Spacecraft Power 101

Introduction

All spacecraft have effectively the same power elements: power generation, energy storage, and power distribution. There are many ways to design each of these systems, but for the most part satellites use solar arrays, batteries, and power conditioning and distribution circuits to achieve this. Both of our deep-space craft, Odin and Vestri, have these core components.

However, Vestri and Odin will be wildly different when it comes to power system design. The differences boil down to build schedule, power requirements, and build experience.

The rushed build schedule of Odin meant that we had to choose components based on what was available to us at the time. This focused the design of the power system to meet the immediate need: make sure that we could integrate the available parts and still meet the launch deadline. We achieved that goal. Odin even powered up and for at least a few hours the power system operated nominally. As you can read in more detail here, Odin did not complete its mission. So when we started the Vestri build, we took a hard look at what went wrong and what we could do to fix it.

The power requirements differ significantly because Vestri is not only a larger spacecraft requiring more power, but it also demands greater reliability. Odin was designed merely for an asteroid fly-by, allowing us to use a relatively low-power chemical propulsion system. Vestri, however, will land on the asteroid, necessitating an electric propulsion system. Therefore, Vestri must generate substantially more power, reduce its minimum power consumption to a much smaller fraction of the solar array capacity, store more energy, and distribute power more efficiently.

We would not understand the following trade space in detail without Odin, which was not only the first deep space craft our team had built, it was the first commercial deep-space craft ever built. We learned a lot during the development and launch, and looking back on it we made some suboptimal decisions in our core design of the Odin power system. We're now improving every aspect of the Vestri power system, leveraging additional development time to implement lessons from Odin and expertise from new team members with deep space craft experience.

This is a detailed summary of how those learnings are being applied to Vestri.

The Power System

The power system is the heart of any spacecraft. It's the most critical system on the spacecraft; if we can't turn the spacecraft on, then we can't do anything else. For the purpose of this class, lets simplify the power system into three major parts: the power generation (solar panels), the power distribution (the routing), and the energy storage (batteries).

Before we jump into the details of the power system, it's important to remember an important difference about AstroForge missions: we go to deep space. When you typically hear the word spacecraft, you're hearing about something orbiting the planet; however, spacecraft traveling to deep space have a completely different set of power problems compared to those that orbit the Earth. Unlike a spacecraft in Low Earth Orbit (LEO), we don't get shaded from the sun, as we don't go behind the planet. This means that the temperature of our spacecraft will be extremely hot on one side, and extremely cold on the other. But we also aren't as structured as a spacecraft in Geosynchronous Orbit (GEO). Due to our thrust vector needing to change, we don't have a dedicated "sun-facing side." We will go into more detail on this when we talk about thermal, but it also plays a major role in how we design the solar array system.

Block Diagrams

Below is a high-level block diagram that represents all the components of the system at a high level. To set the stage for the details below, here are the high-level architectures that were used for both Odin and Vestri. Use them as references as we explain all the critical processes below.

Odin Power Design

Vestri Power Design

Power Budgets

The first step in designing a power system for any spacecraft is a power budget. Power budgets determine the necessary sizing of solar arrays and batteries on spacecraft. These are some of the most detailed and technical documents in spacecraft system design. Variables like temperature, distance from the sun, and lifetime dramatically affect the power budget.

We break the power budget down into what we call modes. These are modes of operations, or states, that the overall spacecraft is in during any phase of the mission. For example, when the thrusters are firing we call this "burn mode" and all calculations are based off the power draw which the thrusters require. The overall mode structure did not change between Odin and Vestri.

The spacecraft has the following 4 primary modes:

• Minimum Power - worst case, minimize power draw to preserve battery and wait for commands from Earth, low duty-cycle beacon

• Safe - Stabilize the spacecraft, communicate with Earth in a predictable pattern

• Burn - When we are using the propulsion system

• Earth Communication - Talking with Earth at a higher data rate, downlink mission data

Minimum Power Mode

Let’s start with the worse case scenario. If things are going very wrong and the battery state of charge continues to drop, even after reverting to Safe Mode, the spacecraft's last line of defense is Minimum Power Mode. In this mode, the spacecraft drops to a minimum power state to extend the remaining battery charge as long as possible. Attitude control ceases, radio transmissions drop to a very low duty cycle, and most avionics systems shut down except for the bare essential systems keeping the satellite alive.

The stark difference between Odin and Vestri's power systems is evident here. Odin had a significantly higher minimum power requirement than Vestri due to its avionics design. Odin featured two power-hungry flight computers that couldn't be turned off, each consuming 35W of power. With these computers and the two receivers operating, Odin's lowest possible power draw was nearly half the capacity of its solar arrays.

For Vestri, we redesigned the architecture to allow the flight computers to turn off and switch to a low-power microcontroller for essential spacecraft maintenance. This design allows us, even with a bigger spacecraft, to consume a fraction of the power above.

Safe Mode

Now let's look at Safe Mode. This is the mode when something goes wrong. Every spacecraft has one.

The behavior of the spacecraft in Safe Mode is simple: point the arrays at the sun and spin about the sun vector while periodically broadcasting telemetry on the transmitters. The receivers stay on throughout, so the spacecraft can receive commands at any time. If the spacecraft successfully finds the sun, it can stay in this mode indefinitely. However, if the spacecraft can't point the arrays at the sun and the battery state of charge continues to drop, then we will revert to Minimum Power mode.

Earth Communication Mode

I will get into this more in the communication piece, but at a high level it's important to understand that the data rate (i.e. the speed at which data is transmitted between a spacecraft and ground stations or other spacecraft, typically measured in bits per second) of the spacecraft is not fixed. Both Odin and Vestri talked in a very low data rate configuration unless they were commanded higher. This is simply the mode where we transmit at a higher data rate.

As you can see we had a massive increase in power on Vestri when in the Earth Communication mode. The main driver here is the solid state amplifier for the radio.

Burn Mode

In Burn Mode, the spacecraft enables the propulsion system and activates the thrusters. This is the mode that gets us out of the Earth system and into deep space. This mode has a huge difference between Odin and Vestri.

There are several things to point out here. First is that, for Odin, the power consumption while firing the thrusters exceeded the power generation of the arrays! This meant that anytime we were thrusting, we had to draw down the batteries. A consequence of this is that we could only thrust for around an hour before we had to stop, recharge the batteries, and start again. This severely limited the burn sequences of the spacecraft.

This was not an oversight. On a chemical propulsion system the firings are short, so being power negative during these bursts are fine. We made sure that if we went below a specific value on the battery we would exit and go into safe mode.

Second, you’ll note that Vestri’s power requirement is significantly higher than Odin’s. The reason for this is simple: on Vestri we use a Hall Effect Thruster, and in the max thrust configuration it alone takes 1240W of power. However, we’ve sized the solar arrays to be able to power the Vestri propulsion system with plenty of margin. This is crucial, since we need to thrust continuously for thousands of hours to rendezvous with the asteroid.

Solar Arrays - How We Generate Power

Both the Odin and Vestri spacecraft use solar panels to generate power. There are a number of ways to generate power in space, such as a Radioisotope Thermoelectric Generator (RTG), but these are cost-prohibitive and subject to burdensome regulation. So, like all commercial spacecraft, we chose to go with solar panels.

Before we talk about the differences in solar arrays between our two spacecraft, let's review the main drivers that affect the power generated by solar panels. The biggest variables are offset angle, distance from the sun, and the temperature of the cells themselves.

Solar Offset Angle: All the power generation figures given so far and discussed below assume the solar panels are directly facing the sun (i.e. 0° offset). As the offset angle to the sun increases, the power generated at the arrays follows a cosine law. The table below gives a simple one-dimensional look at offset angle relative to % power lost.

The above is a really basic chart, but illustrates one important concept. The tolerance at which we need to be pointed at the sun is non-linear — so we don't have to have perfect pointing accuracy to get the vast majority of the solar flux from the sun.

Distance from the Sun: For satellites in Earth orbit, the distance to the sun is a relatively constant value of 1AU (AU = astronomical unit, defined as the average distance from Earth to the sun, approximately 93 million miles or 150 million km). For us, the cannot make this static assumption. As we travel out the asteroid we will get considerably closer and farther to than sun than 1AU. We are designing our spacecraft to operate in a range from 0.9AU to 1.1AU from the sun.

At 1AU, the average solar power density is 1367W/m^2. As this flux is radiating out from the sun, this power density follows an inverse square law. For example, if you double your distance from the sun, the power density drops by a factor of 4. For our mission, bound by 0.9AU to 1.1AU, we will experience power densities from 1725W/m^2 (at 0.9AU) to 1155W/m^2 (at 1.1AU). This is a massive swing, for what is a relatively small difference in distance.

Cell Temperature: The other major factor in solar power generation is the affect temperature has on solar arrays. At higher temperatures solar panels operate less efficiently. The plot below shows the power generation of the Vestri solar cells at 1AU.

We are not in Earth orbit, meaning our arrays are constantly illuminated with the full power of the sun (unless we steer the spacecraft to not point the arrays at the sun). Thus, our arrays will operate towards the hotter end of the curve below. You can see that going from -40°C to 100°C results in a drop from ~0.9W/cell to less than 0.7W/cell, leading to a roughly 20% decrease in power generation due to increased temperature.

To complicate things further, these three factors (offset angle, distance from sun, and temperature) are interrelated and drive further complexity in the determining a power budget. For example, when we are at 0.9AU our solar cells will be hotter so they are less efficient, but there is also higher solar power density so they generate more power. Also, if we orient the solar arrays to have an offset angle, then they will get colder and also experience reduced efficiency.

There are other factors that do affect solar power generation, such as radiation effects like darkening of coverglass and atomic displacement, but we account for them at the mission level, not the spacecraft level. These factors cause us to budget for the cells degrading by 10% over a year-long mission. In reality, and from real-world data, this number is likely closer to 2% per year, or 4% over the entire mission. Yet with something so critical, we want to error on the side of caution. So we design our missions to be under 2 years in total to mitigate the loss of solar cell efficiency.

For the sake of simplicity in the comparisons below, we will assume arrays pointing directly at the sun (0° offset) at 1AU. We will not ignore temperature effects on power generation.

Odin:

The panels for Odin were made by Pumpkin Space Systems and included the standard triple-junction cells on a fiberglass (printed circuit board) substrate. The cells were then covered with glass to protect them against radiation. With this all being done, we expected the panels to degrade at about 10% per year due to radiation effects, meaning for our two-year mission we would lose 20% of the total power.

The solar system was composed of two wings. Each of the wings had three panels. Each one of the panels can create 30 watts of power when facing the sun. In total, we had 90 watts of power per side, and 180 watts of power for the entire spacecraft in the ideal configuration at 1AU.

The panels were constructed as six independent strings, such that if any one string went out we would lose 1/6th of the total power generation.

Vestri:

The solar panels on Vestri haven't changed much in their configuration, but they significantly larger. Vestri is constructed with two wings, and each wing has three panels. However, this time the decision was driven by being able to fit into the fairing on the Falcon 9, not by what was available.

Each one of the panels can generate 341 watts of power at 1 AU. This is done using a semi-rigid cell structure that is on average 25% efficient. For radiation protection, instead of using glass, we are using a polymer spray that will provide the same amount of protection. We expect the degradation on these panels to be around 4% a year, so we will lose 8% of overall power over the lifespan.

We are attaching the semi-rigid cells to a carbon fiber aluminum honeycomb structure coated with Kapton sandwich panel substrate, resulting in an incredibly light design, around 120W/kg including panels, hinges, wiring, and hold down mechanisms. The panels are arranged to have two rows of modules on each panel. Overall we have 12 independent strings across the two wings. The loss of any one panel or string of solar cells will not take down the mission. We will be able to proceed with slightly less power generation capacity.

When designing the solar array system, we wanted the spacecraft to be able to stabilize and communicate with Earth even if the deployment mechanism failed to actuate. If the arrays do not deploy, we have 341W on a single panel, which is enough to operate in Safe Mode and communicate with Earth.

Further, if only a single wing deploys, we have up to 1023W and can still fire up the thruster in a degraded propulsion mode that requires only 450W compared to the 1250W for peak thrust.

This is the single largest lesson from Odin. While the panels themselves were single, or in some cases multi-fault tolerant, the release mechanism was not. For Vestri it was a requirement from the start that we would not lose the mission if the panels did not deploy. While we will be in a severely degraded state — having these large panels will give us months, if not years, to attempt to fix the issue.

Batteries - How We Store Energy

Batteries are a crucial part of any spacecraft. They store surplus energy for use at a later time whenever the power needs of the spacecraft exceed what the arrays are generating in that moment. In Earth orbiting satellites, this is an important part of operating during eclipse - the portion of the orbit where the satellite is in the shadow of the Earth. For these satellites, the batteries keep the system running when the arrays get no power at all.

Our case is a bit different. Under most launch scenarios we experience no eclipse at all, so we don’t need to worry about going behind the shadow of the Earth. Instead, our batteries are there to help the system start up, keep us alive if we lose our solar pointing, and they handle high power surges when the propulsion system starts up.

Batteries are specified in the amount of energy they can store, typically in units of Watt*hours (Wh). A 100 Wh battery, for example, can power a 100W load for 1 hour or a 1W load for 100 hours. It is a useful metric to understand how long your system can operate off of battery power alone.

Odin:

Odin used four 100Wh batteries, two per power string, meaning that each string had access to 200Wh of energy. Odin could not share battery energy from Side A to Side B. A drawback of the off-the-shelf power system used in Odin was that the power system could only charge the batteries up to a maximum of 90% their full capacity, this means that each battery had only 90Wh of usable energy in it. On the other end of the battery charge spectrum, under high loads the battery would ‘brown out’ at battery state of charge less than 25%. This further limited our usable energy down to 65Wh per battery - a significant drawback of our system. With all 4 batteries fully charged, Odin had 260Wh available to use.

For a spacecraft with minimum power mode of ~75W (Odin), this amount of energy doesn’t leave much time for troubleshooting issues while running off of batteries alone. In the event that the spacecraft was running in Minimum Power mode and only using battery power, we had about 3.5 hours before we depleted the batteries.

Vestri:

On Vestri, we are significantly increasing the size of the batteries relative to the minimum power draw of the vehicle. We now have two 500Wh batteries, amounting to over 1000Wh for the entire system. Since we are designing the power management system in house, we are not limited by the performance of off-the-shelf systems. This means that we can share the battery energy from Side A with that of Side B, giving each side full access to 1000Wh of energy. Also, our battery charge circuit can charge the batteries up to their full 100% state of charge, giving us full access to the energy storage.

We’ve increased the resilience of Vestri by also reducing the minimum power mode of the spacecraft. Vestri can reduce it’s power draw to 36W, meaning that if we are operating off of batteries alone in this mode, we can last for over 27 hours on a single charge. This gives us a full day to root cause any issues and identify a fix in the event we are operating on batteries only (an unlikely scenario to begin with).

Power Board - How We Distribute the Power

Generating power with solar arrays and storing it in batteries doesn't do you any good if you can't distribute the power out to the systems that need it, like your computers, sensors, radios, and actuators. That is where the power distribution comes into play.

A regulated power system will keep the bus voltage constant, in most cases 28V. It gives you the advantage of having a constant voltage for all downstream loads, and not having to protect every component for overvoltage damage. But it does lower the overall efficiency, and adds some mass to the system as the power regulators need to be larger.

An unregulated system is where the bus is not held at any special voltage and is instead dictated on the high end by the solar input, and on the low end by the battery. These systems are higher efficiency and simpler, but require that all your components can tolerate variable loads and have built-in overvoltage protection. An unregulated system also does not operate the solar panels at their peak efficiency.

Both Odin and Vestri used a regulated power system. We purchase components from other vendors, such as our reaction wheels, star trackers, and radios. Making sure the supplier was compliant with an unregulated system would require a deep dive into their systems, and getting it wrong could mean a simple loss of a mission-critical system. For the sake of reliability, regulating the power was the safest choice.

Distribution

The power distribution system is also responsible for conditioning the power from solar arrays into a usable format for the vehicle. There are several ways to do this, but so far all of AstroForge spacecraft have used an algorithm called Maximum Power Point Tracking (MPPT). In MPPT based power systems, the power conditioning circuit tries to find the maximum power point of the solar cells to get the most efficiency out of them. This lets us tune the amount of power coming off of the solar arrays to meet exactly our need.

A final aspect of the power distribution system is to answer the question: what do we do with the excess power? If your solar arrays are generating 200W and you’re only using 100W, you have a 100W surplus. That power needs to go somewhere, and in most cases it is heat. Different satellites deal with this in different ways, but an advantage of MPPT based systems is that they can control the power coming off the solar arrays such that the excess power (heat) stays on the arrays. This is done by actually operating away from the maximum power point of the solar arrays.

Odin:

On Odin, we were using commercial-off-the-shelf systems for power conditioning and built our own power distribution circuit. We used two Power Conditioning Units, each running an independent MPPT algorithm, to distribute power to the A Side and B Side devices. However, we wanted both sides (A and B) to have access to 100% of the solar power. To achieve this, we put in place a power sharing diode between the solar arrays and MPPT circuits. In hindsight, this was not a good idea. The MPPT circuit was not designed to operate in parallel with a second circuit, and effectively had no knowledge of the parallel circuit. This meant that each side was trying to pull the Maximum Power Point off of the same solar array, resulting in the voltage of the solar array collapsing.

We caught this issue during testing and formed a workaround by artificially limiting the power going into one side for a given solar string. For example, MPPT B Solar Input #1 was limited to a very low power (100mW) while MPPT A Solar Input #1 (tied to the same solar string) was not limited at all. This enabled the two MPPT circuits to play nicely together, but resulted in around a 5% degradation in total power available to the system. Further, the power sharing diode also incurred some power losses, limiting the peak power generation capability.

Another issue with the off-the-shelf MPPT was the longer switching timing. This ment that during large power transient, such as when the propulsion system shuts down suddenly, there is a surge of energy in the system that has to go somewhere. “Somewhere” in this system meant a shunt, which we called the “overvoltage” shunt. This is basically a large piece of metal bolted to the spacecraft. This effectively wasn’t a issue in the design, but it did cause some issues during testing and integration.

During one assembly of the vehicle, the shunt circuit was damaged during handling and we didn’t notice until we turned the system on. During testing, we regularly subject the power system to large transients that necessitate energy going into the shunt, however this time the shunt circuit was open (meaning power couldn’t flow into the shunt). When this happened, the power had nowhere to go and it blew up the MPPT circuit. Smoke billowed out of the vehicle and we were forced to do an emergency replacement two weeks before shipping to the launch site. On Vestri, we’ve removed the need for this shunt altogether.

Vestri:

The biggest advantage we have going from Odin to Vestri is that now we control 100% of the power conditioning and distribution system design since we are building it in house. We are still going with MPPT, but we control the design. This allows us to make several high level decisions that will vastly improve our robustness.

First, we are no longer sharing solar arrays between the two sides of the power system. In Vestri, we have dedicated strings for MPPT Circuit A and MPPT Circuit B. Out of the gate, we avoid the diode losses and efficiency losses required to enable two independent MPPT circuits operate in parallel. However, we still want full access to the power from Side B in Side A and vice versa. To achieve this we are adding a power sharing circuit that lets us tie the two power buses together, creating a power pool that gives both sides access to 100% of the power and energy available to the spacecraft.

Since we control the design, we are also getting rid of the need of an overvoltage Shunt. Vestri will still be subjected to the large power transients that necessitate a shunt, however we will deal with it in a more clever way. The duration of the power surge is quite short, less than 50mse, so we will be putting that power back into the battery until the power from the solar arrays can be reduced by the MPPT circuit. Effectively the battery will act as our overvoltage Shunt, meaning we don’t need additional circuit or external devices to deal with it.

We will continue to push the boundaries of speed in deep space exploration, with the time between Odin and Vestri launches scheduled for less than 12 months. Through the design, build, launch, and operations of Vestri, we will inevitably learn more lessons and continue to improve our power systems for future mining vehicles. Vestri already represents a step change in capability compared to Odin's design. Everyone is focused on getting every detail of Vestri correct. We will be successful on our next mission.