It’s amazing, but modern spacecraft are equipped with outdated microprocessors that were developed back in the 20th century. In this article we will tell you what is the reason for this state of affairs.
Spaceships are real wonders of technology, equipped with all kinds of electronics. Of course, this includes microprocessors, thanks to which the equipment can perform very complex calculations. However, the chips used in the developments of NASA and other space agencies can often seem laughably outdated.
When we talk about the processor, our desktop computers probably immediately come to mind. Many chips have had a huge impact on the technology industry. Currently, powerful supercomputers with huge computing power have already been developed. It seems logical to use such equipment in such a complex technological field as space research. Landing on the moon or launching and maneuvering a space probe at a distance of millions of kilometers from our planet, of course, potentially require a lot of computing power. It turns out that this is not quite true, and many of you will probably be surprised how little it takes to control, for example, a space station. For example, the newest Perseverance rover, which recently successfully landed on the Red Planet, is based on the RAD750 processor, which is a special version of the PowerPC 750 — the heart of the iMac G3 computers released more than 20 years ago. And the Ingenuity helicopter, which is currently also operating on Mars, is equipped with a Snapdragon 801 processor. These spacecraft, which perform the most complex computing operations, work on very outdated microprocessors. But this state of affairs is unlikely to change even in the future. Let’s find out why scientists from NASA and other space agencies are forced to use such weak SoCs.
Space processors are surprisingly slow
Let’s start with an example that should be well known to everyone. We are talking about an event that took place on July 16, 1969. On this day, as part of the Apollo 11 mission, the SA-506 launch vehicle took the Apollo spacecraft out of the Earth’s atmosphere. And just 4 days later, American astronauts Buzz Aldrin and Neil Armstrong set foot on the surface of the moon for the first time in human history. The mission was successfully carried out using, in particular, the AGC computer (Apollo Guidance Computer), developed back in 1966. The design was quite interesting from the point of view of computer technology, but looking at the technical characteristics of this device, one can only wonder that the mission was a success at all. Just think about it, the chip on board worked with a clock frequency of only 2.048 MHz and had a RAM of only 2048 words. Yes, exactly words. Now it seems just incredible, but at that time it was one of the most modern computers.
It is worth noting that the Apple II home computer, released a few years later, offered similar performance. In other words, at that time the spacecraft had technical equipment that was ahead of its time.
However, this continued until a certain point, and it quickly became clear that a more efficient device does not necessarily have to be the best solution, and sometimes it can be more dangerous. A turning point in the history of space electronics was the determination of the exact values of cosmic radiation and its effects on technology. But how does radiation affect the processor itself?
When the Gemini spacecraft was launched into space, equipped with a simple onboard computer, the technologies used to create it were, from the current point of view, extremely primitive. However, in space, this turned out to be a great advantage. Currently, when creating new processors, more modern technological processes are used, and we can easily buy microscopic processors made using 7 nm lithography. The smaller the chip, the less voltage is needed to turn it on and off. In space, this can cause serious problems. The fact is that when exposed to radiation particles, there is a possibility of unplanned switching of the state in which the transistor will be located. This, in turn, can lead to the fact that the latter will stop working at the most unexpected moment, or calculations performed using such a processor will be inaccurate. And in space, this is unacceptable and can lead to tragic consequences.
An interesting example is, for example, the Intel 386SX processor (a stripped-down version of the Intel 80386), which controlled the so-called glass cabin. It ran at a clock speed of about 20 MHz, which means it could perform tasks at 20,000 cycles per second. Already at the time of its debut in space construction, the chip did not have a particularly high speed, however, more importantly, due to the low clock frequency, the processor was safe.
When exposed to radiation, its particles can damage the data stored in the processor cache. This is possible in a very short window – low timing significantly reduces it, as faster circuits are more susceptible to radiation. In simple words, radiation over time can affect data storage and damage the processor itself. This is unacceptable in the operating conditions of a space station, a launch vehicle or a probe. No one will risk a million-dollar project.
At one time, the effects of radiation were compensated by changes in the production process itself, for example, materials such as gallium arsenide were used. However, each modification was very expensive. In addition, systems for spacecraft are being created in specialized factories in small quantities. Only the introduction of RHBD technology made it possible to use the standard CMOS process in the production of radiation-resistant chips. Methods such as triple redundancy, which allows three identical copies of a single bit to be permanently stored, were also used. When they are needed, the best one is chosen.
The destructive effect of radiation on spacecraft systems at one time led to the failure of the Russian Phobos-Grunt mission. The WS512K32V20G24M chip, designed for military aircraft, was damaged by heavy cosmic ray ions. Excessive current damaged the computer, and it switched to safe mode. Due to communication problems, the restart was impossible, which eventually led to the release of the probe into the atmosphere and its combustion.
Therefore, really durable blocks are used for projects with a long service life. For example, the Hubble telescope was originally equipped with an 8-bit Rockwell Autonetics DF-224 unit with a clock frequency of 1.25 MHz. It soon became clear that this was a bad idea, and NASA had to go through the procedure of replacing the chip with an Intel device. In 1993, the telescope was adapted to support Intel 386, and during the 3A service mission in 1999, a pair of DF-224 and Intel 386 chips were replaced with an Intel 486 chip.
We have already given an example of a space station. It would seem that such a large and complex structure should have a very efficient system on board. However, this is not the case. It is known that on the International Space Station (ISS), the main computer runs on the already mentioned Intel 386 unit. Basically, two sets of three computers are used – one Russian and one American. Let’s also take a look at the much newer New Horizons spacecraft, which flew past Pluto in 2015 and was aimed at the Kuiper Belt. The radiation-resistant Mongoose-V chip with a clock frequency of 15 MHz, capable of performing tasks at a speed of 40,000 cycles per second, was responsible for most of the functions in this device. Its performance is close to that of the processor running the PlayStation console.
When we look even at very modern spacecraft, we see that their designers use solutions that are often several decades old. Not so long ago, the whole world watched the landing of the Curiosity rover on Mars. Few people guessed that inside there is a BAE RAD750 processor with a clock frequency of only 200 MHz, which is an improved version of the IBM PowerPC 750 chip. If you’ve ever owned an Apple computer, you may know this processor from the iMac series. Moreover, it also used the microprocessor from the Nintendo Wii console. Due to the requirements of operation in conditions of increased radiation, its clock frequency has been reduced by more than three times.
We have already mentioned above that the Perseverance rover also runs on a processor that was released more than 20 years ago. In other words, nothing has changed, and spacecraft costing millions of dollars use microprocessors that were released in the last century. As incredible as it may sound, it’s true.
As an example, we decided to understand in more detail what is used as software on the famous Crew Dragon, Falcon and Starlink.
When we hear the name of the Crew Dragon spacecraft, many people remember the three touch screens and the blue graphical control interface that we saw during the broadcasts. There is still a lot of discussion about the feasibility of controlling a spacecraft using touch screens instead of buttons, switches and joysticks. SpaceX chose this option because their goal was to design the ship so that it did not require any control and, at the same time, that the crew always had as much information available as possible. The ship is completely autonomous, and the only thing astronauts have to control is limited by the internal cabin systems, such as, for example, the volume of the audio system. Astronauts should control the flight of the spacecraft and its most important systems only in emergency cases, and SpaceX tried to develop the best graphical interface for these tasks with the help of the astronauts themselves.
However, it should be noted that the key functions of the ship can be controlled using the buttons located under the display. The crew has the opportunity to launch a fire extinguishing system, open parachutes when returning to the atmosphere, interrupt the flight to the ISS, start an emergency descent from orbit, reset on-board computers and other emergency tasks. A lever under the middle display allows astronauts to launch an evacuation system. They also have buttons that trigger and cancel commands entered via displays. Thus, if an astronaut executes a command on the display and it fails, he still has the option to cancel the command by pressing the button under the display. The clarity and controllability of the displays were also tested under vibration conditions, and test teams and astronauts conducted many tests in gloves and in sealed spacesuits.
Probably the most important requirement for the missile and ship control system is, of course, reliability. In the case of SpaceX rockets, this is ensured primarily by reserving systems, i.e. by using several identical components, which then work together and can duplicate and complement each other. In particular, the Falcon 9 has a total of three separate onboard computers. Each of these computers reads data from sensors and rocket systems, makes the necessary calculations, makes decisions about further actions and generates commands to make these decisions. All three computers are interconnected, and the results are compared and analyzed.
Computers are based on dual-core PowerPC processors. Again, both cores perform the same calculations, compare them with each other and check for compliance. Thus, although hardware redundancy is threefold, software and computing redundancy is sixfold. At the same time, it is possible to return the faulty computer to a working state, for example, by rebooting. If the main computer fails, one of the remaining ones takes over.
In case of problems with computers or other systems, the fate of the mission depends on the solution of the Autonomous Flight Safety System (AFSS). This is a completely independent system of on-board computers that runs on a set of several microcontrollers (small computers), receives the same data from sensors, calculation results and commands from on-board computers and controls the safe course of the flight.
To ensure that all computers always have the most reliable data, most sensors are also redundant, as are computers that read this data and then send it to on-board computers. Similarly, computers that control individual subsystems of the rocket (engines, rudders, maneuvering nozzles, etc.) are duplicated by the commands of onboard computers. Thus, the Falcon 9 is controlled by an entire tree consisting of at least 30 computers. At the top of the tree are on-board computers that control the network of subordinate computers. Each of them has its own communication channel with each on-board computer separately.
But as you can see, all onboard computers are based on simple microchips, and not the fancy chips of modern supercomputers.
The future of space chips
Using relatively old processors does not mean that new ones are not being created. It’s just that the process of creating them is very difficult and takes a lot of time. You also need to understand that each structure that will be used in space must meet the requirements of the MIL-STD-883 class. This means passing more than 100 tests developed by the US Department of Defense, including thermal, mechanical, electrical and other chip tests. Most processors that have passed this test are made only from the central part of the silicon wafer. This is due to the fact that it is here that the occurrence of edge defects is least likely.
The list of projects for future spacecraft includes, among other things, HPSC series systems developed by NASA. Processors are expected to be ready at the turn of 2023 and 2024. Their performance should be more than 100 times higher than that of the fastest systems currently used in spacecraft. Americans are focused on developing chips that can help in the conquest of the Moon and Mars. But so far, these are just projects.
The European Space Agency, which has been developing chips based on the SPARK open source architecture for a long time, approaches this somewhat differently. The latest such product is the GR740 model from the LEON4FT family. This quad-core processor with a clock frequency of 250 MHz, equipped with a gigabit network adapter and 2 MB of second-level cache memory, should become a suitable platform for unmanned spacecraft and satellites. According to scientists, the design and characteristics of the processor should guarantee its normal operation even after 1000 years. Scientists guarantee that only after 300 years of operation of the chip, at least one error can occur. This gives confidence in the strength and durability of spacecraft, because a flight to Mars will take about 250-300 days and this is only on a convenient trajectory. Probes sometimes wander for years in outer space.
As a curious fact, it is worth mentioning that in 2017, HPE and NASA launched the first commercial high-performance computer aboard the SpaceX Falcon 9 rocket. A two-socket HPE Apollo 40 server with Intel Broadwell processors and a fast 56 Gbit/s interface arrived at the International Space Station. According to scientists, its performance was only 1 TFLOPS, but for space conditions it was still a lot.
This shows how difficult it is to design chips for outer space, and how hard it is to catch up with, at least, conventional home PC processors.
But scientists are making great efforts to develop the most powerful microchips that will not only support the operation of spacecraft, but will also be reliably protected from cosmic radiation. Perhaps quantum computers will change the situation, but that’s another story.