Homepage Knowledgebase GitHub

ESA incubation status update -- Efficient inertial attitude actuator for small spacecraft

Efficient inertial attitude actuator for small spacecraft

Acronyms and abbreviations

  • ADACS – attitude determination and control system (spacecraft);
  • COTS – commercial off-the-shelf (equipment);
  • FEM – finite element modeling (system or structural analysis);
  • NRE – non-recurring engineering (costs);
  • SRM – switched reluctance machine;
  • PMSM – permanent magnet synchronous machine;
  • PCB – printed circuit board;
  • VSI – voltage source inverter (power electronic circuit);
  • UAVCAN – uncomplicated application-level vehicular communication and networking (standard).


This is a project status update as of December 2019.

Compared to the original outline, the goals of the project have been amended according to the new information and knowledge we gained during the initial technology evaluation and the initial research of the market. As explained in the document ESA incubation mid-term summary (Jul 2019), the current focus is the development of the weight-efficient and cost-efficient integrated flywheel for CubeSats and other compact space vehicles, with the potential for the future extension towards a complete integrated ADACS.

As of Dec 2019, we have invested over €46k into this project; a detailed breakdown is available in the related financial report. Despite the highly detailed records being available, it is hard to unambiguously segregate the costs of this project from our other activities because this work is closely tied to our core competences and therefore it is entangled with many of our other research activities.

Revised problem statement

As indicated in the linked mid-term summary document, the original plan of applying the existing Telega motor control technology for spacecraft attitude control systems has turned out to be non-viable due to the fundamental mismatch of the constraints and design goals of aerial drives it was originally designed for and those of the new application. The crucial differentiator is the typical load profile: the main objective of an aerospace propulsion drive, or any electric propulsion drive in general, is to overcome adverse forces that impede the movement of the vehicle through space, such as drag, friction, or gravity. The scarcity of the energy reserves available on board dictates that the energy conversion should be done in the most efficient way possible. Therefore, the defining trait of an electric propulsion drive is the ability to deliver high power over long periods of time while minimizing the energy losses. Our motor control solution Telega is designed and carefully optimized to meet those objectives well.

Contrary to that, attitude control systems of spacecraft feature an entirely different load profile: an idealized attitude control system requires no energy input unless the orientation of the spacecraft requires adjustment or the attitude of the spacecraft is adversely affected by an external influence, such as tidal forces from nearby celestial bodies or residual atmospheric drag. Considering the typical operating environment and dynamic properties of spacecraft, the great advances of our technology on the side of energy efficiency lose relevance.

Our improved methods of motor control require high-performance computing hardware and very specific analog circuitry. This results in marginally higher costs of their implementation compared to the industry average. The resulting cost difference and the low relevance of the energy saving advantages render our existing solution suboptimal for use in spacecraft attitude control systems.

According to our engagements with relevant businesses at the Innovation Showcase Summit (it was held at Thales Alenia Space last April) and further market research, manufacturers of space vehicles are interested in a complete integrated ADACS solution as opposed to standalone reaction wheel actuators. The long-term goals of the project shall be formulated accordingly.

The article “Starlink is a very big deal” by Casey Handmer provides highly relevant insights into the influence of the ongoing restructuring of the space launch market on the design of new space vehicles. Particularly, it is shown that the ongoing reduction of the cost of launch per kilogram shifts the focus from high-reliability designs with a long design lifespan towards faster and cheaper iterations. Based on that assessment we predict that in the near future the demand for low-cost standard-compliant inertial attitude control solutions for spacecraft will increase considerably.

The ongoing reduction of the cost of hardware, NRE costs, and acceleration of iterations will drive existing and future manufacturers of space vehicles to favor highly modular, interoperable, and standard-compliant equipment over customized or proprietary developments. We see the nascent of this transition already through our other project which is also tightly related to space flight – the UAVCAN protocol. UAVCAN is an onboard data distribution standard that currently dominates the field of unmanned aerial vehicles while finding new uses in space applications and manned (human-carrying) electric aircraft. Following our prediction, at least two major manufacturers of compact satellites – Spire Global and Open Cosmos – are adopting UAVCAN, thus making the first steps towards an ecosystem of easily reusable UAVCAN-enabled COTS equipment for future space vehicles. Based on the empirical observations made on the market of UAV avionics, and assuming that the trend of UAVCAN adoption growth continues, we expect that in the medium term, future manufacturers of space vehicles will increase their reliance on UAVCAN-compatible COTS equipment. Being equipped with this information and having extensive expertise in the field of real-time safety-critical intra-vehicular communications, Zubax can stay ahead of the curve by equipping the prospective integrated ADACS solution with a UAVCAN-compliant interface.

The electric drive design in question shall be built upon our latest know-how that we developed over the course of our extensive research of electric machines. In particular, at the aforementioned summit at Thales Alenia Space it has been confirmed that the improvement of the mechanical vibration profile achievable with our control methods is of high relevance for space applications. This is because vibrations induced by the attitude control system may adversely affect the operation of certain types of satellite payloads, such as narrow-field-of-view cameras. The following chart shows the vibration amplitude of an electric motor when driven by our motor control solution (blue) and an off-the-shelf competitor (orange).

Upon consideration of the above aspects, we have formulated the following updated set of high-level design objectives for the inertial attitude control solution:

  • The drive shall be optimized for continuous operation at high speeds under zero or near-zero load.
  • The unit shall be compatible with the 1U CubeSat form-factor standard.
  • The mass of the unit shall be minimized to meet the design constraints of CubeSat avionics.
  • The complexity and cost of the unit in series production shall be minimized to meet the predicted demands of the future market.
  • The prospective ADACS system shall support the UAVCAN standard.
  • The drive shall be built upon our know-how in energy-efficient, low-vibration, and robust motor control.

Our approach

At the moment, the R&D is focused on the reaction wheel actuator. Design aspects pertaining to the outer ADACS solution are currently outside of the scope, to be addressed in a later stage of the project.

During the initial study it has been identified that the optimal configuration that meets the high-level objectives is an axial flux motor design where the rotor and the flywheel are implemented in the same element, which is manufactured as a monolithic aluminum body of a specific shape, and the stator is implemented as a set of planar coils on a PCB. Conventional designs based on various types of PMSM are considered suboptimal due to the cost and complexity requirements.

The advantage of the monolithic rotor design is that it allows us to drastically cut costs on manufacturing by virtue of not being dependent on magnetic materials and related sophisticated assembly procedures. Simplified machining enables the manufacturer to achieve far greater balancing of the rotor compared to an equivalent solution that contains permanent magnets. Balancing is of extreme importance for a flywheel drive because it is expected to operate at extremely high speeds in the interest of maximizing the angular momentum storage capability.

Initial FEM analysis performed on topologically-equivalent radial flux motor architectures indicates that both SRM and induction motor architectures are feasible. While SRM are simpler to design, their inherent torque ripple makes them unfit for use in the target application due to the vibration profile requirements. Despite that, we chose to implement both types in the first iteration of the project for purposes of initial assessment of the technology.

An early functional prototype of an axial flux induction drive with a planar PCB stator is shown in the following photo (rotor removed).

The following renders show the current candidate for an SRM-type drive design which is yet to be manufactured. Our progress along this particular path has been greatly hindered by the lack of adequate machining capabilities available in our region – the axial flux SRM design requires the rotor of a complex geometry. In the long term, we intend to equip Zubax with a sufficient manufacturing base to remove our dependency on third-party mechanical manufacturing contractors, as they have proven to be exceptionally troublesome for our work. This is unlike our experience with electronics manufacturing contractors, which are generally highly reliable.

The new design requires a different topology of the VSI and a different control logic compared to the original Telega platform. Ideally, in the case of the induction motor design, we should be able to rely on open-loop feedforward control exclusively – this approach is enabled and made feasible by the fact that the parameters of the driven load (the flywheel) are static and known precisely. The model-based feedforward architecture is expected to simplify the mechanics and electronics, thus improving the adherence to the high-level design goals. The fact that the control system will have to be designed from scratch is not a cause for concern since it will be based on the solid technological base and the know-how developed internally at Zubax prior to this project.

The next steps include completion of the design work, assembly of prototypes, and their detailed evaluation in the lab. We expect that the empirical analysis of the first iteration will be completed by February 2020. This is beyond the original deadline; the delay is explained by the changes to the project structure according to the new information we collected in the process.


Our evaluation of the current state of the art and available research indicates that this work is novel and has potential for lasting influence on the design of attitude control systems in small spacecraft.

Currently, there is growing interest in leveraging axial flux motors in various applications, including aeronautical systems, due to their improved efficiency and lower disturbances (1, 2). At the same time, the body of research around axial flux electric motors notably lacks coverage of the axial SRM architecture. Our initial FEM analysis indicates that this approach has non-obvious properties that warrant a deeper investigation. Other types of axial motors, particularly PMSM, are used nowadays in a narrow set of applications, particularly electric aviation (often in a stacked configuration for reasons of scalability) and certain types of appliances where their specific shape can be effectively leveraged (e.g., electric lawnmowers).

Considering the ongoing changes in the industry, partially enabled by the advent of low-cost space launch providers, the relevance of our research will increase in the future. In the beginning of 2020, we are situated in an ideal position to introduce a potentially significant change into the future state of the art.

Final notes

The next steps ahead mostly amount to the formulation of the low-level engineering requirements based on the business requirements we were able to construct so far, followed by the actual engineering work, followed by manufacturing and testing. Assuming that the necessary resources are available, we expect to build a flywheel MVP based on a PMSM integrated axial flux drive by 2021Q1. Following that, the next step would be the integration of a complete ADACS system with the three-axial flywheel assembly. More specifically, the plan would amount to roughly the following steps:

  • Set out the design requirements for the final ADACS-integrated attitude control actuator. This is largely already done due to our earlier engagements with the industry.

  • Design and manufacture the flywheel assembly. This includes machining and assembly of the rotor with permanent magnets and of the PCB-based stator with the control and power electronics. The electronics are going to be based on ordinary automotive-grade COTS hardware, which is a common practice for low-cost LEO space missions. The machining is likely going to be outsourced to Russia due to the lack of alternatives at the moment.

  • Port the Telega motor control solution to the flywheel assembly electronics. As indicated, this is going to be a much-reduced implementation because we intend to operate the drive in the open-loop mode (like the conventional VFD drive) for simplification reasons. We do not expect this approach to cause noticeable degradation of performance because a flywheel is trivial to model accurately (the friction is low and known, the angular momentum is known and constant, no external forces, etc.).

  • Validate the three-axis flywheel assembly on a dynamometric stand. The stand shouldn’t be a problem to assemble using our own manufacturing resources (the mechanics and the electronics are fairly trivial so we should be able to rely on our own 3D printers and existing PCBA contractors). The back-up plan would be to rent the equipment from Sputnix or a similar company.

  • Integrate the ADACS suite. This sub-project will not be explored in detail in this document due to it being out of scope. Our initial assessment suggests that the work can be realistically completed in under 6 man-months.

  • The initial validation will be performed in close collaboration with the aforementioned companies. We do not expect any regulatory overhead whatsoever because this category of space hardware is not subjected to any non-trivial regulation, according to our research.

  • The final stages of developing user documentation, marketing, and other standard productization activities do not warrant any special attention.

As indicated earlier, the release of a bare flywheel unit devoid of a full ADACS does not make business sense according to our analysis of the market. The market, we should note, is hungry. According to our direct exchanges with prospective customers (as described above), our own online research, and consultations with the Sputnix design bureau, the demand for CubeSat space hardware, particularly for the standard 1U attitude actuators, far exceeds the supply. The market state is therefore favorable, we have the required expertise, and we are equipped with sufficient connections with the industry (thanks to ESA, Thales, and Ivo Remmelg in particular) so we are in an ideal position to see the project until successful productization.

We expect that a similar small company like ours involved in a project of this kind could benefit greatly from help with messaging, marketing, and access to a competent manufacturing base.

Manufacturers of vehicles and complex vehicular systems tend to be conservative, preferring established large corporate suppliers (Texas Instruments, Trinamic, Maxon, Siemens, etc.) over startup-scale enterprises. The choice is understandable because design change is hard and the integration of new technology may be extremely costly and time-consuming; by relying on established vendors, the risk of change being necessary is reduced. Considering also the fact that the sales cycle for any mid/large-scale project is at least two years long (that is the minimal time from the initial technology evaluation to mass production), we observe that the environment is extremely hostile to new business and the entry barriers may be insurmountable without external contributions. This problem is often terribly misunderstood by venture capitalists who nowadays tend to be focused on IT-based solutions where the start of a new enterprise is a far less capital-intensive activity and the market is much more accessible (and much more volatile). The state of affairs could be improved by informing the VC public about the specifics of deep-tech industrial startups.

Another commonly occurring problem is the inability of a general-purpose VC to comprehensively evaluate the feasibility and market viability of a technology developed by a deep-tech startup due to the lack of adequate expertise within reach. We perceive that government programs intended to stimulate entrepreneurial activities completely overlook this issue, which is unfortunate because our own experience indicates that the resources poured into the advancement of small scalable businesses could have been applied more sensibly. In particular, it would be of great help if an established institution with a solid scientific authority, such as ESA, could provide a service of independent technology evaluation to bridge the explanatory gap between VCs and startups developing highly complex technologies for the frontiers of modern industry.

A more practical problem we encountered was the lack of easily accessible and sufficiently capable CNC machining facilities in the region. The contract machining services we engaged with were either uninterested in a non-mass-produced design, prohibitively expensive, or simply lacking the necessary means to produce parts of appropriate quality within the required tolerances. The few local firms available in Tallinn were ill-equipped to meet the design requirements, whereas similar services that we were able to locate in Germany quoted inadequate prices. Our short-term solution is to outsource manufacturing to China and Russia, which is terribly inefficient. In the longer term, considering that we routinely require experimental machining, we are planning to equip a private CNC lab – despite the fact that we are not a production company, it seems to be the only viable option. If the ESA offered access to relevant manufacturing services, that might have helped us considerably.

Feel free to quote.

Additional information

Notable projects and companies leveraging our technologies: