup of the marking station itself – namely, a substrate that is out of parallelism tolerance can produce such an error pattern, and is a likely explanation for the higher 4*Sigma and peak-to-peak error of the Y-Axis in that particular grid. Regardless, the results of this testing allow for reasonably confident approximation of laser marking performance when using combined motion systems, and should help in the decision-making process as to which stages are appropriate given the goals of a given application.

While the information detailed within this study details a comprehensive and unambiguous means of identifying error contributions from both the servo and scanner motion sub-systems in an IFOV-enabled platform, other considerations and future study should be considered. Namely, the errors measured are purely static in nature, and do not account for dynamic influences any given platform might experience during practical laser processing applications. Furthermore, the sources of the static errors themselves might be investigated further to identify how they might be further mitigated or at least better understood. These topics will be addressed in future studies and technical papers by Aerotech.