The Age of Asteroids | Hacker News Reader

The Age of Asteroids

newyorker.com

26 pointsiKenshu9 comments

One statement that stood out to me was Chris Lewicki's dig at NASA's "twenty year old computer" running Curiosity, with Lewicki adding the usual jab about his iPhone having more processing power.

The RAD750 is based on a 1996-era PowerPC design, so the statement is technically true, but as far as I know it's basically the state-of-the-art in rad-hardened processors -- it's the same processor used in Boeing's 787 FMC, for example. (SST did just ship a rad-hard M0, which is kind of neat.)

Lewicki's not just talking off-the-cuff here -- he's one of a few people actually qualified to opine on spacecraft engineering -- but I wonder, what does he think should be used in place of the hugely-expensive, arguably-antiquated but undoubtedly robust systems currently in use?

NASA and other space agencies usually build a single satellite for a single task, and it will cost $1B+. Chris Lewicki and other "newspace" companies are suggesting making cheap satellites, but lots of them. If one fails, maybe you lose $1M but you certainly don't lose $1B.

This gap makes it possible to take risks and use newer hardware, specifically hardware that has not been radiation-hardened or even tested in orbit.

Sounds like that's a fundamental risk assessment that needs to be made ASAP. If you're talking $1 million satellites, and this denigrated state of the art most powerful rad hard CPU board goes for a reported $200K, then if the non-tested, non-rad hard (and don't forget temperature extremes), more powerful CPUs just don't work, or don't work for certain missions, you need to find that out.

These systems are not designed by idiots.

Normal CPU's don't last long enough in space to be all that useful for satellites. And even for profit telecom companies stick with radiation hardened CPU’s so it's not like this is a case of 'military' over engineering.

As far as I could tell, the low budget amateur radio types also use rad hard CPUs in their satellites.

So, what's Mr. Lewicki's alternative?

Here is SpaceX's: https://news.ycombinator.com/item?id=8746788

SpaceX does not need radiation hardening for rockets which are only briefly in space. That's hardly the equivalent of satellites which need to be in space for years.

And it's worth mentioning that both approaches make sense for the application being considered.

When you are sending a suite of purpose-built instruments and a one-of-a-kind rover to Mars, taking advantage of a narrow launch window, to accomplish a very well-defined scientific mission, you require the spacecraft to be safe.

If you are sending a bunch of less-expensive spacecraft to perform a less intensive investigation of a bunch of asteroids, you might be more risk-tolerant. But as the cost of the instrument suite goes up, you'd have to be more risk averse.

Additionally, if you expect to be in a difficult radiation environment, things change. LEO is not a difficult environment, but deep space is.

It's a systems engineering problem, and there is no single correct approach. It's an engineering question, with the usual set of tradeoffs, expertise, and modeling. NASA and private industry have been studying this for decades now. (I spent a couple of years working on a software-implemented fault tolerance scheme for science algorithms in spacecraft.)

The most surprising single fact from this article, in my opinion, was to learn that Queen's Brian May is an astrophysicist! That fact alone distracted me slightly while reading the rest, which I also found to be interesting.

Reply on news.ycombinator.com