Can you really use Arduino for a small spacecraft?

Question

A group of Russian geeks intends to shoot a small vehicle to the moon, which is supposed to photograph the places of the Apollo landings. This will be the ultimate proof that people actually visited the moon.

In one of the articles about the topic I've read that you could use a low-cost onboard computer, even an Arduino, provided that it is protected from radiation.

If the purpose of the spacecraft is to

1. separate itself from a group of satellites (several small vehicles will be launched with one rocket),
2. navigate to and fly over a particular place,
3. shoot a couple of photos and
4. transmit them to Earth,

can you really use such a low-end device as Arduino as the onboard computer? If not, why not?

Update 1 (06.02.2016 12:34 MSK): I asked the project lead on how they intend to prove that the images are real. You can find his answer here. Short version:

1. The images will have a much higher quality than those generated by previous analogs. This means that the new vehicle will "see" things (such as details of the Moon surface), which previous didn't. Therefore, the images will contain additional information, which is not known yet (like a small crater, which was unrecognizable on earlier photographs).
2. Radio amateurs will be able to detect (but not to decode) the signal from the vehicle.

Answer 1

Can you have a spacecraft based on an Arduino? Sure you can! ArduSat was two kickstarter funded cubesats that were eventually launched from the International Space Station in November 2013. When you think about it, an Arduino easily outperforms for instance the over forty year old Apollo Guidance Computer

All of your requirements should be doable, if it is simply a flyby mission.

By the way, images of the Apollo landing sites are already taken, for instance by the Clementine probe.

Answer 2

You can, but it will suffer from a number of problems. These problems can probably be overcome with a short term mission. Problems include:

1. Radiation- Degrading the long term effect of the electronics.
2. Single event upsets- This is probably the biggest danger, a high energy cosmic ray strike could cause a bit flip, potentially changing the code running in a mission critical way.
3. Temperature- If the temperature isn't carefully managed, could see some significant damage.
4. Vacuum- Probably not much of an issue, but could cause out-gassing which might have a long term effect.
5. Vibration- Parts could be vibrated off during the launch.

These probably can mostly be overcome by adding in extra protection to the spacecraft. But there isn't anything in particular that would prevent an Arduino from being used to control a spacecraft, especially a very short term mission. It has been demonstrated in LEO, but the radiation effects will be more severe for a mission to the Moon. Bottom line, it could be done, but I wouldn't recommend it.

Answer 3

I concur with Hohmannfan's response. This answer addresses the wider issue of computers in satellites.

Who needs a computer? I don't think there is anything about the mission that you have described in the question that actually requires any "digital computer" at all. It might seem as if image handling and navigation are highly demanding in computing terms, but that is largely because we are accustomed to the idea of a world enabled by high-level software.

I think it is a good starting point in terms of systems engineering education to actually step through the processes involved and ask oneself "what is the most basic implementation possible?", particularly in terms of "what decisions absolutely must be taken on board, rather than by ground command?". Designers of missions in the 60's through 80's often came down on the side of "no computer needed". Its only since the weight, performance and cost of such things has come down that we take it for granted.

Digital All of the logical decisions that need to be made on a satellite could in principal be made by discrete logic gates. Its an engineering judgement as to when the "digital finite state machine" so created has become so complicated that it would be better replaced by a CPU/address bus/data bus architecture.

As an aside, its not obvious that there is any clear dividing line between discrete electronics and a "computer" in the modern sense. This article regarding Pioneer 10 hints at the in-between possibilities.

Much of the computation for the mission was performed on Earth and transmitted to the probe, where it was able to retain in memory up to five commands of the 222 possible entries by ground controllers. The spacecraft included two command decoders and a command distribution unit, a very limited form of processor, to direct operations on the spacecraft.

Analogue Furthermore, decisions relating to progressive quantities - sensor outputs and control loops are in the first instance analysed in control engineering terms. How they are implemented is again a design choice and the old world was full of analogue computer elements.

Environmental compatibility There is the launch environment and then natural radiation once in orbit. The latter includes ESD damage (see here) as well as radiation dose and displacement damage. One interesting anecdote is that the progressively higher performance computers using physically smaller gates and switching times are more vulnerable to these effects than their older cousins. When it comes to ESD, shielding doesn't help when there are peripheral bits of the circuit on the spacecraft skin. The designers need to take it carefully.

By the way, thank you @uhoh for the link, that was interesting.

Answer 4

It should be possible. ESA has tested the rad-hardness of some cousin processors to that used in the Arduino and they turned out fairly well, at least for a relatively short mission. Some current stuff is actually using ancient 8051-architecture chips.

There would be enough processing power to do navigation, maybe even enough to stream out recorded fake pictures.

Edit: A powerpoint summarizing the ESA findings can be found here.

Heavy Ion result for ATMega128

The ATMega128-AU16 has an acceptable behavior for a ISS LEO environment.

SEL once in 481 years

SEU once in 690 years

The AT90CAN128-AU16 is not acceptable for space application.

The chip used in many Arduinos is the ATMega328, which is a cousin of the ATMega128.

Answer 5

Others have covered the hardware difficulties, but I would like to mention the software difficulties. Getting enough margin (CPU and memory) is difficult on arduino class processors. Spacecraft I have worked have required anything between 50% and 90% margin that means that you only get to use between 50% and 10% of the processor. The margin is for things like schedulability and memory scrubbing. The other factor is it is just harder (and therefore more expensive) to write reliable software with less resources. There is a fine line between too simple (can't do all the checking you really should do) and too complex (impossible to test) and the arduino probably favours more of the former.

Answer 6

You don't need that much processing power or RAM to explore space. The Arduino is far more powerful than the processors which have been used on spacecraft historically. The basic Arduino is 16 MHz and 256 KB of flash memory. You can add RAM or flash in significant amounts. The computers used in the Apollo space program come nowhere near this.

Galileo's processors are 8 MHz, while Spirit and Opportunity are 20 MHz processors. The code on these probes was highly optimized and specially developed for the purpose at hand, making very efficient use of the computing resources available.

It's entirely do-able.

Answer 7

From a computational point of view, it is feasible to pack all the algorithms in there. However, there are issues with radiation shielding as you point, but integrated circuits are also affected by temperature, so proper thermal insulation must be considered too, as these are not definitely graded to operate in "harsh" environments.

Answer 8

I think this project is extremely difficult. To get detailed high quality pictures of Apollo landing sites, you need a very low orbit around the moon. The distance between the spacecraft and the landing site has to be small enough for a detailed picture. But those very low moon orbits are not stable due to mascons of the moon. To avoid a crash, the small spacecraft would need a sophisticated propulsion system and a lot of fuel for orbit control. The navigation must be extremely precise to get an orbit over a landing site. There is no moon GPS available for navigation. Finding the landing sites requires complex and very fast image processing to dectect the remains of a landing within a series of pictures. I doubt that an arduino has enough processing power to do this image processing in real time.

Answer 9

My take: You could use an Arduino board, but you would have to remanufacture it to be non-RoHS compliant, i.e. using solder made of lead-tin alloy, using conformal coatings and potting it. As it stands, space electronics have an exemption from RoHS compliance anyway.

The Arduino boards you buy off the shelf are RoHS compliant and use lead-free solders, whose only main structural metal is tin, alloyed with small amount of silver. One problem with that in a space environment is the high susceptibility to growth of tin whiskers from the solder joints, which can cause short circuits by bridging to adjacent tracks. Tin whiskers have been asked about as a question here before. Lead content slows down the growth of the whiskers and inhibits the tin pest.

Unfortunately, the growth and causes of these whiskers is not well understood, and the mission duration isn't a factor - they can occur at any time.

So your COTS Arduino could go on the fritz at any time.

Answer 10

This is something I've been looking into as well. Here's my idea:

Use ATMEGA328 (or ATMEGA16A if durable enough) with triple redundancy, using only thru-hole components on protoboard. As mentioned elsewhere here, use leaded solder. The voting circuit for each digital output can be built with 4 logic gates. An error counter and watchdog circuit to reset a misbehaving CPU can be built from discrete logic chips as well. 4000 series are what I'm looking at at the moment. Add optoisolators where they make sense. Copper tape on the top and bottom of every chip might help a bit.

Conformal coat the whole board when done. Add thin steel RF shielding over the most sensitive components. Use potting epoxy under the shielding. There is some new research (sorry I don't have a link) that suggests rust might help to deflect certain unwanted particles, so if this inner shielding is rusty, that might help. The shielding should be connected to common ground with some sort of protection against sudden spikes.

Put the whole thing in a milled/cast aluminium project box with walls at least ⅛" thick. Connections to the outside should use GX aviation connectors. Keep the steel and aluminium from touching by using copper or brass. Exactly how depends on how you made the hole and how much space is around it. Fill the entire box with potting material. Sand down the mating surfaces of the box and lid for good contact, screw it down, and seal the seam with conductive tape.

It might be a bit heavier than you would like, but all of these methods together give it a fighting chance of surviving.

Edit:

• If something is really mission-critical, use mechanical relays instead of MOSFETs.
• Avoid electrolytic capacitors like the plague. Also avoid polyester and nylon film capacitors. Monolithic ceramic capacitors are 'good enough' but generally try to use as few capacitors as possible.