Hacker News Reader

Comment by orbital-decay | Hacker News Reader

Back in the days rockets were more or less unique, because the tech was new and every bit of efficiency mattered. But the common trope among the designers was always "launch on a modular rocket, assembly from pieces in LEO, depart anywhere". And orbital assembly and modular launchers have followed in 1980s/90s, when the extreme optimization didn't matter so much anymore.

It's interesting to see SpaceX taking the path of a custom-tailored launcher again, trying to integrate everything as tightly as possible (especially speaking of their direct Mars flight concept). This would probably be impossible to scale without designing a new system from scratch, so they have to get everything right. Kind of the Amiga of the spaceflight.

View on HN

The heavy booster can be the same, and any mission beyond LEO will require refuelling, and the tankers can be standardised.

I think the terminal mission component is always going to have to be customised to the mission type. A Lunar lander and Mars lander are very different beasts, but even just basing those on a basic Starship hull is a big win over completely custom from scratch systems such as the Blue Origin lunar lander.

The big idea with Starship, at least for missions going beyond LEO is to drop upper stages in LEO as fuel tanks and use them to fuel the "payload" Starship. In the Artemis mission, it is supposed to take astronauts from the Orion capsule, land them on the moon and back to Orion, with talks about a "gateway" space station. So Starship definitely follows the modular / orbital assembly principle.

The idea of purely modular launchers (like OTRAG) never really took of. Armadillo Aerospace tried to capitalize on that idea, but they never really went beyond John Carmack's hobby project. All attempts at anything but "extreme optimization" failed, including the other end of the spectrum with the humongous Sea Dragon. That's because of the so called "tyranny of the rocket equation" and considerations like the square-cube law.

What I meant is not things like OTRAG (which was an abomination indeed), but practical configurable launchers unified to the certain degree into a family. Think Delta IV, Atlas V, Angara, or even Falcon 9, which is the only surviving rocket from the family that was originally planned.

The reason modularity of that kind wasn't especially successful isn't the tyranny of the rocket equation, but the economy. Turned out the launch market is too small to capitalize on fringe cases, and the added flexibility simply haven't outweighed the loss in efficiency. When the modularity was starting to pick up steam, most thought that the market will expand much faster, and few people expected Iridium to be a miserable failure, for example.

> But the common trope among the designers was always "launch on a modular rocket, assembly from pieces in LEO, depart anywhere".

It’s a design constraint imposed by the reality of launching from Earth. It puts an upper bound on how much weight you can send at one time. That’s pretty much the only choice until we can manufacture directly in space.

What's the source of the constraint on rocket size that you refer to?

The tyranny of the rocket equation means it starts to get crazy if you can't improve the payload fraction and want to send up more mass. I don't think there is a literal upper limit, but you very quickly get to the point where a few extra kilograms means you have to double the size of your rocket.

Isn't Starship planned to launch 100 tons into LEO, just with a two stage rocket, and with both stages being reusable? I don't know the numbers but this seems to suggest efficiency (cost per ton to orbit) improves with increasingly large rockets.

(Though I imagine there could be a size sweet spot beyond which efficiency shrinks again.)

And it needs to refuel to get to Mars despite being the largest rocket ever built and an incredible feet of engineering. The limits are height and space to put engines under it.

But that doesn't limit the rocket size in any way, only the fraction of its mass that can be used for payload. That limit does not depend on total mass of the rocket. Is there something that makes larger rockets less feasible than smaller rockets with the same payload mass fraction?

> "launch on a modular rocket, assembly from pieces in LEO, depart anywhere"

That's still what they are doing, its just that the pieces are fuel.

> This would probably be impossible to scale without designing a new system from scratch

Its not like the amiga of space flight, its like cloud. Instead of horizontal scaling, its vertical scaling. Instead of of building bigger, you just build more and scale that way.

They will add like 10m to the Starship, there is some horizontal scale left.

Explain Amiga analogy please.

The Amiga had a litany of custom chips to achieve its incredible performance. The Mac and PC used far less custom ICs.

> The Amiga had a litany of custom chips to achieve its incredible performance.

Exactly, and with PC technology rapidly catching up, updating them to a level on par with the latest VGA and sound cards was an enormous effort that didn't kill Commodore (internal mismanagement did) but came too late to save them. First prototypes of the AAA chipset are mentioned and visible in the famous "Deathbed Vigil" video shot by Dave Haynie himself at the C= headquarters in West Chester PA when the company was already belly up. I would highly recommend watching it also to non ex Amiga users, also for interesting vintage technical content, but keep tissues ready in case you were.

https://www.youtube.com/watch?v=BaTjwo1ywcI

Let me expand on squarefoot's answer.

When the Amiga launched in 1985 its hardware was unquestionably the state of the art in microcomputers. Macs didn't have color yet, and the best consumer PCs had 16-color EGA graphics and a beeper speaker (or 16-color Tandy graphics and 3-voice sound).

While Amiga stood still, IBM in 1987 introduced 256-color VGA, which the industry rapidly expanded to Super VGA with higher resolutions and more colors. By 1990, it was easy to buy a clone PC with Super VGA and sound card, both from a variety of third-party vendors and both exceeding Amiga's capabilities. While Amiga still had a theoretical advantage in terms of hardware sprites, in practice PC horsepower increased so quickly that its lack of same didn't matter.

To put another way, there is no way a single company with a relatively small customer base can compete against a gigantic ecosystem of many dozens of vendors, all battling to serve a customer base ten times as large by constantly lowering prices and improving capabilities.

You're more or less on point, but the conclusion doesn't follow the premise. It is not a good analogy.

When the original Amiga team left Commodore, they left a finished new chipset (Ranger), with very competitive specs.

However, Commodore discarded it as too expensive, not attributing any value to being first to market or having the best product.

Management systematically acted against the engineers, such as when they unilaterally decided not to have a DSP and forbid the engineers to use the A3000 prototype boards with DSP on them that they had made.

There's multiple books about the story of the Amiga and its mismanagement by Commodore, as it extends much further than that. Lots of cancelled projects, many of which finished like Ranger was.

While custom, it is remarkable the gate count was low.

They were simple and very cleverly designed.

Reply on news.ycombinator.com