10
$\begingroup$

I was reading an article regarding the Shuttle's GPCs and how they stack up against commercially-made hardware on http://www.nasa.gov/mission_pages/shuttle/flyout/flyfeature_shuttlecomputers.html and was finding it quite interesting. One thing that caught my attention was that they have a lot of issues caused by radiation with their off-the-shelf laptops (IBM/Lenovo Thinkpads), with the memory in each machine typically getting fried 2-3 times per mission (and a lot more on Hubble missions), with a high probability that it will happen while crossing the South Atlantic Anomaly.

That got me thinking. The Thinkpads, like most off the shelf computers, are plastic. Sure, they have RF shielding, but that exists mostly to appease regulators regarding RF emissions and would offer little to no radiation protection.

However, there are laptops on the market with metal cases (most notably the Apple laptops, made from a milled block of aluminium).

Would an off-the-shelf machine in a metal casing hold up better in space compared to an otherwise identical machine in a plastic case? I understand some plastics can be quite effective radiation blockers, and metals can actually make the situation worse (though I'm by no means an expert on nuclear physics, I'm just interested). Are there any cases on record of a matal-clad laptop going into space, and if so how did it hold up compared to the plastic-clad thinkpads?

Apologies if this is the wrong SE site to be asking this on, but while it is to do with computers I think the answer to this question lies in physics so it seems to me that this is the correct place to be asking a question like this.

$\endgroup$
4
  • $\begingroup$ Electrical Engineering would be valid as well, but since this is about space E&M radiation, I would venture to say it's about equally suited for both sites. $\endgroup$ Commented Jul 15, 2011 at 15:32
  • $\begingroup$ I thought that the radiation only caused the RAM contents to be modified/corrupted - I didn't think it permanently damaged the modules... $\endgroup$ Commented Jul 15, 2011 at 16:46
  • $\begingroup$ I would say this is definitely on topic here, and probably off topic on the trilogy sites - it is about radiation, after all, not computing. I'm not sure if the electronics people would want it. $\endgroup$
    – David Z
    Commented Jul 15, 2011 at 18:42
  • $\begingroup$ Vaguely related: Cosmic Rays: what is the probability they will affect a program? on Stack Overflow. And you'll note that there was considerable argument about it's appropriateness. $\endgroup$ Commented Jul 18, 2011 at 21:53

1 Answer 1

7
$\begingroup$

The bad news: Space radiation is much harsher compared to boring gamma rays from our primitive nuclear reactors. Space radiation has much higher energy levels, and you cannot completely shield it, even with 10 meters of lead (which is in fact not very effective for neutrons).

The good news is that an individual gamma photon, for example, usually would not introduce memory error by itself, even when going right through a memory cell. The real problem is boron. If the stable isotope 10B catches a neutron from space radiation, it will decay into lithium and an alpha particle (and a “harmless” gamma photon). This alpha particle is the one causing soft errors, because it’s heavy and causes huge damage to semiconductor when passing though it.

The solution is to only use 11B in ICs. (Boron is used in packaging and insulation layers on the ASICs.)

Another problem, which in some cases needs special ASIC designs (for special prices, of course) is photocurrent - gamma rays are like light, and will generate electricity right inside the silicon chip (like in a solar cell), which might randomly turn transistors on. In some cases, two randomly turned on transistors might short the power rail to ground, and the chip will burn to ashes (= latchup; usually happens with ‘parasitic’ BJT transistors in MOSFETs).

And finally, a common solution is using ECC memory everywhere, which can autofix errors on-the-go.

$\endgroup$
4
  • $\begingroup$ The only downside to EEC is the slight performance penalty - although the difference is largely negligible, it's worth mentioning for completeness. It should also be noted that this will only work under light radiation exposure; heavy enough and EEC will not help you. $\endgroup$ Commented Jul 15, 2011 at 16:44
  • $\begingroup$ I can take it from your answer then that an aluminium-clad laptop would be just as susceptible to radiation damage as a plastic one? You're right about ECC RAM, it would be better suited to space, but I don't believe there are any off-the-shelf laptops that support it, which is why I didn't bring it up, or ruby ICs for that matter. My criteria for the question is that the computer in question has to be factory-spec, or close to it (The NASA page mentioned unspecified "modifications" to the thinkpads, but those could have nothing to do with space radiation). $\endgroup$
    – GordonM
    Commented Jul 18, 2011 at 8:50
  • $\begingroup$ Sorry, by ruby I meant sapphire. $\endgroup$
    – GordonM
    Commented Jul 18, 2011 at 9:00
  • $\begingroup$ Yes, aluminium is as susceptible as plastic. In fact, plastic does better job at shielding neutrons, which is more serious issue. AFAIK more or less recent AMD-based laptops with integrated memory controller in CPU should be able to deal with ECC memory right out of the box (although might be limited in BIOS). Sapphire IC's are very expensive, so we never see them in laptops, but AMD-based CPU's are made on similar technology (Silion on insulator), which should be less sensitive to radiation. Anyway, on low earth orbit usual off-the-shelf NB might just work, it's not that nasty. $\endgroup$ Commented Jul 18, 2011 at 9:11

Your Answer

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

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