Victor Soh Cheng Ying

Year 4 Mechanical Engineering

A0277970M

2. Mechanical Design, Analysis, and Manufacturing of CubeSat Frame Structure

2.1 Objective

Given that the camera is one of our mission’s primary payloads (where the Simera MonoScape100 is the only viable option for our needs), proper integration is critical to avoid vibrations or structural shifts that can distort imagery.

2.2 Requirements

To meet the mission objectives and integration standards, the G5 structure must adhere to the following engineering requirements:

Table 2.2.1 Table of requirements

2.3 Limitations of Existing Solutions

Standard COTS frames are found to not be able to accommodate the imager without major modification, risking vibration-induced misalignment and inadequate structural support. This is due to the size of the imager at 98 mm at its widest.

Figure 2.3.1 Engineering drawing of Simera MonoScape100 camera showing the widest dimension of 98 mm

Figure 2.3.2 Engineering drawing of GomSpace Nanostructure 6U (GomSpace A/S, April 2018)

Figure 2.3.3 Image of previous design to accommodate for Simera MonoScape100 imager

Aside from COTS frame modifications, other existing solutions include purchasing a larger frame which is 12U (the next size up from 6U which does manage to fit the imager). This offers more space and fits the camera without structural changes. However, there are not enough additional subsystems to justify the 12U space requirement for the G5 mission scope. Launch costs also increase drastically with size. A 12U launch can cost up to 2 times more than a 6U one (approximately an additional SGD 500k to SGD 1 million).

Figure 2.3.4 Image of 12U COTS CubeSat frame (Satsearch, 2025)

2.4 Proposed Solution

A fully custom-designed 6U space grade CubeSat structure, manufactured in-house at the NUS Central Workshop is the proposed solution to the issues prefaced above. This enables the following:

Direct payload integration
Structural integrity
Modularity and ease of assembly
Shorter lead time
Lower cost

Figure 2.4.1 Image of new proposed design for 6U CubeSat structure frame

2.4.1 Housing of Components

A prior design (based off a standard COTS frame) was modified to house all components with great modularity for all components. Despite that, the frame had to have a cut in it to accommodate the desired earth observation imager (i.e., Simera MonoScape100).

Figure 2.4.1.1 Highlight of prior design with modified COTS frame with payload

Figure 2.4.1.2 Highlight of prior design with modified COTS frame without payload

Upon the redesign, the issue with the previous one has now been fixed successfully in the current iteration by providing sufficient space between the imager’s widest part and space in between the frame.

Figure 2.4.1.3 Image of redesigned 6U CubeSat structure frame showing sufficient space to accommodate for the imager without compromising on structural integrity

2.4.2 Compliance of Hard Requirements from ExoLaunch GmbH

Since the CubeSat is to be integrated into the deployer from ExoLaunch’s ExoPod Nova, the requirements stated by them are mandatory to be fulfilled. The following are the hard requirements set by our intended CubeSat deployer (with reference to the ExoPod Nova deployer in particular):

Rail-to-rail dimension tolerance to be ± 0.1 mm for the CubeSat rail width (X) and height (Y) with the CubeSat rail length (Z) tolerance at ± 0.5 mm (6U rail dimensions are generally the same as stated in the 6U section from the CubeSat Design Specification Rev 14.1 except for Z);
Surface roughness of Ra 1.6 or lower (lower Ra value is smoother) after annodising;
All rail surfaces must be Type III hard annodised;
Any holes or edges on the CubeSat rail must be adequately chamfered.

Figure 2.4.2.1 Table of maximum CubeSat dimensions from ExoLaunch

The subsequent figures show the critical dimensions of the redesigned 6U CubeSat frame which are within the stated maximum dimensions from ExoLaunch.

Figure 2.4.2.2 Screenshot of CubeSat rail height (Y) from redesigned 6U CubeSat structure frame

Figure 2.4.2.3 Screenshow of CubeSat rail width (X) from redesigned 6U CubeSat structure frame

Figure 2.4.2.4 Screenshot of CubeSat rail length (Z) from redesigned 6U CubeSat structure frame

Not only are the dimensions of the current redesign in line with the maximum CubeSat dimensions, but also the overall mass and COG is well within the stated recommended limits (as seen below) at approximately 9.52 kg and (XYZ : -5.81 mm, 1.54 mm, -16.21 mm) respectively.

Figure 2.4.2.5 Table of maximum recommended distance of the COG from the geometrical center (ExoLaunch, 2024)

Figure 2.4.2.6 Screenshot overlay of maximum acceptable mass of any given 6U CubeSat and the mass of the redesigned 6U CubeSat together with its COG
For the surface finish requirement of Ra 1.6, NUS Central Workshop has adequate capability to achieve the required surface roughness. The main structure frame parts that need to be machined fit within the limits of the machine table and range of motion of their machines.

Figure 2.4.2.7 Screenshot of overall dimensions of X plate

Figure 2.4.2.7 Screenshot of overall dimensions of Y plate

Figure 2.4.2.8 Screenshot of NUS Central Workshop CNC machining capabilities (NUS, 2025)

A local company will provide hard anodizing, with a thickness of 30-60 microns. Half of this penetrates the surface, so a maximum of 60 microns must be accounted for in post-process dimensions. These values require verification using a profilometer. We recommend creating test pieces for anodizing to validate the coating thickness and determine the necessary machining offsets before final production.

Figure 2.4.2.9 Local company for Type III hard anodising (ATC, 2025)

Figure 2.4.2.10 Surface Roughness Guide for CNC Machining (Geomiq, 2025)

2.5 Analysis of CubeSat Design

The frame was validated through FEA analysis in Ansys to confirm its structural soundness under launch conditions. A full-scale aluminum mock-up was then CNC machined to verify assembly viability, ensuring the 95mm camera and kill switch could be properly installed with correct clearances and alignment before final manufacturing.

2.5.1 Simulation and Results

The completed frame assembly, comprising all brackets and structural rings, was subjected to static structural, modal, and random vibration simulations. Point masses were positioned at each stack’s center of mass and connected to their corresponding mounting points using RBE2 (Rigid Body Element) remote points, which guarantee full load transfer experienced by the point mass to the mounting interfaces, in order to accurately depict the internal mass distribution of the satellite’s components.

Figure 2.5.1.1 Point masses connected to mounting points using RBE2 (black lines)

Load conditions were applied according to ExoLaunch’s qualification requirements, as illustrated in the figure with a maximum load of 2183.5 N force on each of the two rail faces. The load conditions for random vibration can be seen in Appendix A.

Figure 2.5.1.2 Deployer loads in datasheet and in ANSYS environment

The assembly was constrained using fixed supports at the top two rail contact faces and at the front and rear of all 4 rail ends, replicating the clamping mechanism of the ExoPort deployer. This setup reflects the real launch configuration, where the CubeSat remains fully restrained and stationary throughout ascent. The results are summarized in the table below.

2.6 Killswitch Mechanism Design

Launch vehicle safety protocols require that all CubeSats remain electrically inert until deployment. To satisfy this, a new mechanical kill-switch mechanism was designed and fabricated.

The previous design, shown in blue, positioned the kill-switch flush against the deployer rail. However, after discussions with stakeholders and our NuSpace, this configuration was deemed mechanically risky and had a high likelihood of the switch being jammed, bent, or locked during integration into the deployer slot, potentially resulting in switch failure affecting mission success.

Figure 2.6.1 Comparison between the earlier flushed-switch configuration (top) and the improved plunger-based designs (bottom)
The new updated mechanism employs a compact spring & plunger-based assembly consisting of two threaded halves joined by a hex-coupling interface, enabling simple installation using a standard hex coupler tool even in tight clearances.

Figure 2.6.2 Plunger assembly with annotations

This kill-switch uses a spring-loaded plunger in a stepped bore. When compressed in the deployer, it holds a microswitch “off,” isolating power. Upon deployment, the spring extends, closing the switch to activate systems. A redundant switch on the opposite rail ensures reliability. The design is a simple, robust solution compliant with SpaceX and ExoLaunch requirements.

Figure 2.6.3 Working prototype of new killswitch design
Upon finishing the working prototype of the killswitch mechanism design previously, the minimum distance of the plunger to the killswitch hinge lever and the stroke length of the plunger is analysed in CAD. This is to ensure that the selected distances are exactly where the killswitch needs to be for proper engagement of the killswitch by the plunger. It is also vital to have a minimum clearance between the plunger and the killswitch hinge lever to avoid any unnecessary killswitch engagement in its deployed state.

The clearance distance on The distances and stroke length can be observed from Figure 2.6.4 and Figure 2.6.5 respectively. For a better illustration of the plunger movement, refer to Appendix B Figure B1-2 and Figure B1-3.

Figure 2.6.4 Minimum distance of plunger to the killswitch hinge lever

Figure 2.6.5 Stroke length of the killswitch plunger

Referring to GomSpace’s Nanostruc 6U CAD for its killswitch plunger hole diameter and depth, it is observed that their killswitch plunger ratio is 2.48 (2 d.p.). Thus, the hole diameter and depth of the hole (whereby the plunger will have a fine roll fit, H7/g6) will respect a similar aspect ratio as the one decided by GomSpace. An aspect ratio exceeding 1 will make sure that there is minimal deflection in the plunger when engaging the killswitch (which is sprung).

Figure 2.6.6 GomSpace Nanostruc 6U CAD showing its killswitch plunger hole diameter

Figure 2.6.7 GomSpace Nanostruc 6U CAD showing its killswitch plunger hole depth

Figure 2.6.8 Chosen hole diameter for the killswitch plunger design

Figure 2.6.9 Chosen hole depth for the killswitch plunger design

An undercut is decided to be included on the killswitch plunger to ensure that it will seat flat on the face of the structure frame (as shown in Figure 2.6.12).

Figure 2.6.10 Undercut width chosen for the killswitch plunger design

Figure 2.6.11 Killswitch mechanism section of the satellite with the plunger and spring

Figure 2.6.12 Seating face for the back of the killswitch plunger

Figure 2.6.13 End view of the X plate showing the killswitch plunger hole

As seen in Figure 2.6.14, the plunger engages the killswitch as intended with no issues of the plunger binding up within the hole.

Figure 2.6.14 Installed view of the killswitch mechanism in its deployed and stowed state

2.7 Design for Manufacturing and Assembly (DFMA) Considerations

2.7.1 Manufacturing Considerations

One of the main considerations done for the full frame structure redesign is the minimum fillet radius chosen together with the fillet placement depending on the part orientation relative to the endmill orientation. Fillets are only applied if the endmill orientation is perpendicular to the cut surface. Furthermore, all chosen fillet radii (for internal corners specifically) also considered the endmill size necessary to mill it.

Another vital consideration done was the minimal wall thicknesses of 1 mm is observed for all features and machined components to ensure that the machinability of the thin features are still acceptable.

Figure 2.7.1.1 Close-up look on the killswitch mechanism location

Figure 2.7.1.2 Close-up look on fillets only applied when the endmill orientation is perpendicular to the cut surface

Figure 2.7.1.3 Smallest internal fillet corner radius chosen for the redesigned 6U CubeSat structure frame

2.7.2 Assembly Considerations

One of the assembly considerations done was to place all of the main socket head cap screws (as seen in Figure below) such that they can be easily accessed from the outside. This would allow the use of any torque wrench without any issues or restrictions.

Figure 2.7.2.1 Screenshot of highlighted bolt holes that allow for better accessibility on the redesigned 6U CubeSat structure frame

Figure 2.7.2.2 Showcase of torque wrench being used on the redesigned 6U CubeSat structure frame

Thread engagement of at least 2.5D (approximately 2.83D, as per what Figure below) was chosen since the hole is tapped into aluminium (soft compared to steel). This is to ensure that the stress from the bolt is distributed more uniformly to avoid the fasteners from stripping out the tapped threads in the frame structure.

Figure 2.7.2.3 Close-up on the thread conditions of the main bolt locations of the redesigned 6U structure frame
Last but not least, the tolerance stack-up between the X plate, Y plate, and the structure ring is analysed. Even at the largest size of the structure rings (111.7 mm), there will be a minimum clearance of 0.05 mm between the Y plates and both sides of the structure rings. However, taking the maximum dimension in between the Y plates (112.45 mm) and the smallest structure ring dimension (111.6 mm), there is a maximum deviation of 0.85 mm which may cause issues such as stretching the ring when bolting from both sides of the Y plates.

Figure 2.7.2.4 Screenshot of the X plate, Y plate, and structure ring with annotations

Figure 2.7.2.5 Fit between the frame structure and the structure rings before bolting in from the other side of the ring

Figure 2.7.2.6 View of the interface between the frame structure and structure rings where they are bolted together

Figure 2.7.2.7 Assembled frame structure with all of the structure rings to see if there are any interferences between the two

Referring from Figure 2.7.2.5 to Figure 2.7.2.7, it is confirmed that the chosen tolerances for both the frame and structure ring (as highlighted in Figure 2.7.2.4) are effective at ensuring no issues during assembly while also not having a very large gap between the parts.

2.8 Vibration test plan

Vibration testing is an essential procedure for confirming that hardware can withstand the dynamic pressures of launch and operation, especially in the heavy equipment, automotive, and aerospace industries. In order to find latent faults without causing over-testing damage, a structured test plan makes sure that components are subjected to realistic stresses.

Figure 2.8.1 Planned vibration test plan flowchart

In order to replicate real-world vibration conditions, a typical vibration test follows a methodical sequence across three orthogonal axes (X, Y, and Z). The procedure starts with putting the test article on a shaker for the Z-axis of the CubeSat, as shown in Figure 2.8.1. To find the baseline resonance profile and make sure the system works, a low-level sine sweep is first carried out. The test-level random vibration is then applied. After that, same process is repeated for the Y-axis and, lastly, the X-axis. A final low-level sine sweep is performed at the end of the test for each axes to look for any alterations or damage caused by the random vibration.

Figure 2.8.2 Typical configurations for vibration tests (Instar, 2025)

Specialised high-force systems are needed for large or heavy test objects. The 350 kN vibration table from TÜV SÜD Singapore, which can support payloads up to 3,000 kg on a 96-inch square table, is shown in Figure 2.8.3. It has a sine force rating of 350 kN, a random force rating of 315 kN, and a frequency range of 0 to 2000 Hz. For sectors including aerospace, automotive, and off-highway vehicles, these systems provide thorough testing, enabling engineers to customise profiles to meet particular customer requirements.

Figure 2.8.3 Description of the 350 kN vibration table available at TÜV SÜD Singapore (TÜV SÜD, 2026)

Designing efficient tests requires an understanding of what goes wrong during random vibration. Unseating electrical connectors, flipping mechanical relays, loosening threaded fasteners, and material rupture are common failures, as shown in Figure 2.8.5, especially for compact, high-frequency goods (over 100 Hz) like PC boards and avionics. It’s interesting to note that huge primary structures for single flights can fail during a test if they are not well protected, yet they rarely fail in flight due to random vibration.

The fact that peak acceleration during random vibration is usually substantially beyond the 3σ (three times RMS) level (typically, 4 to 5 times RMS) is a crucial design and analytical point from Figure 2.8.4. To avoid failure, further conservatism in the analysis is required if a design is based just on 3σ. Additionally, since fatigue failures account for the majority of random vibration failures, test exposure times should be limited as short as possible to prevent overstressing the hardware.
In conclusion, confirming hardware dependability requires a carefully thought-out vibration test that makes use of appropriate flow, appropriate fixtures, high-capacity equipment, and knowledge of peak acceleration and fatigue processes.

Figure 2.8.4 Highlighted section on typical peak accelerations in terms of multiples of the random vibration test RMS value (Instar, 2025)

Figure 2.8.5 Highlighted section on common failures expected when undergoing random vibration tests (Instar, 2025)

Figure 2.8.6 Highlighted sections on the types of accelerometers commonly used and guidelines on their placements (Instar, 2025)

2.9 Future Work

For the future work of this section, the following are the future work planned for the mechanical design, analysis, and manufacturing of the CubeSat frame structure:

Further refine the random vibration test plan together with a vibration test setup checklist and work closely with the test jig design before finalising it
Proceed with the physical testing to qualify the frame structure for flight
Determine whether the whole frame structure including the structure rings need to be anodised or just the X plates