July 14, 2007 notes
Ive been crazy busy at Id with WWDC and E3 last month, and
as a forewarning, the next six weeks looks equally busy, so updates might be
We got in a bit of trouble last month. After we delivered all of our telemetry data from
the permitted free flight in Oklahoma
to AST for analysis, they pointed out a couple anomalies: our initial battery
voltage was at a level that should have triggered a no-go, and the vehicle had
rolled to a point that should have triggered an abort. These were due to the fact that we had
replaced the lead-acid batteries with lithium-ion batteries to save weight, and
relaxed the roll control dead band to reduce ullage
gas depletion. I had written about these
in the last couple updates, and we had tested them with tethered tests, but we
had not updated our permit with the changes.
AST agreed that we had not endangered anyone, but we were in violation
of our permit rules. We have agreed on adopting
new procedures to ensure that engineering changes get propagated to the AST
paperwork, but we have dropped a few rungs in ASTs eyes.
We set out to test a modified injector with reduced film
cooling to give more equal propellant utilization and higher Isp. This should have given us 220+ second flight capability while carrying the 25 kg X-Prize Cup payload, but
immediately after takeoff it was clear from looking at the exhaust plume that
something wasnt right. After a
teardown, we found that the metal Helicoflex seal had
a raised, pinched area in it, and that had caused a crack in the graphite
chamber when it was pressed into place.
This was almost certainly caused by careless handling of the seal on a
work table while we were swapping injectors.
In the future, we are probably going to just use new seals on each
engine. They are fairly expensive at
about $130 each, but we shouldnt be taking the engines apart much more.
We made the decision to not bother testing the hotter engine
before XPC, and just go with the flight-tested pattern. We have dropped 20 pounds off the vehicle
since we did the 192 second flight, and that wasnt to complete depletion, so
we have 200+ second capability as-is, which should be plenty for the 180 second
level two flights. My wife scolds me
sometimes about not just sticking with a working solution, and she is usually
I finally worked out everything necessary to use the Septentrio attitude sensing GPS system with our electronics
box. The antenna / receiver combination
is much more sensitive to interference than our previous GPS system, and we
werent able to hold a fixed integer ambiguity lock for the attitude solution
until I completely removed power from the old GPS board that was still inside
the box. I was a little surprised to
find that level of interference between two receivers. We later found that our wireless video
systems also cause interference with the GPS.
We are going to have to make some corner reflectors for the video
It took me an appallingly long time to get the 3-axis attitude
values from the GPS into the correct reference frame for the existing flight
control software. It was frustrating
having to leave the box and antennas propped up at different angles out in the
middle of the parking lot while working on the software inside, but the worst
part was that I felt that I had gone through every possible orientations of the
quaternion output before I finally just switched over to the Euler angle output
and eventually got it working. Im
supposed to be good at all that 3D math. Sigh.
We brought out Texel,
our second Quad vehicle, and propagated all the updates we have made to Pixel
over. We are in the process of making
sure that our team skill sets are fully redundant, so I started training Russ
on running the laptop flight control software to fly the vehicles (Joseph is
also training Phil on driving and operating the crane truck). We went out for a tethered test, but we
managed to get a bit of a bang on startup, which cracked the graphite chamber.
There were three factors that were different from our last
dozen or so successful flights: We had
reduced the starting pressure from 300 psi to 275 psi, because during the free-flight in Oklahoma we noticed
that the engine throttled almost all the way down to the minimum throttle level
when transitioning from a hover to a decent at higher altitude. This caused the engine to seem like it almost
shut down, so we were trying to bring the throttle position up a bit by
operating at a lower pressure. This was
also the first time Texel
had alcohol in it since last October, so the line to the igniter was completely
dry. Finally, the lox pre-chill before
engine start has been done manually, and Russ did it a little shorter than I
This was the last of our pristine graphite chambers, and
unfortunately we had a miscommunication with Cessaroni
Aerospace about pre-ordering more blanks to have on hand, so we are stuck with
a bit of a lead time. We are probably
still ten days out from getting a good sized shipment in.
To address each of the issues we believe led to the startup
bang (which wasnt much as hard starts go, not damaging any metal, but the
graphite is fragile), we are now doing the following:
We are back to 300 psi minimum
pressure. The engine cough doesnt seem
to hurt anything, so we are going to stick with what has been demonstrated to
work. We will eventually have to revisit
this for in-flight restarts, which will always be at a lower pressure. We may need to have a different idle throttle
position based on current tank pressure, but we will work that out post-XPC. We are going to bleed fuel into the igniter
early in our flight prep process to make sure it is going to instantly get fuel
during the start sequence. I have made
the lox pre-chill part of the automated start sequence, so it will be exactly
the same duration on every start.
While we dont have any perfect chambers, we do still have
one with an heavily eroded throat that resulted from a
pinched o-ring on an earlier engine. We
certainly werent going to try and do back-to-back 180+ second flights with it,
so we decided to finish up the first modular rocket and try and get it in the
Ready to fly, the module weighs 455 pounds, over 100 pounds
of which is in the legs. If you also
take out the electronics box, then the raw tanks, plumbing and engines that you
would replicate in a larger configuration are about 320 pounds. I am pretty happy with how the engineering
turned out with the modules, essentially everything is right at the base of the
vehicle, and you could basically move everything to a different tank set with
almost no changes. Each module can fly
as an independent rocket right now, but if pure differential throttling with no
gimbal movement works out on the four module system,
our next generation of modules will have much tighter and lighter
packaging. I am also seriously
considering a module arrangement with a fixed engine beside the tanks, so we
could let the bottom of the spheres just bump into the ground with rubber
pads. Landing gear weighs more than you
think it should, and keeping the height low helps landing stability and
transportation. With a fixed engine
design and chem.-etched tanks, we should be able to get to a mass ratio of 5
with the modules.
We headed out to our test site, once again getting a bit
lucky with the weather here. I noticed I
had one of my three year old sons toy robots in the back of my SUV, so I
decided to tie-wrap it on for the rockets maiden voyage. Russ can ride a little bit later. J
Without having to worry about the propellant balance
problems that the Quads have, we were able to do four short flights with a
single propellant load, just re-pressurizing in between to make up for the long
engine purges after shutdowns.
On the first hop, it sort of fell off the stands and went
into a severe roll. It seemed to catch
it and roll back, but drastically overshot.
I suspected that my roll gain calculation might be wrong for very high
rates, allowing the overshoot.
On the second hop, we moved to a smoother patch of concrete
that wasnt chewed up from previous flights, and stuck a shim under the short
leg to make the vehicle more stable on the stands. It took off straighter, but immediately went
into a strong roll. Clearly we had the
roll thrusters backwards, even though we checked that at least three times,
comparing against the quad.
We switched the roll thrusters for the third flight, and it
took off fine. The reason the roll
seemed to correct and overshoot on the previous flights was that when it passed
180 degrees of roll, it was interpreted as -180, which caused it to change
firing directions for another 180 degrees.
It still flew off to the side a bit on liftoff, but that is certainly
due to the heavily and asymmetrically eroded chamber nozzle we are using. Stability was great, and I drove the vehicle
around under the tether a bit.
We did a ground liftoff and landing for the fourth flight. Everything went fine. We need to get a softer grade of rubber for
the landing pads, because even with a lot of holes drilled in them, the total
surface area per pound is higher than on the Quads, and it hits fairly
jarringly. I have been logging peak accelerometer
values for our ground landing flights for a while now, and I think I am finally
almost comfortable with using that as a ground-contact signal for automated
engine shutoff to reduce the landing bounce and worries about the control
inversion when a leg is on the ground.
If we had brought more helium, we would have gone ahead and
flown a 90+ second flight today. We have
a few things to improve before the next module flight, and I want to move future
work with the 3-axis GPS over to the module so the two Quads stay identical
with the flight proven configuration, but I was thrilled with how well
everything came together. The new start
sequence gave ultra-smooth pressure rises on all four flights. I tend to think that the longer, automated
lox pre-chill was the main improvement. The
higher thrust to weight ratio of the vehicle is very apparent it really leaps
into the air on throttle up. I still
think vertical drag racing with two of these will be really damn cool.
Now that we have proven that the module works as designed,
we are going to build up the remaining four modules that we have parts on hand
for. James is going to work almost full
time on the fabrication work, so I wouldnt be surprised if we have the tank
sets bolted together next month. All the
plumbing, wiring, and testing will take more time, but Im pretty sure we will
have a four module system done for display at XPC. A tethered test at XPC might be possible, but
winning the Lunar Lander Challenges is the main priority. Again, it is a little unfortunate that our
development scheduling gets hampered by the fixed scheduling of the LLC at the
The four module vehicle will probably be flying the Space
Diving mission next year. If we dont
wreck it in testing high altitude flight, we are considering doing a staging
demo, flying the fifth module off the four module cluster, which would be a
solidly 100km+ vehicle. The commercial manned
100km vehicle should be a cluster of six to nine improved modules (higher mass
ratio and aspect ratio) in a single stage for full redundancy, with a nice
cabin on top.
Matt is finally out from under his own crushing work load,
so we have some pictures this month:
We are consolidating all the propellant loading controls in
one place. We will be making a metal
Here is the current engine design. The tapped holes on the bottom allow a
pressure test plate to be bolted on to check for post-valve leaks. It might also be useful for a nozzle
extension at some point.
Im sure most people are appalled at us using rubber bumpers
for landing gear, but I rather like them.
Both of our electronics boxes are converted over to the
lithium batteries and honeycomb box walls.
Everything on the bottom is just stuck down with RTV instead of strap
clamps. 30 pounds lighter that the
The roll thruster mounts were just the cut off pieces from the
quads, after we moved them inboard. We
are going to rebuild these to hang the valves down more vertically and use 90
degree fittings for the nozzles.