Didn't this use mostly off the shelf parts? If so, I wonder how this will impact costs on future missions. If they can do more with consumer hardware, they can save budget to apply toward more science.
Applied Ion Systems is throwing these kind of rocks with small-scale electric space propulsion. It's interesting to see both the excitement and energy from eager researchers and hardcore hobbyists (cubesat folks), and the oftentimes rude and nasty pushback from industry.
There is a lot of cost savings in taking COTS parts and qualifying them for space vs. designing new space-qualified parts, we will see more of this in the future especially with expensive niche technologies with a lot of crossover such as optical communications.
Yes, it did. I like the sentiment but I wonder how much conflict of interest would undermine this idea. Imagine how many companies are involved in developing space grade one-off hardware! Also, why would a highly bureaucratic structure undercut the amount of money that they themselves are asking for (and receiving) out of a budget? Savings are not aligned with the interest of such structure. Its not that for the amount you have saved you can allocate rest of the funds for something else (usually this is how it works with publicly funded projects AFAIK)
Every bit of money and platform resources (rad hardened CPUs are giant, slow and power inefficient compared to even semi modern COTS) is money and resources that NASA can spend on scientific payload on the same platform.
NASA absolutely does have some incentive to find savings in control hardware and software.
Finally, while Ingenuity does use a non-hardened Snapdragon, many other of its critical electronics components are still rad-hardened. The FPGA and dual MCUs (that actually do the low level control and I/O I assume) are both rad hardened. In addition, the COTS components that were used where screened by NASA for their performance in radiation.
The Snapdragon is really just there to control the radio, and do image processing. Critically, these are functions that have -some leeway- for timing, giving the option to just restart the Snapdragon if a watch dog detects a problem.
All of this to say is that rad-hardening isn't going away, but will probably stick around in many critical niches. What Ingenuity absolutely do is validate that modern COTS processors have a role to play in radiation elevated environments, including in semi-critical applications.
HN is dominantly a web/SW crowd plus some mobile frontend, and "it uses a Snapdragon" gives many a wrong idea. In embedded device projects a lot of time is spent planning and designing around a heavy compute element running Linux like this, especially if the device has a safety concept or other mixed criticality concerns. It will have a substantial moat around it.
On HN if you say "systems architecture" most folks go "Oh you mean like, whether we use microservices?". In embedded, while there is a lot of overlap and analogues, it's also all of the above, plus power state management and other aspects. It's not very shiny, but that profession makes all your cars, airplanes and alien planet multicopters.
"It's not very shiny, but that profession makes all your cars, airplanes and alien planet multicopters."
If you have a good head for it, it's a pretty darn good career. You might not make $500k/year like you would at google, but the money is still decent and reliable.
This also drops the bar for other space agencies, from other countries and private alike. Getting cheaper hardware also means more launches and more testing. Instead of sendind a multi million project to space, you can send basically a smartphone (an epheumism, ofc) and some big antennas, and do it in bulk
"off the shelf" in aerospace means that you can buy it from an aerospace manufacturer, and don't have to build it in house. There's considerably more engineering behind these products than the equivalent consumer electronics
The original Elon Musk biography describes how a legacy aerospace engineer joined spacex and was tasked with a part:
> He got a quote back for $120,000. “Elon laughed,” Davis said. “He said, 'That part is no more complicated than a garage door opener. Your budget is five thousand dollars. Go make it work.’”
From some perspectives, yes. From others it's not so bad. I love having cover from the top to do engineering and qualification to have better solutions. Normally it's "we don't have time or resources to make it, get back to the spreadsheet mines".
For LEO you can scoot by pretty easily with non-hardened solutions and better systems engineering and software. For deep space you'll need to be more clever.
> It got there by challenging conventional wisdom, she adds. Most launchpad air conditioning systems, for instance, cost nearly half a million dollars, but SpaceX execs wondered why it cost so much more to cool an area the size of a conference room than the $75,000 it cost to cool their entire headquarters and manufacturing plant. The company brought the cost down to about $35,000, says Shotwell.
I wouldn’t take them seriously, except that SpaceX has a full decade of very reliable launch services at a fraction of the cost. You might ask how they did that, and SpaceX says “here are some examples”. It’s not just reusing rockets.
If there’s any company you should pay attention to when they report successful cost cutting, it's spacex. You can hate or love Musk but Gwynne Shotwell is the real deal there. Not sure how she keeps Elon from blowing the whole business up but somehow she does and quite successfully at that.
If she says you can keep cool whatever needs cooling for 10% of the Boeing price, you’d better believe it.
Consider how many governments signed 8+ figure contracts to develop apps. Why would this kind of waste be any different in other industries? It's humans all the way down ;)
Not sure I understand your response. There is a tremendous amount of launch data from several years of launches (and landings) that demonstrate whatever approach to design and manufacture SpaceX is using results in incredibly robust and capable space systems. I think rather than fear, a better response is to see what can be learned from SpaceX and see if it can be applied elsewhere.
That depends on the validation and QC of their internal engineering. It is just as terrifying to spend USD120000 on a part that might be no better that what you can design&build in-house.
Here's the background:
There were numerous ways in which SpaceX's strategies diverged from space industry norms, and almost all of them had direct implications for the cost of its launch systems. First, whereas most aerospace companies give their designs to myriad third-party contractors who create the hardware for them, SpaceX produced roughly 80% of its launch hardware inhouse. SpaceX builds its own motherboards and circuits, vibration sensors, radios and more. In most industries vertical integration increases the costs of firms by not enabling them to benefit from competitive bidding between efficient suppliers. In the aerospace industry, however, the entrenchment of norms around using parts specialized for the space industry ("space grade"), and the bureaucratic rules defined by government contractors, had kept supply costs high — very high. SpaceX decided instead to build many of its own parts, or to buy parts not considered "space grade" and modify them to achieve "space grade".For example, rather than paying $50,000 to $100,000 for an industrial-grade radio, SpaceX was able to build its own for $5,000, and shaved 20% of the weight off at the same time.
SpaceX's willingness to produce their own parts came as a shock to suppliers. For example,Tom Mueller recounts a time when he asked a vendor for an estimate on a particular engine valve: "They came back [requesting] like a year and a half in development and hundreds of thousands of dollars. Just way out of whack. And we're like, 'No, we need it by this summer, for much, much less money.' They go, 'Good luck with that,' and kind of smirked and left." Mueller's team created the valve themselves, and by summer they had qualified it for use with cryogenic propellants. "That vendor, they iced us for a couple of months," Mueller said, "and then they called us back: 'Hey, we're willing to do that valve. You guys want to talk about it?' And we're like, 'No, we're done.' He goes, 'What do you mean you're done?' 'We qualified it. We're done.' And there was just silence at the end of the line. They were in shock." As noted, a big factor driving savings at SpaceX is that it often builds its components out of readily available consumer electronics rather than equipment alreadydeemed "space grade" by the rest of the industry. Twenty years ago "space grade" equipment would have had far superior performance characteristics compared to consumer electronics, but today that is no longer the case-standard electronics can now compete with more expensive, specialized gear. For example, at one point SpaceX needed an actuator that would steer the second stage of the Falcon 1. The job fell to engineer Steve Davis to find the important part, and since he had never built a part like that before he sought out suppliers who could make it for them. Their quoted price for the device was $120,000. As Davis recalls, "Elon laughed. He said, 'That part is no more complicated than a garage door opener. Your budget is five thousand dollars. Go make it work.'"20 Davis ended up designing an actuator that cost $3,900. Another example is provided by the computers that provide avionics for a rocket. Traditionally NASA's Jet Propulsion Laboratory bought expensive, specially toughened computers that cost over $10 million each to operate its rockets. Musk told engineer Kevin Watson that he wanted the bulk of the computer systems for Falcon 1 and Dragon to cost no more than $10,000. Watson was floored,noting, "In traditional aerospace, it would cost you more than ten thousand dollars just for the food at a meeting to discuss the cost of the avionics."21 Watson was inspired by the challenge, however, and ended up creating a fully redundant avionics platform that used a mix of off-the-shelf computer parts and in-house components for just over $10,000. That same system was then also adapted for use in the Falcon 9.
About the actuator:
Steve Davis, the twenty second hire of SpaceX, needed an actuator that would trigger the gimbal action used to steer the upper stage of Falcon1. He went to find some suppliers and got a quote a $120,000. “Elon laughed”. Davis said. “He said, ‘That part is no more complicated than a garage door opener. Your budget is five thousand dollars. Go make it work.’” Davis spent nine months building the actuator and the final actuator approved by Musk ended up costing $3900.
Seems like they don't include the engineering time in the part cost calculations - so cheating a bit!
I've experienced building something in-house that is far better than what you could otherwise get. Back in the 00's I wrote a JavaScript framework because the existing ones were all crappy in a variety of ways. Even as a one-person effort (and I'm no 10x engineer) I could write something that was wayyy better in a bunch of important ways (albeit not pretty enough design). My work was engineered better than the open source and commercial frameworks that I evaluated/used. Better loading, better recovery from network and other errors, better memory behaviour, better size, better speed, better integration, better diagnostics, better browser support, better user interface, customised for our needs. It did exactly what we needed for our project and mostly worked flawlessly.
It does bode well for sending cheaper "nice to have" experiments on missions, though.