August 27 and 31 meeting notes
Armadillo Labor Day cookout: http://media.armadilloaerospace.com/2002_08_31/hotdog.jpg
200 Isp, 45 second radiative burn
We tested our fourth generation kerosene injector on
Tuesday. Basically, it has an even tighter venturi, down to 0.6"
(with a 0.5" throat), and instead of the fuel coming in as a high velocity
jet from the side of the venturi, a stainless pipe comes in from the outside on
the catalyst pack side of the venturi, and it has a pretty sizable hole in the
bottom, before a plugged end. The idea is to have very low velocity fuel
stream straight down the middle of the very high velocity (nearly sonic)
peroxide decomposition, much like an SSME lox post injector in reverse. The way we bring the fuel pipe in is pretty
neat we drilled out a swagelok tube-to-pipe fitting so the tube can slide
completely through the fitting, making the ferrules into a bulkhead seal. This technique opens up a few other
interesting design possibilities for future work.
We put the radiative chamber together with the high
temperature bolts, the stainless clamp ring, and the silica gasket material
(not the gas-filled O-rings, which hadnt arrived yet). We also wet sanded the
silicide coated flange, and used a torque wrench to evenly tighten the flange
to 6 ft-lb on each of the #10 socket head cap screws. All of the runs sealed
absolutely perfectly. When we finally
got around to checking, 6 ft-lb is a lot higher than cap screws that size
should be torqued, even in a high end alloy, but it did the job. For 18-8 stainless #10, 2 ft-lb is
recommended, and about 3 ft-lb for 316 stainless.
We started off very rich with 0.080 peroxide / 0.050
kerosene (but not as rich as we were running the regen engine), and immediately
on lighting, it was obvious things were working a lot better, because the
engine started glowing much more rapidly, and much higher up. There was
still a visible bias opposite the fuel inlet, probably because the flow had
enough momentum that it was not "turned" completely axially when it
came out of the drilled hole.
Our oxidizer-only Isp was 246, up from a previous best of
192! We shrank the fuel jet a good chunk to 0.045, and the oxidizer-Isp
increased a bit more. We shrank it some more, and it finally nosed over a
bit, which would still be at increasing "true Isp".
Thrust was 45 lbf +/- 5% at 250 psi feed pressure.
We will have to get our sight glass hooked back up on the
new tank so we can measure the kerosene flow, but this was with an 0.040"
kerosene jet, while a 0.060" jet resulted in a measured O:F of 2.2:1 on
Saturday, so that would put us around 5:1 O:F.
On Saturday we double checked the load cell calibration and found it to
be a couple percent optimistic, but that still means we have a true Isp right
at 200 seconds, which I am quite happy with.
Long video of the entire burn: http://media.armadilloaerospace.com/2002_08_31/burnedTZM.mpg
We loaded up four liters of peroxide and let it run.
The burn was nice and monotonous, but after 45 seconds of hot fire, the plume
started deflecting slightly, and some sparks started coming off the
nozzle. I killed the kerosene, and let the rest run out as monoprop.
We burned the silicide coating off at the nozzle, which
allowed the TZM to vaporize a hole. We have another TZM chamber already
machined, but Im not sure that I want to spend another $3000 for the silicide
coating. We obviously have to tone down
the heat transfer with some combination of richer fuel mixture, lower chamber
pressure, better film cooling, and thermal barrier coatings if we want to run
radiatively cooled. At the moment, our
regenerative designs are looking good (and much cheaper!), so we are leaning
that direction. Being able to see the
combustion distribution by the glow of the chamber has been quite useful, so we
might make up a batch of disposable stainless steel chambers that we can just
let glow red for a few seconds.
We may be near the theoretical max Isp for this pressure
(220 at a leaner mixture), but we wanted to get rid of the uneven combustion,
so we modified the fuel injector tube to actually take a 90 degree turn (cut,
bend, weld), instead of having a hole drilled in the side. This gives
straighter, and even lower velocity fuel injection, which should improve the
On Saturday, we installed the modified injector on the regen
engine for tests. We used water cooling
instead of peroxide cooling, because we arent expecting this small of an
engine to be able to run steady state with the improved combustion temperatures
against the reduced peroxide flow.
We had several frustrating runs with leaks around the
injector, leading us to believe that the injector was left somewhat warped by
the long radiative runs. This has been
a chronic problem for us, so we are going to go ahead and make an injector out
of stainless that should hold up better.
We were able to get good enough data to learn something
interesting the new fuel pipe with the 90 degree bend was performing worse
than the drilled pipe, only slightly better than the side injector. We believe this is due to it sticking down
into the venturi, so there is less opportunity for high shear mixing before
everything expends out into the main chamber.
We are going to test this theory by making the next injector tall enough
that the bent pipe exits well above the plane of the venturi.
We have been buying parts for a 1000lbf regen biprop, which
we should be testing within a month.
Altimeter Auto Throttle
We did a normal five gallon test hop of the lander to make
sure everything was still working properly after the skid on the ground after
losing the computer last time. We had
replaced some components, and changed from AN jets between two aluminum
fittings to custom drilled jets in a single pipe fitting to make the engine
mounts more rigid. Everything worked
We prepared to try the auto-throttle again. We loaded a somewhat lighter load of
peroxide, because the last time we tried this, June
8 Notes, we had a pretty bad crash due to improperly simulating the valve
behavior. I was prepared to kill the
auto throttle immediately on this test.
After warming the engines, I briefly clicked the altitude
hat up, which would cause it to set a target altitude half a meter above the
current altitude. The vehicle leapt off
the ground, and I rapidly killed the auto throttle, but on reviewing the video
and telemetry, it had behaved exactly the way it was supposed to, bobbing a
couple times in hover before I had killed the throttle.
I lifted it up again with the auto-throttle, and it did
indeed hover just where it was supposed to.
However, it started rolling fairly rapidly, so I told the auto throttle
to land, which it did properly. After
looking at the data, it appears that the roll fiber optic gyro was going
completely nuts. We power cycled the
entire system, which seemed to clear up the problem.
We loaded up for another run, which started out perfectly,
with it sitting at a nice hover, and me flying it side to side a bit, but the
roll gyro went away again mid-flight, starting the fast roll again, and causing
it to tip over as I set it down. We
broke another hose, but everything else was fine.
On investigating the next day, it appears we may have a
thermal issue. The logged IMU
temperatures climbed from 88 F to 112 F over the course of the tests. This is still well within the rated limits,
but the roll axis would be the one mounted on the bottom, which is bolted to
the plastic case, which may well be significantly hotter than the location of
the sensor. Crossbow recommends
mounting to an aluminum plate.
With the auto-throttle working well, we should be able to do
a perfect DC-X style flight path out at the 100 acre site now. We will also be able to develop high speed
auto-landing, which will allow the computer to save the vehicle automatically
in the event of telemetry failure, or any failure on the remote pilot laptop
system (like the joystick failure that caused the splat crash Dec
22, 2001 Notes ). It will also enable soft, upright landing
after a parachute cutaway.
The laser altimeter we are using, a Laser Atlanta Advantage,
is unfortunately a bit of a pain for our application. We converted it from a battery pack to using our main system
power, but it still requires us to manually press a membrane switch to turn it
on, and the continuous reading mode is a commanded toggle that requires us to
start the altimeter and the flight control program in a specific order. The housing shape, like a police radar gun,
is also annoying from a vehicle integration standpoint. I also get a few bogus values from it occasionally. I may wind up buying a more expensive one.