May 2, 2004 notes
Jet vanes win
On the Tuesday before Space Access we tried hard to get a
perfect hover of the big vehicle in before we left, but we ran into a couple
problems. I replaced the A/D board, but
I had a jumper misconfigured, resulting in the thermocouple signals being
clamped off early, and fixing an error that I had introduced the previous week
while setting up the jet vane code path had resulted in the throttle positions
being off by about 13% (which was why the vehicle was almost hovering at 25%
throttle last week it was actually 38% throttle). The combination of these two factors along with the normal issues
resulted in us not getting all four engines warmed up properly for a liftoff.
This Tuesday, I get both of those problems corrected, but we
also finished up assembly and wiring of the jet vane test vehicle. We have a nice insulated box made of
phenolic / nomex honeycomb covered in fastblock insulation surrounding the
vanes and protecting the actuators and wiring.
We had some intermittent problems with the throttle valve stalling
partway open, so we didnt tank it up to a liftoff pressure, but we did hang it
from the lift and let it tilt itself side to side under its own control. I boosted the roll control gain a lot on the
second test to let it keep itself from twisting under the tether.
Russ finally realized that the valve stalling was just the
internal thermal cutout tripping due to the control system cycling it back and
forth so rapidly for 30 seconds or so. We bridge past this on all of our new valves as part of our
prepping procedure, but the ½ ball valves we were using on this vehicle were
almost two years old, from before we discovered that issue. After we fixed them, there were no
On Friday night, my laptop died. It had given some signs of ill health earlier in the week, blue
screening once on Tuesday and somehow killing a joystick attached to it on
Thursday, but on Friday the hard drive would only make sick sounds and not
boot. Luckily, I had just backed
everything up before going to Space Access, so I only lost a few evenings of
work. I lugged a desktop system in to
the shop on Saturday, but I am probably going to buy a ruggedized laptop with a
transreflective display for Armadillo soon.
We have had instances where a normal display has been very difficult to
read, so that will be a good benefit as well.
Unfortunately, most of the ruggedized laptops seem to by only 1024x768
resolution, and I have been spoiled by the 1600x1200 resolution on my previous
Even though we knew why the valve stalling had happened, it
was a strong enough reminder about potential failures that we decided to plumb
in a master cutoff valve on the jet vane vehicle. Stacking a second valve didnt quite fit, so we had to do a funny
angle weld on top of the engine to get everything in. On the big vehicle, the master cutoff computer board and battery were
in a separate box from the main electronics, connected by a cable. We had already had a couple instances where
we either forgot to connect the cable or forgot to turn the master cutoff on, requiring
us to take the hatch back off and fix it, so we had extra incentive to go ahead
and integrate it into the main electronics board. Previously we used separate batteries for the actuators (valve
drives, ignition, etc) and the sensors / computers, but we have had some issues
that have encouraged us to at least unify the grounds. The computers and sensors all take power
from isolated DC/DC converters, so we decided to run the actuators straight
from the main battery. This seems to
work fine. I now use the smaller
battery for the master cutoff, and I mounted the master cutoff board in one of
the few remaining flat spaces on the board.
This was a good simplification, and we now have a master cutoff in all
For flight testing we again hung the vehicle under the lift,
but since it wasnt going to be able to dip the lift down very far, we were
able to just rest it on our old 12 heavy foam cubes for stabilization before
liftoff. They blew out nicely on
throttle up. The engine is leaking a
bit around the spark plug hole, but not too bad.
The first hop tilted over quickly, so I increased the tilt control
gains quite a bit.
The second hop lifted off and hung at an angle, picking up
horizontal velocity, so I cut it off quickly.
I probably could have steered it back, but it was easier to just kill
it. On reviewing the video and
telemetry, we found that for the time that it was in the air, it was flying
This is the behavior we would expect from either an offset
center of gravity or vanes that werent calibrated straight up. We knew the mounting of the electronics had
biased the weight forward, so we clamped ten pounds of weight on the back side
for the next test flight. This time it
flew basically straight up, still with the extreme smoothness, but we lost GPS
lock almost immediately after liftoff.
We went through several tests trying to figure out what was
going on. I could fly it on manual throttle
just fine, but the Ashtech G12-HDMA was losing lock within a quarter second or
so of liftoff. Communication with the
GPS continued without a hitch, but the values would come in all the same for a
ten second or so stretch. We tried
mounting the amplified GPS antenna on some soft foam and even completely
enclosing it in foam for acoustic protection, with no change. Russ had an idea that sounded really logical
that it wasnt liftoff that caused the problem, but instead it was the
condition of high chamber pressure on throttle up that caused the spark plug to
misfire, generating RF noise. I changed
the code so I only had the spark going during warmup and not during liftoff,
but it didnt change anything. I have a
new GPS antenna arriving soon that we will test with on Tuesday.
We decided to go ahead and do a long run with me manually
controlling the throttle and steering.
I had it throttled up for 19 seconds, but I wasnt perfectly centered
under the lift when I started descending, so the last couple seconds pulled
taut on the tether. Still, it was at
least a 16 second perfectly controlled flight.
If it hadnt been on a tether, this test would have required a federal
launch license complete with environmental assessment, which rather clearly
shows how silly the burn time limit is.
It looks like AST is going to grant us a burn time waiver, but only for tests
at SWRS (southwest regional spaceport), which is a two-day trip away.
A couple things noticed on the telemetry:
It looks like it could use a little bit more gain on the
roll control, but the tipping angles are just fabulously smooth. The Y angle displays a constant offset
because the CG still isnt completely centered, but this will be automatically
taken care of when I start tracking horizontal velocity from the GPS, or I can
add a windowed integration term to the control system.
I throttled up with the engines at 750 C (I warmed them to
800C, but I had to wait while someone ran out to take the lens cap off the on-board
camera), and it didnt dip at all as the throttle opened. During the run, it slowly crept up to almost
1000C, so it was running with excellent efficiency.
Interestingly, the GPS regained lock for a second and a half
midway through the flight, then lost it again for ten seconds.
Something else very interesting on all the tests of this
vehicle is that the engine warmed up completely predictably on every single run
(nine so far), without any need for a large slug of propellant to get it
started. There are two factors that
probably combine to do this: this
engine has the compressed ring catalyst in the hot section instead of a loose
pack, and the main throttle valve has a slight leak, which has the engine somewhat
warm before I even throttle it up. We
usually hear a pop at some point during pressurization, which means that the
tiny trickle of propellant is actually igniting even before I start the spark
plug due to the methanol breaking down into hydrogen gas, which can ignite on
catalyst at room temperature with the released oxygen. If that is the key factor, we can add tiny
calibrated leak orifices into the engine, but we may also have to watch out for
tiny flows causing burning inside our cold catalyst, which tends to prune them
up after a lot of runs.
Overall, I am very proud of this work, we went from concept
to highly successful flight test in just over one month when it wasnt even our
top priority. The behavior of this
system, both in terms of startup complexity, plumbing / wiring complexity, and
flight control responsiveness are so superior to the differential throttling or
attitude engine systems we have used before that we feel it is a clear winner. When a jet vane moves, you get torque on the
vehicle instantly, as opposed to throttling or starting an engine, which
requires a fairly long liquid column to be accelerated, then the gas volume of
the engine to change pressure before you get vehicle torque.
The two classic objections to jet vanes are that they erode,
and they sap a few percent of the vehicle performance due to drag. With our low temperature engines, there is
no erosion whatsoever. We are willing
to pay a bit of performance for this vehicle layout, and we mitigate it a bit
by being able to run the throttle 100% open, instead of having to stop a bit
short to allow differential throttling, and the straight shot plumbing we will
get under the big tanks.
A downside that I had been considering was that we couldnt
make it redundant like we could the quads of differentially throttled engines,
but I realized that wasnt actually the case.
You could put two engines with their own jet vanes next to each other,
and if one was shut down, the other could still fly the vehicle, just at an
angle of attack that put the offset thrust through the center of gravity. It really isnt any different than engine
out with gimbaled motors, except you can get by with only two instead of
three. Jet vane roll control with an
offset engine would still work, but it would have some cross axis coupling into
the other axis due to the side forces not being completely balanced.
There are several more things to be done with the test
vehicle: get the GPS working reliably,
start using a differential GPS base station, calculate vehicle orientation in
GPS space so it can do accurate automatic horizontal position hold, and start
doing accelerating boosted hops at our test site. We are probably also going to stick the laser altimeter back on
it as a backup for the GPS.
We thought long and hard about it, and reached the
conclusion that continuing the flight tests with the big vehicle in
differential throttled mode would just be a waste of time. We are going to cram hard on converting it
over to using one of the 12 engines in jet-vane mode. We have to build a large vane board, modify
the tank bulkhead, build longer landing support legs, modify a pair of 2
valves for cutoff / throttle joining, rewire everything, and mount the
engine. We hashed out how everything is
going to go, and we think that we can have it up in the air in two to three
weeks. From there, we just need to
build a Really Big Engine to make the vehicle space shot capable.
We need the equivalent of a 24 diameter engine, making
16,000 to 20,000 lbf. To make a single man to 100 km vehicle with the 850
gallon tank, or two of them on the 1600 gallon tank for a three person X-Prize
vehicle. We are setting up to do some
tests with liquid catalyst injection again as a possible option for the really
big motors, avoiding the expense and lead times on all the engine catalysts. We had lousy results when we tried that
before, but now that we know the importance of flameholders for these engines,
I think we can make it work. It is
probably still not a good idea, but we are going to see how it goes.