4

I'm working on a cubesat project intended to operate in low earth orbit. We're considering a few different options for our flight computer, ranging from Raspberry Pi to something like STM32L476. Despite being told repeatedly that "Raspberry Pis have a decent heritage in space," I'm not fully convinced.

It's hard to get a good picture of just how hazardous the conditions in LEO are to electronics.

  • What range of radiation doses would we expect?
  • How spaceworthy are microprocessors and microcontrollers for non critical and somewhat cost sensitive space applications?
  • Would a watchdog timer be reasonably sufficient to protect against faults at lower radiation doses?
  • More specifically, assuming we design our software accordingly and use STM32's watchdog peripherals, how long should we expect to survive in LEO from strictly failure caused by radiation?

Here are some resources I found, but they don't answer my questions completely.

Some more information about my particular situation:

  • We are very sensitive to cost so we cannot use all space-rated components for no reason. Vorago's radiation hardened MCUs that are about $1k per piece are out of our consideration as well.
  • Thermal issues will be considered separately, but I have a reasonable idea of what to expect and how to perform testing. It's also quite easy to test compared to ionizing radiation.
crossroad
  • 1,390
  • 7
  • 19
  • 6
    > *Despite being told repeatedly that "Raspberry Pis have a decent heritage in space," I'm not fully convinced.* Thanks, me neither. It's a mystery to me why someone would pay several thousand € for a rocket ride but insist on using the cheapest computer available to them, based on *checks notes* a TV set-top box SoC that's not even rated for industrial thermal range, on a board that multiple high-speed design experts have on separate occasions and independently called "a surprisingly well-working piece of amateur art" in discussions with me. – Marcus Müller Feb 08 '23 at 08:53
  • From the software/microcontroller point of view check out this thread on SO: [Compiling an application for use in highly radioactive environments](https://stackoverflow.com/questions/36827659/compiling-an-application-for-use-in-highly-radioactive-environments). – Lundin Feb 08 '23 at 08:57
  • 2
    NASA is your friend. https://llis.nasa.gov/lesson/824 – Peter Smith Feb 08 '23 at 08:59
  • And to quote myself in the SO link above: "In general, use a MCU suitable for these kind of systems, and not some generic mainstream fluff you received in a packet of corn flakes." – Lundin Feb 08 '23 at 09:03
  • Crossroad, Usually if you are being considered for a cubesat you will have a contact to help out with these questions or else you are expected to have such an expert within the team. I worked with someone (Robert is his first name) who is an expert on the LEO environment -- decades of experience. I know some of his horror stories but I don't remember concrete details. What's the expected project lifetime? – periblepsis Feb 08 '23 at 09:19
  • TonyM I'm attending college in California. I do realize its true that RPi have flown in space many times, but the consumer level quality put me off. @periblepsis To be honest, we're just a team of university students. We don't have access to very much resources beside a ticket to space. We're aiming to fly within the next few years but the exact timeline is still tentative (Actually our launch was delayed many times, so now we are reconsidering the whole approach. Structural design is basically done but electronics wise we are "open-sky") – crossroad Feb 08 '23 at 09:22
  • @TonyM nope, not myself, but I have close friends working in the new-space segment at a larger southwest German satellite manufacturers, plus multiple friends with DLR comms background. That's how I get to drink tea or beer and talk about satellite failure models occasionally. – Marcus Müller Feb 08 '23 at 09:22
  • @TonyM ah, AstroPi! Yeah, that's a thing :) but it's a sensing payload, not a flight computer – in essence, most bits it handles are foreseeably image data, so that bitflips are tolerable noise in the measurement data, not e.g. a satellite losing attitude control. – Marcus Müller Feb 08 '23 at 09:26
  • 1
    Not an answer to your question, but you can get around any single event upsets with redundancy. SpaceX uses bog-standard X86 chips with triple redundancy, but their design is based around satellite redundancy as well by having thousands in orbit so having one suffering all three X86'es fail is accounted for, a luxury you don't have in a cubesat. What you probably need a number for is the risk of permanent latch up. – winny Feb 08 '23 at 09:26
  • Oops @MarcusMüller, I first read this question while half awake and didn't register the big important words 'flight computer' so I've deleted my comments as they're daft in that context :-) I've got a fast processor/FPGA board flying up there, as well as a couple of FPGAs at root control level. As I said, it's surprising the commercial tech that's up there in microsats now but, as you say, not for a flight computer. – TonyM Feb 08 '23 at 11:56
  • @TonyM - But the commercial tech that's up there in microsats is in missions that are a lot more forgiving than what some of us are dealing with. There's world of difference between throwing something up there and hoping it works for a couple of months and guaranteeing, with some degree of certainty, that an assets will perform it's function over 8 years or more on orbit. – SteveSh Feb 09 '23 at 14:15
  • @SteveSh, yes I know - had you not read my design work mentioned in my comment? – TonyM Feb 09 '23 at 16:07
  • I wasn't questioning your experience, but just emphasizing the differences between short duration microsats and long duration applications. – SteveSh Feb 09 '23 at 16:34

2 Answers2

1

What range of radiation doses would we expect?

You need to find out and write down your specs (what level of radiation are we looking at?, what are you trying to do?), temperature/cooling considerations (although temperature is not actually defined in space). Please research further for your specifications.

Additional edit: The amount of radiation exposure a satellite in low Earth orbit (LEO) experiences depends on several factors, such as the altitude of the orbit, the shielding of the satellite, and the current level of solar activity. Further research is needed and more information should be available to define the radiation that the electronics will have to handle.

An example of radiation based on altitude:

Effective dose rate depending on the altitude

Image Source: https://www.researchgate.net/publication/253302207_Space_Radiation_Measurement_on_the_Polar_Route_onboard_the_Korean_Commercial_Flights

If you would want to do a professional job using raspberry pi, you must look at each one of the raspberry's ICs to check what each can handle (based on your specs).

How spaceworthy are microprocessors and microcontrollers for non critical and somewhat cost sensitive space applications?

They all have their limitations and maximum radiation tolerances. Checking the datasheets of space graded ICs should solve your question. Space graded ICs have just different silicon inside.

Would a watchdog timer be reasonably sufficient to protect against faults at lower radiation doses?

I suggest using space graded ICs instead of trying to look for workarounds to that. You will avoid hassle and brain pain if you use proper ICs to do your job instead of looking at the errors later on from the watchdog.

----open source project(s)----

There are people making satellites already. If you follow a open source project like the below, you might see what ICs they are using and take some ideas: https://gitlab.com/librespacefoundation/upsat

----Videos----

Maybe this could help: EEVblog #896 - Space Electronics

And this: EEVblog #897 - Radiation Effects On Space Electronics

This video starting from 4:50 about space graded: EEVblog 1435 - The World's Most EXPENSIVE Catalog IC ?

Christianidis Vasileios
  • 2,645
  • 2
  • 9
  • 32
  • Just to address the first part, "what level of radiation are we looking at" is what I'm wondering, as all I know is that we're going to LEO. I will deal with temperature issues separately, but I already have a good idea of what to expect and how to thoroughly validate that aspect. – crossroad Feb 08 '23 at 09:08
  • By the way, another issue that I barely mentioned in the question, is the fact that we are somewhat sensitive to cost. Vorago makes radiation hardened MCUs that I'd be very interested if not for their $1000 per piece price tag. I cannot justify spending that level of money without knowing closer to the exact odds that we're dealing with. – crossroad Feb 08 '23 at 09:18
  • 1
    I suspect space grade IC's are a non starter for OP's project due to 1) cost and 2) availability/delivery times. A lot of providers of space grade IC's only build them as needed and so lead times can easily exceed 52 weeks. – SteveSh Feb 08 '23 at 13:49
  • 1
    @SteveSh On top of that I think OP should research more. Its difficult to build a space graded electronics project while your questions are "how much radiation I will be facing?". Research is needed beforehand or a lot of money will get wasted. – Christianidis Vasileios Feb 08 '23 at 14:36
0

I'm going to take the first question.

What range of radiation doses would we expect?

The radiation environment in LEO is very well characterized and understood. Given the construction of your cubesat, the materials used, and the details of your orbit (for example, does it go through the SAA) you can come up with a fairly good estimate of what level of the various types of radiation your parts/design will see.

For more details on the LEO radiation environment, I would suggest you investigate the JPL or GSFC (Goddard Space Flight Center) websites.

SteveSh
  • 9,672
  • 2
  • 14
  • 31