September 3, 2006 notes
Tank Insulation
We settled on using Pyrogel
insulating blankets from Aspen Aerogels to insulate
our tanks: http://www.aerogel.com/products/pdf/Pyrogel_6250_DS.pdf#search=%22pyrogel%20insulation%22
After Phil figured out the right pattern to make the gores
for the spherical tanks, it went very well.
We used a spray adhesive to attach it, and we tested all the combinations
of dusty side / non-dusty side and painted / non-painted for best
adhesion. Surprisingly, putting the
adhesive on the dusty side worked best.
The material still does shed some dust in the wind, but it is reasonably
rugged, more so than the fastblock insulation we were
previously using, and it is only about $5 / square foot, which is a tenth the
cost of the fastblock. It also sheds water fairly well.
I was a bit surprised at how much of a difference insulating
the tank made on our boiloff rates. I had been presuming that much of the boiloff was due to heat conduction from the rest of the 90
pound tanks that don’t get cooled that well during filling, but it turns out
that the entire tanks get cooled a lot better with insulation. We used to have 10 psi
in the tanks after filling, even with the vents open, but now it is just 3 psi and soon drops to 1 psi. We also insulated our test stand tank.
http://media.armadilloaerospace.com/2006_08_12/insulating.jpg
Engine Tests
We did some noise measurements to see just how disturbing we
are to our neighbors. In our parking
lot, noise is over 100 db, but it drops to 90 at the end of the building, and
is only 80 at the main street. Traffic
noise is 70, and this is an industrial neighborhood, so that really isn’t too
bad. Testing below ground level in our
receiving well with the water below it is certainly helping a lot, because our
old vertical testing was louder than this with significantly smaller engines. We need to check the noise levels for the
vehicle flight tests, which will be worse due to the exposed plume.
The element drilling of the stainless steel injector went
better than I expected. I botched the
very first element, which we welded over, but all the rest of the 38 unlike
impinging elements were milled and reamed without a problem. I had problems with the 1/32” film cooling
holes, which are about a 6:1 aspect ratio with this setup.
The elements are 45 degree fuel / 0 degree lox unlike
impinging. 38 unlike impinging element
pairs, 19 film cooling holes.
http://media.armadilloaerospace.com/2006_08_12/stainlessParts.jpg
http://media.armadilloaerospace.com/2006_08_12/stainlessBottom.jpg
http://media.armadilloaerospace.com/2006_08_12/stainlessElements.jpg
We had a rather odd set of tests with this engine. The first run had poor performance, slightly
worse than the last aluminum injector.
The second run had fantastic performance, 50% better than the first run,
even though the only difference was more propellant loaded to make the run last
longer. It wasn’t just a data collection
anomaly, because the video clearly showed no visible film cooling in the plume
because it was superheated much more. We
did more runs at various pressure and tried to
replicate the high performance run, but it never happened again.
The stainless injector suffered some burning right at the
igniter exit hole, but it only slightly yellowed elsewhere. The burning at the igniter hole might have
something to do with the high combustion efficiency, perhaps making a continuous
ignition torch during the odd run.
I was very disappointed that we didn’t get a consistent high
performance with this engine. We had
always rapidly melted the aluminum unlike-impinging injectors, so I had hoped
that the Isp would be much
higher if the stainless kept it from melting.
One of my last theories about the injector performance was
that the unlike fans were hitting each other very early because of the small
distance between holes, and this essentially just turned the pattern into a
single converging cone with less surface area than if the individual fans had
more room. To test this, I made another aluminum
injector with the exact same 38 unlike impinging element pairs / 19 film
cooling holes as the last couple injectors, but with the following changes:
Tapered manifold inserts to keep the propellant velocity
equal and higher around the manifold to increase cooling.
Elements twisted so the fans don’t intersect.
Flush drilled 50 degree impingement, 30 degrees from fuel,
20 from lox, so the resulting fans still won’t spread out towards the chamber
walls, but also won’t converge towards the center as fast as the 45 / 0 angles
we were using.
http://media.armadilloaerospace.com/2006_08_31/twistedParts.jpg
http://media.armadilloaerospace.com/2006_08_31/twistedElements.jpg
Note the deeper film cooling cutout to make the holes lower
aspect and easier to drill.
http://media.armadilloaerospace.com/2006_08_31/twistedSpray.jpg
http://media.armadilloaerospace.com/2006_08_31/twistedSprayFlash.jpg
In testing this, we saw a short burst of very high
performance from the engine, then it got ratty. I believed this to be a sign that the lox was
boiling in the manifold, due to the tapered manifolds drawing more heat out and
/ or the combustion being much better. I
added a set of tiny showerhead holes at the inner border of the lox manifold to
hopefully create a gas cooling layer.
We had several very high performance runs with this
configuration.
A three-camera shot of a long, high performance engine burn:
http://media.armadilloaerospace.com/2006_08_31/twistedRun.mpg
We finally made the inside of the graphite chamber glow red
hot, even with all the film cooling. The
outside got hot enough that the radiative heating
caused all the phenolic tubes insulating the tie rods
to blacken and crack above the nozzle, making a rather disturbing
snap-crackle-pop sound after the run.
The bolts needed to be re-torqued
after the run. We don’t know for sure if
the tie-rods actually stretched due to the chamber thermally expanding (I
really don’t think they got very hot with the phenolic
tubes over them), or if the bottom clamp plate relaxed a bit more. I’m leaning towards the clamp plate being the
problem. Cesaroni
Aerospace is going to modify the design for some of our new chambers so we can
grab the chamber from a flange at the top, leaving most of the engine hanging
free below. This will save a decent
amount of weight, and avoid most of the heat concerns.
When we increased the feed pressure to 400 psi, the injector burned through around the middle of the
lox manifold ring. I made a new injector
that had twice the film cooling holes to get the chamber temperature back down,
and I staggered the lox cooling holes with half of them on the middle radius
where it burned through previously. For
some reason, I was having trouble drilling the holes, and I wound up adding a
big countersink on them, thinking that might be a positive thing to cause the
lox to spray out and vaporize faster.
http://media.armadilloaerospace.com/2006_08_26/staggerHoles.jpg
The non-countersunk holes were machining errors,
they don’t go all the way through.
Note the doubling of the film cooling holes.
After doing some flight tests with this engine, we saw that
there was erosion on the face:
http://media.armadilloaerospace.com/2006_08_26/erosion.jpg
The countersunk holes are a mistake, they act as flameholders. We are
trying an experiment by welding over all this mess, facing it off, and
re-drilling just a single circle of straight holes at the outer band. I didn’t have any trouble drilling the
straight holes this time, but we haven’t tested the injector yet.
I had already modified the stainless injector with this same
pattern before we found the erosion, but it doesn’t seem to be burning it yet.
New Electronics
We have completed the new electronics box and transitioned
the current vehicle over to it, and the second box is almost complete. The old box got stripped of its GPS and IMU
and dedicated to the test stand. This is
the first time we have ever had two complete sets of electronics at one time.
The final push to retire the old electronics was that we
needed more driver channels to independently control each roll thruster to
allow us to move propellant from side to side for dynamic balancing. The old box required us to tie together pairs
of roll thrusters and purge solenoids, due to lack of
channels (we burned one pair of drivers with a short). With the new box we also get current sensors
on the drivers, more analog channels available, watchdog shutoff of up to four drivers
(so we can do a four engine differentially throttled vehicle next), and what
should be a higher reliability construction.
One of the three PC boards we had assembled at a board shop
had a significant error – the identical looking 5v and 12v DC-DC converters
were swapped. It is a very good thing
that Russ checked that before we put an expensive computer or GPS on the
system. The power and ground planes are
so generous on this board that Russ wasn’t able to unsolder the parts with his
equipment, so we have to send it back to the assembler to be fixed. They did turn it around in a single day,
which was appreciated.
We planned on moving to the new Thales
DG14 GPS boards, but despite what they claim, they don’t seem to be 100% software
compatible with the G12 boards that we have been using, so we have an older G12
in there until I can figure out why my init sequence isn’t working.
I was just looking for some manuals online, and I found that
Thales Navigation was just sold to Magellan a couple
days ago. This particular board family
started out with Ashtech, was sold to Thales, and has now been sold to Magellan, just over the
years we have been using them. Changing
to a different model wouldn’t be a large hardship, but I hope we don’t get
orphaned.
http://media.armadilloaerospace.com/2006_08_26/newElectronics.jpg
http://media.armadilloaerospace.com/2006_08_26/newElectronics2.jpg
http://media.armadilloaerospace.com/2006_08_31/boxOutside.jpg
http://media.armadilloaerospace.com/2006_08_31/boxInside.jpg
Flight Tests
Disconnecting the paired tank ullages
helped significantly with the vehicle balance problem, and if we carefully
balance the weights before liftoff, we can have good flights. Now that we have sufficient driver channels,
we can cause propellant to transfer from tank to tank in flight by firing only
a single roll thruster instead of a pair, causing a drop in pressure in one
tank, which allows propellant to flow over from the opposing one. We are probably going to do a manual trim for
this, rather than a very low frequency automatic control loop.
With the tank ullages disconnected,
the lox absolutely MUST stay below the saturation point, because at saturation,
any change in the heat input to the tanks will result in different amounts of
boil off, and a couple psi difference
in the tanks leads to a large propellant transfer. Our insulation is pretty good, but we won’t want
to sit on the pad for a half hour after pressurizing. Because our tank pressure drops during
flight, there is the possibility that we could not be at saturation at liftoff,
but reach saturation during flight. That
would be bad.
Flight 5: August 12, 2006 hop1
Added tank insulation
Increasing oscillations, hit ABORT_TILT, then quickly hit
SHUTDOWN_TILT
http://media.armadilloaerospace.com/2006_08_12/hop5.mpg
Flight 6: August 12, 2006 hop2
http://media.armadilloaerospace.com/2006_08_12/hop6.mpg
Cut all the gains in half to make it less twitchy
Much smoother
Ran to propellant depletion
Flight 7: August 15, 2006 hop1
Discovered that one of our bubble levels doesn’t read right
upside down when centering the gimbal.
Increased gimbal range to 0.1 – 0.9
instead of 0.2 – 0.8 for a bit more control authority. The linear actuators don’t have limit
switches, so I don’t want to stall them at their limits in flight.
Removed purge during igniter startup to prevent an initial
transfer of propellant between tanks. We
had added this when we thought we had fed some lox into the fuel manifold
during a pre-chill while horizontal testing, but that turned out not to be the
problem, so removing this doesn’t hurt.
Nice flight, drifted a bit, manually shutdown as it was
starting to pull on the tethers.
http://media.armadilloaerospace.com/2006_08_15/hop7.mpg
Flight 8: August 26, 2006 hop8
When we moved to the new electronics box, our hacked-up
tether shield didn’t fit anymore, so we tried a new idea to keep the tethers
from catching on anything – we put a five foot piece of PVC tubing over the end
of the tether, so the vehicle would have to go at least five feet higher than
the bungee stretch before a tether could loop over one of the protruding parts
of the vehicle. This seems to work well
so far.
Used staggered cooling hole twisted injector.
We added our “payload” weights for the Lunar Lander
Challenge to the vehicle as barbell weights right at the vehicle feet. We might as well move our CG a little farther
away from the gimbal point…
Flew to propellant depletion.
Suffered injector erosion.
No video.
http://media.armadilloaerospace.com/2006_08_31/ignition.jpg
http://media.armadilloaerospace.com/2006_08_31/liftoff.jpg
http://media.armadilloaerospace.com/2006_08_31/flight.jpg
Flight 9: August 29, 2006 hop1
Stainless steel injector.
Added automatic propellant pumping code.
It drifted a lot on liftoff, and I didn’t like the way the
plume looked, so I manually aborted it.
Inspection didn’t reveal any problems.
The hard drop on the tether caused the engine to slip a bit
in the gimbal mount, resulting in an off-center
engine that needed to be fixed. We need
to tighten the set screws harder.
http://media.armadilloaerospace.com/2006_08_29/hop9.wmv
(all combined two-camera videos are
wmv files instead of mpeg due to aspect issues)
Flight 10: August 29, 2006 hop2
Drifted to tether limit, simultaneous ABORT_ROLL_CONTROL and
ABORT_TILT
A rolling liftoff pipe knocked over the second camera at the
end of this video, so we welded little bits of angle to the pipes so they don’t
roll as far.
http://media.armadilloaerospace.com/2006_08_29/hop10.wmv
Flight 11: August 31, 2006 hop1
Disabled automatic pumping code. It may have been doing the wrong thing on the
last two flights, causing increased drift.
Doubled position hold gains relative to
orientation hold gains.
Perfect 35 second flight!
http://media.armadilloaerospace.com/2006_08_31/hover35_dual.wmv
A view from just one camera for folks without windows media
player:
http://media.armadilloaerospace.com/2006_08_31/hover35.mpg
There is still some stable tilt in the system, so we are
going to try and trim that with manually commanded propellant pumping, because
I didn’t like what the automatic system was doing.
Flight 12: September 2, 2006 hop1
Packed up and headed for the Oklahoma Spaceport, hoping to
do a > 90 second flight.
Joseph now has a CDL with hazmat certification, so we can
transport large quantities of all our consumables wherever we need. We outfitted our current trailer with all the
placards, logs, and extinguishers required, but we are probably going to move
to using a rented crane truck in the future, which will allow us to move enough
consumables for multiple full duration flights.
There is definitely something to be said for having everything you need
on one truck, instead of relying on things to get delivered to your test site.
Unfortunately it rained the entire day. We tried to fly the vehicle a couple times
when it cut down to just sprinkling, but the GPS PDOP values never
cooperated. The flight computer
currently will not initiate a flight if the PDOP is over 275, and it will abort
and auto-land a flight if it jumps over 400 during the flight. We occasionally got a PDOP under 275, but it
never stayed there long. We may be able
to relax these numbers, but this is the first time we have ever seen it this
bad, so I am probably going to stay conservative. We probably shouldn’t be flying in the rain
anyway. We got the vehicle off the
stands twice, but it did a GPS abort within a couple seconds. We are probably also suffering a bit from
vibration at the higher thrust levels required for flying with a full
propellant load, so we are also going to add foam insulation to the new
electronics box. The upgraded GPS boards
are supposed to be more vibration resistant, so it will be nice to move to them
after I figure out what changed in the initialization sequence.
500 pounds of ethanol / 700 pounds of lox loaded
ABORT_GPS due to PDOP exceeding 400, steady auto-land
sequence.
Repeated twice before the lox reached saturation and the
liftoff attempts became immediate tip-overs.
http://media.armadilloaerospace.com/2006_09_02/hop12.mpg
http://media.armadilloaerospace.com/2006_09_02/traveling.jpg
http://media.armadilloaerospace.com/2006_09_02/setup.jpg
http://media.armadilloaerospace.com/2006_09_02/groupShot.jpg
http://media.armadilloaerospace.com/2006_09_02/lift.jpg
http://media.armadilloaerospace.com/2006_09_02/missionControl.jpg
http://media.armadilloaerospace.com/2006_09_02/ignition.jpg
We are sometimes getting some raw ethanol pushed out by the
purge at the end of the flight when the combination of gimbal
angle and tether swing is just right. If
the film cooling holes were at the extreme outside edge of the fuel manifold it
would always get pushed out very quickly, but they are inset enough that if
everything is hanging at an angle there can be some fuel retention on one side,
which gets blown out on the next swing from the tether.
Our shock absorbers are frozen up again. We have had corrosion problems with them from
sitting in puddles of water after the lox frost melts. We tried greasing them up, but it didn’t seem
to do the trick. We may need to
fabricate the side load assembly ourselves.
We have found that the Jefferson
piloted solenoid valves we use for roll thrusters and engine purges will
actually open briefly if you hit them, which is extremely disconcerting. Our pressurization hose tends to hang close
to the roll thrusters, and if it swings against them we will usually get a burp
from the thruster, which can disturb our balance. Again, I am looking forward to moving back to
a differentially throttled vehicle next.
Pretty much everything is ready to go on assembling the next
vehicle. James has finished welding all
four spheres, and all the parts are here.
We have often said that it would be cool to do a time lapse video of
assembling the entire thing, but in reality we wind up moving the vehicle
around to different parts of the shop during assembly too much.
In the NVIDIA-theme, our two vehicles for the X-Prize Cup are
named “Pixel” and “Texel”
(Pixel get an added bonus by being a Heinlein character).
http://media.armadilloaerospace.com/2006_08_26/texelParts.jpg
News
