SpaceX Falcon 9 v1.1 - SpX-6/CRS-6 DRAGON - Discussion Thread

[meekGee]

It is unarguable that active guidance (the rocket doesn't need to communicate with the barge for that) is more precise, since by using absolute coordinates you add the errors in barge and rocket station navigation.

I don't know if SX have an active guidance system or not but I can provide a data point.

Radar altimeters can provide accurate height above ground and velocity. In the 1950's 4 antennas (front, back, left, right) at slight angles to the airframe to give deduced "ded" reckoning were simpler and more reliable than inertial systems (back when INS meant spinning metal lumps in very precise bearings).

Modern units use mm microwaves and are available for drones. They have ranges in the 500-1000m and weigh a few Kg.

Such a system requires no comm link to the barge and should work properly against any surface you'd choose to use as a landing site IE no boulders strewn across it.

Thanks.

I agree, if there's a terminal guidance system, it doesn't require comm to the barge.

There are plenty of ways to implement it (I like the one you describe) and it will increase accuracy.
We are also pretty sure that a Z-axis terminal guidance system is in fact implemented.

Jim says he has inside info that an X-Y system isn't implemented.

That's a good place to leave it...

[dgates]

I would note in passing that they apparently did not run out of hydraulic fluid this time!  I guess they can cross that problem off the list! 😀

[iamlucky13]

I am wondering how Carmack made this determination. Was he given the data by SpaceX engineers, or is it based on his own experience at Armadillo with a totally different rocket?

I can see why Musk respects Carmack, but I can't see how you can say it was a stuck valve by just looking at a few images and then the Vine clip. So, it is a SWAG, or else he is being given data that was not publicly released.

The tweet is from Musk to Carmack

I was able to read the tweet last night, but now I'm getting a "page does not exist" message from Twitter when I try to access it again.

I wonder if Musk was just flipping jargon at Carmack for fun and took it down when he got flooded with questions about what it means.

[OxCartMark]

What's the standard remedy for valve sticking?

A spritz of this...

Before I open my yap on this I want to preface my comments by saying that I'm strictly responding to the question in the first quote above (edit: at least initially, before I broadened my comment below) and not trying to tell the valve experts that are at work on this how to do their job which they are certainly vastly more qualified than I to do.

Any proportional valve in general industrial, automotive, or similar use will have stiction.  If you were to vary the input signal by simply ramping up a DC current you'd find that the output would change in discrete jumps.  For many valves in many applications (that probably amounts to most) the jumps in output are too large to be acceptable.  So rather than send a varying DC signal to a proportional valve its common to superimpose an AC signal or to PWM the signal such that the armature and valve element (typically spool) aren't just being asked to move when the valve output is to change but rather they are always in some small motion.  This way the friction applicable to moving the valve from one desired setting to another is dynamic friction, not static friction.  The video below will show some detail on this (action starts at :36).  Note that while this solution works for many situations the dynamics may be upsetting for a rocket engine, I don't know.  Also, dither can be applied to a solenoid operated valve but wouldn't be a solution to a valve with ball screw drive or similar.  In any case the people that are dealing with this know so much more about this subject than I do that I'm feeling exposed in just writing these words.



Also, recall that CRS-1(?) had a stuck valve that took a lot of cycling / hammering before it became active.  From what little info I have seen posted, I think that valve was supplied by the same supplier that provided the propellant valve now in question.  If its true that one supplier provided both stuckified valves then there might be some movement on SpaceX's part to bring this in house.  But the time to develop a valve in house would probably be much longer than the time it will take the supplier to modify and qualify their existing valve.

[llanitedave]

Had he mentioned the Turbo-Encabulator maybe there would have been fewer questions.

[cambrianera]

I need to watch the video a few more times to see if I can make out the moment of leg deploy, but I have to wonder if the late deploy, into a steady sea breeze, isn't what causes the pitch excursion due to the sudden increase in drag at the base of the vehicle relative to the wind. Those rapidly-telescoping legs and their bases have much greater area drag than the relatively featureless cyclinder of the rest of the stage.

IMHO you nailed it.
To expand it, I believe that deployment of the legs moved down the Center of Pressure of the stage, zeroing its distance from Center of Gravity; it also increased total drag.
This means three correction had to be done in same time:
lateral acceleration due to the increased wind drag;
rotational acceleration  due to the gimbal of the engine (now excessive);
angle of the stage (being not vertical but slightly angled to compensate momentum due to wind).
Corrections were fast, but not enough.

[OxCartMark]

I don't know how this whole twitter thing works.  Really.  But when I saw the previously linked Carmack / Musk interchange last night my first thought was to give not so much credence to the propellant valve theory because somewhere shortly before or after that comment Carmack(?) was asking whether the final firing was being done with just one or more than one engine.  My take away at the time was that anyone asking that question knows less about this than nearly anyone here.  But this morning I see it mentioned that the bipropellent valve comment came from EMusk, not Carmack.  Is that correct?

[hrissan]

Good analogy on what happens when there is an unexpected delay in control loop...

Fast forward to 0:25, stupid youtube fails to embed start time correctly.

[Jarnis]

I don't know how this whole twitter thing works.  Really.  But when I saw the previously linked Carmack / Musk interchange last night my first thought was to give not so much credence to the propellant valve theory because somewhere shortly before or after that comment Carmack(?) was asking whether the final firing was being done with just one or more than one engine.  My take away at the time was that anyone asking that question knows less about this than nearly anyone here.  But this morning I see it mentioned that the bipropellent valve comment came from EMusk, not Carmack.  Is that correct?

Yes, it was a Musk tweet to Carmack.

[OxCartMark]

I need to watch the video a few more times to see if I can make out the moment of leg deploy, but I have to wonder if the late deploy, into a steady sea breeze, isn't what causes the pitch excursion due to the sudden increase in drag at the base of the vehicle relative to the wind. Those rapidly-telescoping legs and their bases have much greater area drag than the relatively featureless cyclinder of the rest of the stage.

IMHO you nailed it.
To expand it, I believe that deployment of the legs moved down the Center of Pressure of the stage, zeroing its distance from Center of Gravity; it also increased total drag.
This means three correction had to be done in same time:
lateral acceleration due to the increased wind drag;
rotational acceleration  due to the gimbal of the engine (now excessive);
angle of the stage (being not vertical but slightly angled to compensate momentum due to wind).
Corrections were fast, but not enough.

I thought I'd done a convincing job of debunking that theory already.  At least I convinced me.  Not sure if you didn't see it or didn't believe it.  In case you didn't see my dubunk, it goes like this-

In the video you can't see the legs, other than one which is slightly visible when its nearly on the deck.  This leads one to think that the legs didn't come out until the last moment which is incorrect.  The video compression has erased the legs from all but that one glimpse even though they are definitely out.  They can be seen to be mostly deployed  in the still picture that came out immediately before the vineo.  The stage is hundreds of feet up when they come out.  There's an 8 second difference between legs out and landing.  I don't think the last minute CP shift theory works with that much legs in the breeze time, does it?

~Changing subjects now~

There has been discussion of the stage rotating around its long axis during descent.  I'm not sure I can see it, maybe I see some rotation as it nears the deck. If there is substantial rotation then;
a) We're back (partially at least) into a problem that centerfuged the propellant away from the pickup ports and reduced thrust at the end of the first (legless ) attempt at at water landing.  Which was corrected on subsequent attempts by a doubling of the N2 thruster system capability, and later grid fins.  But if there was some centerfuging which reduced fuel availability (keep in mind that the tanks are ~99% empty) the result would be the inability to achieve 100% thrust, the same result that would come from a sticky propellant valve.
b) Put your structural engineering hats on for a second and let's think about leg failure.  Legs can fail by folding up but in that case the forces are all in line with the cylinder and leg and pivot points, and the cylinder would (assuming it doesn't have some ratchet or brake mechanism) absorb quite a lot of energy.  .or. Legs can fail by the introduction of side load.  It wouldn't take much side load to overload the legs because the two mounting points at the base of the leg are so close together relative to the length of the leg.  It wouldn't take much rotation or translation of the stage to exceed the side load limits of a leg.  I'll bet on this vs. folding up as the leg failure mode.

[MP99]

In this case, it was stiction, i.e. static friction. The valve didn't start moving when it was told to, and so the rest of the control system lagged behind. The fix could be as simple as commanding the valve to open, perhaps not even fully, a second earlier to overcome stiction and be ready to respond to input immediately when received. Of course it could also be more complicated than that, but from Elon's tone, it shouldn't be a huge deal.

Of course, then they run into the next thing to go wrong...

So is the valve sticking/stiction for this particular valve something that's manifested itself before? Otherwise, what bad time for it to suddenly/randomly make an appearance. Maybe it was the high intensity operation and demanding nature of this landing which caused the valve to underperform.

It's my guess that engine is gimballing far more than it ever would in flight. Possible factor?

Cheers, Martin

[cscott]

The fact that the tweet was deleted isn't necessarily terribly significant:

1) someone might have tapped Elon on the shoulder and whispered "ITAR".  Better safe than sorry (even though ICBMs don't need to land!).

2) Elon might have realized he might be prejudicing his engineer's investigation by talking too publicly about initial findings.  Employees tend to listen to the CEO, even if this was not meant as "CEO" conversation.

3) A second engineer might have said, "oh, but it might also be XYZ" and Elon decided he wasn't so certain after all.  (This is the option that will fuel discussion here on NSF until June.)

4) Elon switched to private email for his chat with Carmack, and cleaned up after himself.

5) Too many folks (including reporters) started flooding his inbox with questions after the stiction comment, and he decided he'd rather try to unsay it than have it on the record and appear in news stories.  (It's the landing attempt, not the successful CRS launch, which led most of the new stories I saw.)

After all, he was sure that Carmack received the tweet, so there was nothing to be gained by leaving it up.  And maybe he and Carmack then did a little deep dive into the valve mechanics in a non-public forum.  We just happened to overhear some private conversation... lucky us!

ps. Elsewhere on NSF somebody else noted that Elon is in the habit of deleting his twitter @- messages after they are received, which lends support to option 4.  How boring!  I'm sure we'll try to believe option 3 anyway.

[cambrianera]

@OxCartMark,
In the picture that surfaced before the video, legs are half deployed (that means picture was taken during deployment).
Deviation of rocket in the video happens right after that moment.
But this is my opinion, I've debunked nothing.
And clearly if you don't agree with me or with Herb, you are full entitled to that.

[Kim Keller]

For those commenting on late deploy of the legs, that event happens at Landing -6 seconds. The vine starts just after leg deploy.

[eriblo]

For those commenting on late deploy of the legs, that event happens at Landing -6 seconds. The vine starts just after leg deploy.

And full deployment takes at least 2-3 seconds from previous videos. It might be a little unsymmetrical and likely represent the biggest change of aerodynamics since the stage went subsonic. Personally I wouldn't rule out a contribution to the need for the questioned "late maneuvering" (which might or might not have had extra control loop lag).

[CapitalistOppressor]

So the valves have been nowhere near as extensively tested in the regime of small, fast, and frequent changes.

It could have been extensively tested in McGregor, on an engine test stand.

Could environmental factors be at play here?  I'm not an engineer or familiar with rocket engines, but the stage ascended into a near vacuum then reentered the atmosphere tail first at hypersonic velocities. 

Even if this valve isn't exposed to the outside environment, I'd still think that there would be rapid temperature changes and possibly other effects that are unique to the flight profile and which would not be simulated on a test stand.

[iamlucky13]

The fact that the tweet was deleted isn't necessarily terribly significant:

I agree, but it was still worth pointing out the tweet is gone.

Given Carmack's own deep interest in engine development and control, it definitely could have been part of a real discussion.

Also, several of your points actually are significant. #2 and #3 would mean the cause is not certain, which would not be a surprise just a couple hours after the engineers had their first look at the telemetry. That in turn would mean we shouldn't get too attached to the valve stiction comment, although some good discussion did come out of that.

In fact, I've worked a bit with industrial proportional valves, and had encountered the issues caused by stiction, but not to a critical degree that I had to resolve. So the dithering video was some new, interesting info to me.

[Kim Keller]

And full deployment takes at least 2-3 seconds from previous videos. It might be a little unsymmetrical and likely represent the biggest change of aerodynamics since the stage went subsonic. Personally I wouldn't rule out a contribution to the need for the questioned "late maneuvering" (which might or might not have had extra control loop lag).

I disagree with your deployment time. It's more like 1.5 - 2s.

[MP99]

I'm still open to bet on this... 5 years from now, if a SpaceX barge is in service, it will be able to broadcast live video. Want to take that bet?

How high was F9 when they called acquisition of signal at the barge?

A drone at that height above the barge could transmit to the Cape, and other parts of the coast are closer / would need less height.

Cheers, Martin

[edkyle99]

The Vine video reminds me of Pilot Induced Oscillation effects.  In this case, it's an autopilot.

 - Ed Kyle