Looking at this pretty awesome video of the CRS-6 first stage almost making it back to the barge, it sort of seems that if something could have steadied it just a bit, it would have been okay. Attitude jets are visible firing at the upper end of the stage at 6 seconds of the video. Does the tip-over point to an issue with the attitude controls, or just a matter of refining them?

  • 2
    $\begingroup$ At some point the afternoon of the landing, Musk exchanged some tweets with John Carmack which suggested they'd determined that slow throttle valve response ("valve stiction") on the attitude jets (the white puffs at the top of the stage) was causing "phase lag" (i.e. late steering response, leading to overshoot and oscillation in attitude control). Those tweets appear to have gotten deleted since -- maybe Musk spoke too soon -- so I can't link them, but they suggested that Musk thought this was solvable on the stage. $\endgroup$ Apr 15, 2015 at 23:04
  • $\begingroup$ so there is no reason to think they are in any way underpowered or could need assistance? I'm sort of thinking i should edit again to make those the focus... maybe... $\endgroup$
    – kim holder
    Apr 15, 2015 at 23:07
  • 1
    $\begingroup$ Well, I'm inferring a lot from statements that Musk decided shouldn't be public at this time, but yes, my impression is that they think this is a control/guidance issue solvable without adding any additional active elements to the system. You can see the attitude thruster valiantly trying to hold the thing up from 0:10 on in the video, and alllllmost doing it. $\endgroup$ Apr 15, 2015 at 23:15
  • 1
    $\begingroup$ I don't think they think it's easy at all. After CRS-4 (?) Musk said something like "next time hopefully it'll blow up for a different reason" and prior to this launch he estimated < 50% chance of success. That sounds like an extremely realistic viewpoint. In the meantime, as someone else pointed out, a failed F9 reusable is still a successful F9 expendable. $\endgroup$ Apr 16, 2015 at 4:16
  • 1
    $\begingroup$ Also, I was slightly mistaken about the tweets -- Musk called out a slow biprop throttle valve (i.e., presumably, the main engine), not an attitude jet valve. $\endgroup$ Apr 16, 2015 at 4:17

3 Answers 3


If you watch the video again you'll see the main engine is gimballing - that is, changing the angle at which it points. It goes through a few sweeps as it attempts to correct itself.

There's a well known problem in physics, the "inverted pendulum" problem, where a body can be held with its upper end vertical, simply by controlling the movement of its lower end. (Try this at home by balancing a broom handle.) This is being done here by sideways deflections of the main engine.

In addition, there are the gas thrusters at the top which can help. (I suspect these assist in crosswinds but I'm not sure.)

What seems to be wrong here is the main engine - either its pointing angle went wrong or it didn't maintain the correct thrust while at an angle, causing the lower end to start over correcting.

Short answer: they need to do a bit more work on the software and on the understanding of the dynamic behaviour of the system (possibly including nonlinear behaviour in the rocket engine). The rocket probably already has everything else it needs to work.

  • $\begingroup$ Issue was due to stiction on the first stage BFV... accounting for control lag was not fully implemented in the landing algo. $\endgroup$ Apr 16, 2015 at 8:13
  • $\begingroup$ Yes Elon tweeted something about valve stiction, but he didn't really explain it... I think his tweet means an error in engine thrust, and not in pointing - but I would prefer to wait for a fuller explanation to emerge, which is why I didn't want to quote this. :) $\endgroup$
    – Andy
    Apr 16, 2015 at 8:21
  • 1
    $\begingroup$ The video seems to show the rocket overcompensating, the last portion of the main engine burn had too much thrust (valve stuck open, presumably). Because the engine was gimbaled, this resulted in sideways velocity as the rocket touched down. $\endgroup$
    – Hobbes
    Apr 16, 2015 at 11:58

If you ignore the engine in the video and instead watch the F9's body, you'll see it go through more than a full cycle of oscillation (counter-clockwise, clockwise, counter-clockwise), and those oscillations appear to be increasing in amplitude as it approaches the ground.

Motion like that is not typical of a controlled landing.

Thus, in the unlikely event the F9 was successful at avoiding a tip-over, it would have been by sheer luck, because the trend was one of losing control, not regaining it.

Until there is more information released on what caused the loss of control, it is not possible to say what would have saved this particular landing attempt.


The reported issue is that the valves controlling the Merlin 1D had some internal friction (Stiction) that meant it was slow to respond to computer requests. It was ordered to correct one way, was slow, so over-corrected. The computer ordered a correction to fix it, was slow, over-corrected the other way. Until it came in as you see.


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.