Jun 26, 2001 Meeting Notes
In attendance:
John Carmack
Phil Eaton
Russ Blink
Neil Milburn
New Supplies:
100 lb load cell
10,000 lb load cell
another 1000 psi heavy duty pressure transducer
More –6 hose and hose ends
-10 aluminum hose ends
More Teflon O-rings for compressed gas bottles
Silicone O-ring cord stock creation kit
Stainless machine screw / nut / washer sets
Loctite
I thought I had found a good supplier of 70% semiconductor
grade (unstabilized) peroxide with the Kanto Corporation, but it turned out
they only had 35% in that grade. I am
still interested in seeing what we can do with cheap 70%, possibly with hybrids
or biprops.
We also looked at a video of Juan Lozano’s (http://www.tecaeromex.com/ ) peroxide distillation
setup last week. I still like just
getting our peroxide out of a drum, even if it is fairly expensive, but a team
on a lower budget should probably consider it.
I am buying a fairly sizable lot of peroxide from X-L so we
won’t risk running dry for a while.
Flying the manned vehicle is going to go through four or five gallons
per flight, which will chew through our supplies quickly.
We discussed possibly migrating away from the Speed Flex Teflon
hose and AN fittings to one of the hose styles that McMaster carries, so we
could reduce the number of suppliers I have to hit, and so we could just get
permanently assembled hose lengths instead of building them ourselves and
having half of them leak when bent.
I have requested rough quotes for some large items (4’
spherical tanks, 4’ by 8’ cabin, 4’ by 20’ tubing) from three filament winding
companies:
http://www.amalgacomposites.com
http://www.lincolncomposites.com
http://www.advancedcomposites.com
After the first manned vehicle is worked out, we are
probably going to do a narrow, high performance demonstrator, then build
something pretty sizable. I am thinking
that we would build the first cut with E-glass, work everything out, then build
version 2.0 out of carbon fiber.
Broken Load Cell Analysis
I don't think we banged the load cell last week, I think we
actually overpowered it with the equivalent of a monoprop "hard
start".
The first blast was LOUD, louder than when we were firing
the big motor at 90 pounds of thrust, so I would not be surprised if it made
well over 100 pounds of thrust for a very brief time, which would be enough to
wreck the load cell.
Under the right conditions at startup it is possible to
generate much higher chamber pressures than during steady state operation,
because without any chamber pressure, several times as much peroxide will flow
as during steady state operation.
Normally this just results in the initial cloudy burst that
comes from the engines for a quarter second or so, but if there is sufficient
catalyst to operate on the much larger mass of peroxide, it may all go up as if
it were in a much larger engine.
This is potentially dangerous when the motors get closer to
the optimized weights, although it is somewhat balanced by the fact that the
motor casing will still be at its room temperature strength instead of
operating temperature strength.
I believe three factors contributed:
The uncompressed catalyst allowed a lot more to flow deeper
into the pack than with our older designs. We had the compressed foam
spreader at the top, so we were still getting good distribution, rather
than just tunneling straight through. If the pack was compressed, it
would have provided frictional back pressure to prevent too much peroxide from
flowing in even without chamber pressure.
We had over twice as much catalyst as we needed, which
provided a large volume to hold the peroxide as it flowed in, and it could work
on decomposing a lot more at the same time.
We had just done a water run, so the pack was wet and cold,
giving the peroxide time to flow all the way through the pack before reaction
started.
With the low back pressure catalyst packs, 450 psi tank
pressure, and the -6 plumbing, we could easily have pushed well over 20 ml of
peroxide into the engine during the warm up pulse.
As we moved to larger plumbing, I had increased the warm up
pulse to 200 msec from 100 msec, because 100 msec times were sometimes barely
long enough to get anything from the tank to the engine. This was
dependent on the exact arrangement of the feed hose, and could vary.
My recommendations for future tests:
We should raise the test stand tank so the line to the
engine can be strictly "downhill", avoiding most of the trapped air
issue, allowing the warm up pulses to be shorter. We may need to extend
the angle iron we have the tank secured to. At the same time, we may want
to add a hook for the fill cart line on the test stand so we don't have to
worry about it laying down and letting some of the tank drain back into the
fill line.
After we have done water runs, I should manually make very
short (25 msec) pulses until the plumbing is burped and the pack has gotten
some amount of heating before starting the automated test runs.
This probably won't be an issue at all with the motorized
ball valves, because they take long enough to open that they can't flood a cold
engine. However, the piloted solenoid valve may be particularly bad with
this effect, because it can't make a short pulse, and it still opens completely
full very rapidly.
Perforated Injector Plate Motor Tests
media.armadilloaerospace.com/2001_06_26/NewEngine.jpg
media.armadilloaerospace.com/2001_06_26/EngineTest.jpg
We tested a new engine configuration today, with very good
results.
The motor has a ½” throat and a 2” exit cone, and a 3”
chamber diameter. This is the same
throat dimensions as our last 50 pound thrust motor, but the chamber is a bit
wider so we don’t need to worry about the pack retainer hole area being near
the area of the throat. It is also
shortened to only the length needed to hold the catalyst pack, and the nozzle
is integral with the chamber.
When we used the highly compressed catalyst packs, we could
get by without any injector spreading plate at all, because the pack was dense
enough that it didn’t really allow tunneling.
When we used uncompressed catalyst packs by themselves, we
got very wet runs.
A couple weeks ago, we had very good results by placing four
tightly compressed unplated foam discs on top of 15 uncompressed catalyst
discs. We achieved a much higher flow
and catalysis rate that with the other packs, but the thrust curve dropped off
sharper than it should have, we believe due to the pack compressing itself as
the peroxide pushed against the compressed discs at the top.
We concluded that with a loose pack, it is necessary to have
a good spreading injector ahead of the pack, but it should be supported independent
of the catalyst.
Earlier engines had a coarsely drilled metal plate at the
top, but that had little effect.
This current engine is designed around using a sheet of
micro etch perforated stainless steel (McMaster 92315T101) clamped between the
top closure and the catalyst pack. The perforations
are only 0.006”, but they are so close together that the steel is
semi-transparent.
It can be cut with scissors, so we were concerned that it
might just rupture on the initial valve opening. We did a water test without any catalyst in the engine, and the
metal sheet held up fine.
We loaded up 200 ml of peroxide and fired the engine. And broke another 100 lb load cell. Sigh.
This time we banged the cell on the initial pulse.
We switched to the 500 lb load cell, and wrapped some heavy
rubber bands around the rear of the slide so it could never pull away from the
load cell. This gives us a four to five
pound bias in the graph numbers, but it will prevent us from ever banging a
load cell again. Interestingly, if you
look closely at the data logs, you can see the recoil after thrust terminates as
it stretches the rubber bands back by several pounds.
media.armadilloaerospace.com/2001_06_26/Perforated.xls
Run 1: 200 ml at 450 psi gave 58 pounds of thrust and ran
smoothly
Run 2: 500 ml at 450 psi started out the same and dropped
slowly with blow down, but was a little more ragged
Run 3: 500 ml at 600 psi gave 75 pounds of thrust, a
somewhat surprising linear scale up with pressure
Run 4: 500 ml at 600 psi with pro race (8.8 amp) solenoid
was only slightly lower than the big shot (30 amp) solenoid
The pro race solenoid has a poor characteristic in that it
leaks a bit when below 400 psi (their docs mention this) due to a hard Teflon seal.
The engine starts catalyzing FAST. Most of the thrust is there within 50 msec of the valve opening. We can probably do away with warm up pulses
completely, except after we have run water through the engine.
When we took it apart, the catalyst pack was completely even
under the injector plate. It looks like
we have quite a bit more catalyst than we need in this motor.
There was a slight dishing of the perforated steel plate
when we opened the motor, but not much.
The bottom pack retainer plate holes were too large, because
they allowed some of the pack foam to be pushed down enough to rupture. This may have been why the later runs were a
bit rougher than the first one. We are
going to look at using perforated steel for the retaining plate as well, either
on top of a normal plate with larger holes, or clamped as with the top closure.
Russ and Phil are going to work with this design a bit more,
then make four identical motors to this spec for the manned vehicle attitude
control engines. The next engine will
be the 1.5” throat engine, which should make around 600 pounds of thrust. We won’t be able to fire that until we get
the new test stand set up on my new land.
Software
Pwm2.exe, the test stand software, now has switchable calibrations
for all of our load cells – 100 lb, 500 lb, 3000 lb, and 10000 lb.
I wrote some code to directly compare the FOGs and our old
Gyration gyros. A graph of the simultaneous
rate output when I am spinning them together is at:
media.armadilloaerospace.com/2001_06_26/GyroVsGyro.xls
It doesn’t look like a lot of difference between the $100
and $1500 gyros (my wife’s reaction: “Honey, you got ripped off!”), but the
slight lag is important for stabilizing the flight control, and the little
bumps that are smoothed out do make a difference when accumulated. The shock and G sensitivity are also
critical issues.
I have a test program that properly handles the 3D rotations
of the axis instead of just integrating the rates. I need to get the flight computer moved over to this before we
start doing any significant translational maneuvers. I draw a 3D view of the axis live as you rotate the sensor board,
and display both the integrated and derived values. Our old sensor board does noticeably drift as you move it
around. I can’t test the three FOG axis
directly on my laptop, because I only have two A/D channels, so I am going to
have to set up a sensor forwarding program on the flight computer to compare.
New Electronics Box
media.armadilloaerospace.com/2001_06_26/BoxOpen.jpg
media.armadilloaerospace.com/2001_06_26/BoxInside.jpg
media.armadilloaerospace.com/2001_06_26/BoxBack.jpg
The new electronics box built around the KVH fiber optic
gyros is basically finished.
I am not going to do this again without getting a decent
drill press. I put the last two
together with a hand drill, but I have had enough of not-quite-straight and
not-quite parallel holes…
It’s fairly heavy at 14 pounds. The gyros are bulky, necessitating a larger box, we have a few
more boards inside, and I added a larger lead-acid 12v battery because the 40
amps we could draw when all solenoids were open was quite a bit more than the
lithium batteries we were using are supposed to put out.
I’m using AMP Circular Plastic Connectors for everything
now. You need a special crimper for the
contacts, but they work very well. On
the electronics box right now I just have the four connectors for the engine
solenoids, but we will eventually be adding several more for the motorized ball
valves, tank sensors, and remote GPS (because the box is too electrically noise
to have it mounted inside). I also have
all our load cells wired up with CPC for the test stand.
In the process of building the box out of Home Depot machine
screws, I ran across two completely defective screws, and two other screws
broke while I was tightening them. I
ordered a large set of stainless steel machine screws from McMaster that will
hopefully be of better quality.
When I was finally all set to start logging the
data from the flight computer, I was appalled to see up to 8 bits of noise in
the data from a 16 bit A/D.
I was scared that the gyros ($1500 each axis!)
were messed up, but even wiring a battery directly to the A/D still had all the
noise.
I moved the wires and cables around, but it
didn't make much difference.
When I powered the PC104 stack directly with a
5v bench power supply instead of through the 12v DC/DC converter, the noise
went away (well, down to two bits, which is fine).
I had been concerned about DC/DC converter noise
as it would apply to being used as a reference signal to some of our sensors,
but I had been assuming that the A/D board would condition its own power signal
well enough. Bad assumption.
Russ made a few quick modifications to the DC/DC converter
board today that I am going to check out, but we are probably going to have to
add our own power supply to the SSR driver board.
We probably won’t have things together enough to fly this
weekend, but it should be next week for sure…
News
