White Paper on automating micro-optical assemblies

Precision and versatility: automating micro-optics alignment with the Meca500

WRITTEN BY:
Francois Séguin, President and Founder, XVP Photonics
Joseph Bélanger, Photonics Director, XVP Photonics
Sébastien Allen, Process Development Director, XVP Photonics

Bridging the automation gap

Optical alignment and packaging are often the dominant cost drivers in photonic product manufacturing, an issue that can be alleviated with process automation. However, setting up a fully automated assembly line at the launch of a new product is often unrealistic due to the significant cost and limited functionality of high-precision multi-axis optical positioners. Unfortunately, this can create a situation where introducing a new product on the market is compromised by lower profitability and difficult manufacturing ramp-ups. Due to the cost and functionality limitations of typical photonics motion actuators, it’s hard to come up with a plausible return on investment scenario for automation. New products must grow from initial semi-manual, operator-attended manufacturing before eventually achieving full commercial success and unlocking the budget needed for automation, a potentially risky path to scale up. XVP Photonics has identified an opportunity to bridge the gap betw een these two extremes with the launch of a low-cost, highly precise and compact six-axis robot: Mecademic’s Meca500. Offering a wide range of articulated motion and six degr ees of freedom, the Meca500 can perform many more tasks than just simple alignment, such as parts handling, glue dispensing, and moving the alignment site over a wide area to cover complete multi-component assemblies, all from a single actuation platform. In order to establish the range of addressable photonics applications using the Meca500, its performance was characterized optically and modelled in various alignment scenarios. The work reported in this paper aims to provide a definitive answer to the question: can the Meca500 be relied on to perform optical alignment and assembly tasks autonomously, even in the context of budget-constrained, low-volume product introduction?

Quantifying robot performance

Various optical measurement methods were used to quantify the Meca500’s performance. They were needed to establish how well the robot could repeatedly hit a target position from varied trajectories and, once on target, how finely it could increment and maintain this position for active optimization and fixation. Those measurements resulted in complementary information to the robot performance data already available from Mecademic. Beam centroid measurements in linear and angular space were performed (Figure 2 and Figure 3) to evaluate the Meca500-OBSI, a black version of the Meca500 designed for optimal light absorption and greater security in optical applications. In the first setup, a retro-illuminated single-mode optical fiber is manipulated in front of a stationary, high-power microscope system, providing high-resolution measurements of transverse linear positions. In the second setup, a retro-illuminated pre-collimated fiber assembly is presented at the input of a Fourier transform optical bench. This setup creates an image of the beam at infinity on the surface of a bare CMOS sensor, producing a highly resolved angle-to-displacement measurement insensitive to linear spatial jitter and repeatability of the beam position at the system input. Those techniques can’t provide values for the robot’s absolute accuracy, which typically requires a CMM-type instrument. However, they offer contactless nm/micro-radian scale resolution, owing to optical magnification and centroid detection.

Various photonic alignment applications presenting typical assembly challenges
a) Optical μ-bench assemblies
b) CMOS mini camera objective
lens alignment
c) Fiber optic array to integrated
optics waveguide
d) Laser diode collimation

Documents repeatability and shows recorded angular positions for 1000 consecutive approaches to the same target coordinates. Using a Rayleigh distribution fit, a
repeatability of 50uRad (0.003°) at 99.7% is obtained. Those numbers compound a contribution from the on-target stability. To measure stability, angular measurements from a stationary position are recorded consecutively 1000 times (Figure 5 - left) and analyzed again with a Rayleigh fit. The results show the robot holding position with a jitter of 15uRad (0.0009°) at 99.7%. Finally, to understand how finely the angular position can be incremented, irrespective of jitter or
repeatability, the angle was incremented by 1uRad from a set position, and each resulting angle was averaged over 400 measurements. Figure 6 shows that by removing jitter, the increments track within an error of 5uRad (0.0003°). Here, a Gaussian fit was used with a 3 condition (99.7%). Figure 2: Measurement of guided mode centroid in transverse linear space using a microscope

Measurement of collimated beam centroid in angular space using a Fourier transform optical bench

The same methodology was used to measure the robot’s performance in transverse linear space. The Transverse linear repeatability is covered in norm ISO-9283 section 7.2.2 - Pose repeatability. Using this norm (30 samples, Gaussian fit and repeatability equal to the average plus 3 times standard deviation), a repeatability value of 4.9um is obtained.

• Repeatability – measured error vs. target for over 1000 repositioning trials.
• Angular Error. The red circle represents a radial spread of 50 μRadians (99.7% of data).
• Linear Error. The red circle represents a radial spread of 5.7 μm (99.7% of data).
• Stability – pointing and position variations over 1000 samples for a robot immobilized at fixed coordinates.
• Angular Stability. The red circle represents a radial spread of 15 μRadians (99.7% of data).
• Linear Stability. The red circle represents a radial spread of 0.9 μm (99.7% of data).
• Addressable angular increments – angular position incremented by 1 μRadian, each measured position averaged 400 times to remove stability contributions.
• Error between requested step and measured step is 5uRadians at 3 (99.7% of data).
• Systematic long-base error is also visible.
• Addressable linear step increments – position incremented by 0.2μm, each measured position averaged 400 times to remove stability contributions.
• The error between the requested and measured steps is 0.4 μm at 3 (99.7% of data).
• The Meca500’s impact on optical alignments

Repeatability can be relevant for alignment tasks, such as providing collision-free approaches to the alignment site. However, many optical alignments rely on active feedback, where repeatability doesn’t influence alignment quality. The following analysis focuses on stability and resolution as the main qualifiers of the Meca500’s applicability for micro-optics.

To quantify the robot’s applicability to optical alignment tasks, the impact of linear and angular errors on beam quality and coupling w as calculated for three cases. Case 1 shows direct coupling between identical single-mode waveguides; Case 2, coupling between pre-collimated single-mode waveguides, and Case 3, the collimation of a single-mode waveguide using a perfect lens. For Cases 1 and 2, the coupling loss is calculated using a well-known equation.

Results for Case 1, show that the angular error of the magnitude expected from the Meca500 has almost no impact on loss. Ho wever, the linear error has a significant impact on mode sizes similar to those of a telecom fiber or smaller, limiting the robot's applicability to direct coupling applications.

For Case 2, shown in Figure 10, the linear error has little to no impact on assemblies comprising collimators with a focal length above 0.2 mm. Whereas certain combinations of large focal lengths with small mode sizes can lead to significant losses, typical c-lens assemblies of SMF28 fibers, for example, can be aligned with significant margins of tolerance with the robot.

Meca500 measured performance.

2 Marcuse, D. (1977). Loss analysis of Single-Mode fiber splices. Bell System Technical Journal, 56(5), 703–718.
https://doi.org/10.1002/j.1538-7305.1977.tb00534.x
Linear (μm) Angle μRad (deg)
0.4
0.9
5.7
5 (0.0003)
15 (0.0009)
50 (0.003)
Resolution
(addressable increment)
Stability
Repeatability

For case 3, a simple Zemax model based on a par axial lens was used to quantify the impact of linear errors on beam divergence and pointing. The contribution from angular errors is negligible when considering tilts below 1mRad. From Figure 11, we can see that the impact of linear errors on divergence is below a few percents for focal lengths above 0.5 mm. Above 1mm focal length, pointing error will scale essentially inversely with focal length (1um/1mm = 1mRad and so on).

Simplified typical optical alignment cases.
Case #1: Waveguide Coupling
Case #3: Light Collimation
Case #2: Collimated Light Coupling

Case #1 – Impact of linear and angular errors on coupling loss between single mode waveguides.
Case #2 – Impact of linear and angular error on coupling loss between pre-collimated waveguide assemblies.

Case #1: Impact of a 1um Linear error (Δz & Δr)
Case #2: Fixed Loss
Case #1: Impact of angular error
In summary, for Cases 2 and 3, the Meca500’s performance seems more than adequate for
optical systems with focal lengths above 0.5 mm and mode sizes larger than 5 μm
(approximately the size of a standard mode in SMF28 fiber). However, due to the magnitude
of stability errors on target, its performance is marginal for applications of direct
waveguide-to-waveguide coupling (Case 1).

Impact of linear error on beam divergence and pointing, waveguide collimation.
Table 2:
Summary of the impact of angular and linear errors on three typical alignments cases.
Case 1 - Waveguide to waveguide
Case 2 - Pre-collimated to pre-collimated waveguide
Case 3 - Waveguide collimation
Impact of
Linear error of 1um
Loss < 0.2dB Negligeable Pointing error < 2mRad
Divergence increase < 3%
Negligeable Negligeable
Impact of Angular error of 15uRad
(Mode size > 5um) (Focal > 0.2mm)
Loss < 0.1dB
(for most cases)
(Focal > 0.5mm)

Case #3: Impact of a 1um Radial linear error (Δr) Case #3: Impact of a 1um Gap linear error (Δz) Precise, efficient automated micro-optics assembly. The Meca500 offers a highly effective solution for aligning micro-optics in various applications, as demonstrated in this paper. The robot’s precision and adaptability make it an ideal tool for meeting the demanding requirements of micro-optic assembly. Additionally, the robot provides increased versatility and functionality beyond what typical optical positioners offer, all while maintaining a cost-effective approach through a single actuation platform. By integrating the Meca500 into your next project, you can gain a significant advantage in automating your micro-optic assembly from the outset. Its advanced capabilities enable you to streamline processes and achieve greater efficiency, making it a valuable asset for enhancing the overall performance and productivity of your assembly operations.

About XVP Photonics
XVP Photonics provides a full array of services to help customers develop novel products and manufacturing processes. Our scalable support ranges from on-site expertise interventions to full turnkey solutions, including setting up complete production lines. We offer world-class expertise in photonics, optomechanical design, electronics, firmware and software, system integration, packaging and reliability, assembly processes, and automation. We develop reliable photonic products and the tools, processes, documentation, and training necessary to successfully transfer an efficient and reliable manufacturing process. With a proven track record, XVP Photonics has repeatedly delivered on demanding deployment requirements and helped customers successfully qualify and manuf acture high-reliability products in fields such as telecom, aer ospace, automotive, and medtech. To learn more about XVP Photonics' services, please contact [email protected] or visit https://xvp-photonics.com/

About Mecademic
Mecademic is a true game changer for companies and institutions executing Industry 4.0 strategies. Our industrial robots help our clients automate operations that require high-precision tasks and delicate manipulations, boosting productivity, reducing costs, and improving the quality of their products and operations. Through innovative engineering and architectural simplicity, we offer our customers a plug-and-work system, a tiny footprint, and reduced energy consumption. As a result, our
robots are in over 40 countries, enabling multiple applications of some of the world’s biggest brands, emerging disruptors, and academic institutions. To learn more about Mecademic's industrial robots and microautomation solutions, please contact [email protected] or visit https://mecademic.com/