Open Source Rocketry

I recently stumbled across some fascinating videos by amateur rocketeer Joe Barnard, whose YouTube channel is chock full of interesting projects. Armed with a 3D printer, model rocket components and some fairly simple custom electronics, he has created some amazing results. One interesting video series is his model rocket silo project (more video links given later in the article), including the launch of a fin-less vectored-thrust rocket from that silo that reminds one of a submarine-launched ballistic missile.

What really caught my eye, though, was his three-engine vectored-thrust Falcon Heavy model, shown to the right in mid-flight (the center engine did not ignite during this flight). In that pic (taken from a video linked far below), the thrust vectoring for this fin-less model is clearly visible, particularly with the right-most engine. Other test flights show more dramatic vectoring, more on this later. To his credit, Joe doesn’t filter out his failures, but instead documents his process, warts and all, including crashes, flameouts, fires, control losses and so on.

Joe’s work is a good example of an idea that has been bubbling around in my head for a while:

Modern technology, particularly open-source software and hardware, can allow implementation of advanced weaponry, at a small nation-state level, on par with first-world military weapons, with only about a decade or two lag, and constrained only by the available budget.

Joe’s rockets are missing three things to add smart missile technology to a small nation: scale, power and control algorithms. The first two are merely budgetary issues; scaling his airframes and engines is merely a checkbook problem, as is mass production. After a certain point, these things (including off-the-shelf warhead and materials science technology) do not improve much with increasing budgets; economies of scale merely make them cheaper. The third element, control algorithms, is where all the excitement lies, and is almost free, compared to the other two. Further, with the rise of open source software (such as various guidance and flight control software packages) and computing hardware (particularly with the introduction of the RISC-V platform), this genie has burst completely out of a naive and arrogant arms control bottle.

The United States, particularly its political class more so than the technologists, has a long and well-documented history of arrogance with presuming a special capability with respect to military technology. The most famous example of this arrogance was the Manhattan Project, where the political leadership believed that the US-UK nuclear axis would retain a nuclear monopoly for decades, despite warnings from the nuclear engineers and physicists who knew better. Physics and math work the same for everyone, and once German nuclear physicist Otto Hahn published the results of his 1938 fission experiments, that genie was already out of the bottle. The rest was just budget and engineering. Even if Hahn hadn’t published those results, physics at the time was ripe for the discovery of fission, so it would have been discovered independently by many other physicists within months anyway. Science and invention is like that: ideas get ripe when their time comes, and many minds come to the same conclusions very quickly. Papers and patents only document “first”, and sometimes only by the slimmest of margins, although that distinction usually doesn’t count for very much, given that the US, not Germany, was the first to use nuclear weapons in war.

Espionage makes a difference, but only in terms of cost and schedule, and even so, early adopters usually pay that toll the heaviest. A demonstrated fact that a thing can be done is usually enough to spark the innovation while early adopters pay for a lot of redundancy and blind alleys that later adopters do not. Early adopters also pay for development of processes and practical field models, while later adopters are free to innovate on that foundation at much lower cost, usually by simply studying public photos, videos, official statements and observable deployments. Early adopters must sift through and pay for a large number of options from a practically unlimited menu, while smaller nation-state later adopters can tailor their efforts to al a carte items specific to their needs. This is why the US spent decades and untold amounts of R&D and fielding costs to produce stealth and drone technology, while later adopters seem to almost flippantly introduce sufficiently capable options at much less cost and much more quickly. GPS, cruise missiles, phased array radars, data-linked command and control, stealth-piercing radar, you name it. Same, same, same, same, same.

It has been decades since I have held a security clearance, but during my 1980s-era Naval Academy courses for my Control Systems Engineering degree I was often struck by how modern control algorithms, developed predominately during the 1950s and available as public domain well-published knowledge, can be applied in straight-forward ways to practically any control problem one might imagine. Advancements in computing technology since then have only affected the speed at which control loops can be operated, and the power requirements to accomplish these tasks. In the case of guided missile technology, the required computing power hit about the size of a thumbnail somewhere in 1982 or so. The physics of guided missile control are relatively low data rate kinds of problems, so the major advancements since then have been reducing power consumption (and thus reducing size and weight, or alternatively increasing range and payload) and improving sensors and actuators (thus increasing accuracy, maneuverability and survivability), all of which matured in the very early 2000s.

From a controls perspective, all that Joe is missing for his multi-engine vectored-thrust rocket is the idea of a state observer model, from which the actions of all his engines can then be coordinated. He has the computing power, he has the actuators, he has the sensors. This one idea, which replaces the individual cookie cutter PID loops, as they are known, is like a hot-rodder replacing stock items from under the hood but otherwise leaving most of the car intact. The actual control loop details, based on a well-studied missile problem known as the inverted pendulum, have been available for about sixty or seventy years now, and can be simulated and tested fairly well using open-source software tools once the state model for his rocket has been determined. This latter process is also accessible using open-source software tools and some fairly simple bench and flight model testing to determine various state parameters.

The point is not to criticize or arm-chair manage Joe, the point is that going from Joe’s rockets as they exist today to a small nation-state weapons program is a fairly small and open-source step now, despite having at one time been a large and vainly classified leap from Hitler’s crude ballistic and cruise missiles, jet interceptors and other drawing-board concepts such as surface-to-air missiles. The math was more or less complete by the mid-1950s, the computational power available by the mid-1980s, and the sensors and actuators readily available in the early 2000s.

These things now, quite literally, no longer require rocket scientists.

As promised, here are the links to some of Joe’s rocket project videos. First the silo development project:

Next, launching the fin-less rocket from the silo:

And finally the impressive Falcon Heavy Model flight #2, with lessons-learned:

Leave a Reply

1 Comment on "Open Source Rocketry"

Notify of
Sort by:   newest | oldest | most voted

[…] Joe’s work is a good example of an idea that has been bubbling around in my head for a while:MORE […]