24
$\begingroup$

Recently, there has been growing interest in using Raspberry Pis in CubeSats. NASA is even in on the trend. But, considering that outer space is pretty harsh on electronics (i.e. radiation and cosmic rays, wild temperature swings, space-dust and micro-meteors), is the Pi really viable as a space satellite?

Based on what I remember, space-worthy electronics usually feature a way to recover from a disaster, ECC memory, redundant systems, redundant communications, some form of radiation hardening, and could handle temperature swings. Basically, as reliable as they can make it.

As far as I know, the Pi has none of those. Yes, the Pi can probably handle the temperature part, but one wrong bit-flip can crash whatever is running on it and render it unusable by the lack of recovery. They can probably use the watchdog service in that case. But is it enough?

$\endgroup$
10
  • 1
    $\begingroup$ Partly related: raspberrypi.stackexchange.com/q/325/24224 $\endgroup$
    – Aloha
    May 2, 2016 at 13:48
  • 1
    $\begingroup$ Related: stackoverflow.com/questions/36827659/… $\endgroup$
    – Ghanima
    May 2, 2016 at 14:35
  • 3
    $\begingroup$ Pretty sure not all possible orbits are exposed to harsh radiation. $\endgroup$
    – goldilocks
    May 2, 2016 at 15:36
  • 2
    $\begingroup$ @called2voyage That is exactly the case in fact. There are orbits without harsh radiation, but all satellites have to deal with cosmic rays, which can really cause some issues to spacecraft. $\endgroup$
    – PearsonArtPhoto
    May 2, 2016 at 15:53
  • 1
    $\begingroup$ Temporal redundancy and software-based error detection and correction can compensate (at least partially) for less hardware support for resilience and physical redundancy. $\endgroup$ May 2, 2016 at 16:50

4 Answers 4

19
$\begingroup$

Most cubesats are student projects that they want to have for only a matter of a few months at most. That being said, the Pi would likely be very susceptible to Single Event Upsets, SEUs, and thus might have some serious issues. It is low cost, and would fit inside of the spacecraft, with a bit of shielding it might be okay, but only for a few months for a low importance build. Bottom line, it might work for a student project, but I wouldn't have it be the flight computer of anything important!

It is worth noting that a cell phone has been used as a flight computer for a cubesat. This sat was known as PhoneSat. These are comparable to the Pi in terms of SEU and radiation danger. They did in fact work for some period of time. I haven't found a great source, but the lifetime was about a week for every one of the 5 PhoneSat launched. The first 2 only had batteries, the last 3 had solar power as well. They were deliberately placed in orbits that would keep them up only for a few weeks, but I can't find any reported issues.

Small note, it might be better to use an Arduino than a Pi for spaceflight, as the Arduino is somewhat better for integrated electronics. Pi works great if you have access to a video port, but space is a long ways away, the odds of getting an HDMI cable all that way are pretty slim;-) But there might be other advantages to the Pi. At least keep that in mind when making the decision.

$\endgroup$
12
  • 1
    $\begingroup$ It's a lot easier now than it ever has been. You could probably build a decent cubesat for around \$2000, and a pretty good one for around \$10,000, assuming no labor costs. ULA announced they will launch cubesats for free. $\endgroup$
    – PearsonArtPhoto
    May 2, 2016 at 17:07
  • 1
    $\begingroup$ A major reason not to go with Pi is its power consumption, >500mA @5V. Whatever you save on the little board you must spend on extra power source to keep it running - both in terms of mass budget and costs (solar cells). $\endgroup$
    – SF.
    May 2, 2016 at 17:41
  • 2
    $\begingroup$ You can't compare Pi and Ardurino. Ardurino is a microcontroller, the best one having 512kB flash and 96kB RAM. Pi (first model) is a computer with 512MB RAM, up to 32GB flash (SD-card) and a full operating system. Version 3 has 1GB RAM and a 64bit quad core @1.2GHz... My Pi saw monitor&keyboard just once: to set it up for network. Since then, we're communicating via console (plain text). Ardurino can have more I/Os and analog inputs, but it's easy to add them to a pi, too. I also see some advantages of the arduino over the pi concerning radiation. $\endgroup$
    – sweber
    May 3, 2016 at 7:14
  • 3
    $\begingroup$ "Small note, it's probably better to use an Arduino than a Pi for spaceflight," I develop software for spacecraft professionally and this is objectivity false. There are MANY times you need far more power than an Arduino can provide. $\endgroup$
    – Sam
    May 3, 2016 at 13:55
  • 1
    $\begingroup$ The sort of shielding that will fit in a cubesat is basically pointless as it is too thin to reduce radiation significantly. $\endgroup$
    – Hobbes
    May 3, 2016 at 14:28
6
$\begingroup$

Based on what I remember, space-worthy electronics usually feature a way to recover from a disaster, ECC memory, redundant systems, redundant communications, some form of radiation hardening, and could handle temperature swings. Basically, as reliable as they can make it.

Using radiation hardened computer processors is the traditional approach. This results in an avionics system that is rather massive and bulky, very expensive, and very, very slow (by modern standards). Their use is not mandated, even for human-rated systems. SpaceX, for example, does not use radiation-hardened processors in their Dragon vehicle. What is mandated, for human-rated systems, is being able to show that the avionics system is radiation tolerant.

CubeSats typically are low cost school projects that aren't intended to last for long. A \$250000 radiation hardened processor is about twice the budget for a typical CubeSat project budget, including launch. Interconnecting a couple of \$35 Pis for redundancy would be well within budget and would be a great student project.

$\endgroup$
3
  • $\begingroup$ Dismissing CubeSats as "low cost school projects" is ridiculous. There are a number of companies planning massive deployments. $\endgroup$
    – ventsyv
    May 2, 2016 at 19:54
  • 3
    $\begingroup$ @ventsyv - I was not being dismissive. That said, "low cost school projects" is exactly what CubeSats were for the first 13 years of the history of CubeSats. Other organizations have looked to CubeSats for the same reason they are appealing to colleges and universities, their low cost. The launch cost of a CubeSat is quite small compared that of even a micro satellite (10 to 100 kg). $\endgroup$ May 2, 2016 at 22:17
  • 2
    $\begingroup$ @ventsyv: The ones containing Pis (and relevant to the question) almost certainly are! A 'massive' commercial deployment would justify a purpose-designed board and have the budget for one. $\endgroup$
    – FLHerne
    May 3, 2016 at 11:30
3
$\begingroup$

In low orbit the Earth's magnetic field is an effective shield against solar radiation. It sweeps up the radiation and traps it in areas known as the van Allen belts. We avoid putting spacecraft or people there... One part of the belt does dip into low orbit altitudes and it sometimes causes problems. That's most of the radiation flux you have to worry about and you can shield against it as an extra precaution. Don't picture lead. Plastic is used. Lead doesn't work for this type of radiation (or rather, it actually make it worse).

Cosmic rays are more troublesome. They are a much smaller portion of the radiation in space. This includes all the radiation that does not come from our Sun. The Earth's magnetic field offers no protection against most of it. The 50 miles of our atmosphere shields us from about 90% of it. In spacecraft there is no practical shielding against it. Electronic devices are designed to accept or correct for the erratic behavior cosmic rays can cause. But even a device like the Raspberry Pi, built without much fault tolerance can run for a few days or weeks before a critical bit gets flipped that causes the CPU to lock up. I presume that the mission controller have the ability to remotely reset the device. Another simple solution would be to have a separate watchdog circuit simply reset the board once an hour unless it gets a signal to prevent reset.

$\endgroup$
1
2
$\begingroup$

Silicon valley has been moving into the space industry and it's bringing a new philosophy to the field.

The emphasis is being put on quick development time, not reliability. The cost of replacing a relatively cheap cube sat is seen as well worth it the ability to quickly prototype, test and move onto the next iteration of the product.

The mentality is very much "Build it, fly it, test it, repeat"

Most companies that work in this space expect their satellites to last only a couple of years and plan to make up for it in numbers.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.