April 11, 2004 notes
Up in the air (sort of)
The thermocouple amplifiers still didn’t arrive. Axiomatic said they would ship next-day
delivery on Wednesday, but when I called on Friday looking for a tracking
number, no one was answering the phones.
They have now missed both their quoted ship date and their verbal ship
date. I went with their product because
they offered 12V power and a decent case, but I am probably going to look
elsewhere for the next batch.
We did a careful valve cracking calibration of all the vehicle
valves. We had done this before, but I
had botched the correction code, and we weren’t all that precise. To make a better check than just listening
for gas escaping the nozzles, we stretched disposable gloves over all the
nozzles and watched for when they started to inflate. This worked very well.
There were two correction factors that the code applied to the throttle
position that I wound up removing: the A/D raw to native conversion nudged the potentiometer
range in a percent on each side so that the values would stay truly in the 0%
to 100% range even with some noise, and the flight control software scaled the
desired throttle by 70% so it would never be in the nearly fully open ball
valve range with little authority. I
removed both of these (I limit the controlled throttle explicitly now instead
of implicitly), and now the cracking points are all at approximately the same
level we see on the test stand data feedback.
The individual valves cracked at different points ranging from 11% to 15%
before correction, which obviously makes them not warm up the same at low
throttle levels. With correction, they
all crack at 11%.
It turned out that the last warmup test that cooked the
wiring to the master cutoff valve also shorted out the master cutoff motor
drive by exposing bare wire. When we
were examining the wiring, we probably twisted it enough to let it dead-short,
frying the transistors on the motor drive board. This was the hack job cutoff computer that I had patched together
out of a PC104 computer stack and one of old motor drive boards, so it wasn’t a
bad thing to replace this with one of our new single board microcontroller
based master cutoffs that Tommy laid out.
The one amusing thing about the new board was that Russ’s quick program
for it had a 64k rollover issue, so as we were working on the vehicle, every
once in a while the cutoff valve would move for no apparent reason.
Based on the difficulties we had drawing a vacuum at the
test site last weekend with the leaking valve, we rearranged our fill plumbing
to use the nitrogen powered venturi pump to draw a vacuum. This draws a very fast vacuum to about –10 psi,
while only using a small amount of nitrogen in proportion to our pressurization
load. We tried using it on the big
vehicle tank, and somewhat to our surprise, it was a bit faster than our
previous drum pressurization loading, and any subsequent multi-drum loads would
go much faster, because we pull the vacuum from a top port while loading from
the bottom, resulting in an accelerating fill rate.
http://media.armadilloaerospace.com/2004_04_11/vacuumLoad.jpg
We decided to go ahead and test the vehicle at the
shop. We are probably pushing our luck
from a noise standpoint, but everything went fine. We doubled up our shock straps so the vehicle only hung about 12’
below the lift forks, and it had plenty of room to swing around. We had originally planned to use a larger
crane and let it have 22’ of strap, but it turns out that we can get a lot of
testing done with the shorter hang.
Since we didn’t have on-board thermocouple amplifiers, we made
a long extension to allow us to look at two of the thermocouples with our test
stand acquisition equipment
When we were loading the propellant, it was convenient to
just wheel the entire scale over to the loading station, so we could watch the
weight drop as the filling progressed.
To avoid the manual throttling during the warmup, I had
programmed two buttons for specific throttle positions: 15% for the steady
warmup, and 40% for the slug of propellant to get it going. The first test wasn’t able to warm up with
these settings, with temperatures that wouldn’t continue climbing at 15%
throttle. I found that they all warmed
up fine if I manually ran them up to 25%, but at that point they were making a
pretty significant amount of chamber pressure and going through a lot of
propellant.
I raised the steady fraction to 20% for the second test, and
they all warmed up, but the engines were still fairly hot, so it isn’t
necessarily representative. We also saw
another issue with the slug / idle warming method – if some of the engines are
fully hot and you are just trying to get a problem one going, hitting the slug
button can cause the vehicle to almost hop in the air. I’m afraid we are going to need to develop a
completely software controlled warming / maintaining procedure that individually
tailors the throttle to bring all the engines up to temperature, then maintain
them there. We were going to have to
develop the maintaining profile for the unpowered descent of the vehicle
anyway, so this isn’t a big problem.
When we got all the engines warmed, Joseph lifted the
vehicle several feet off the ground so it wouldn’t impact if anything shut
off. This was somewhat problematic,
because the vehicle started rolling and tipping a bit, and it picked up some
more angle during the throttle up.
The main throttle was being controlled by a GPS velocity feedback
loop, aiming for 1.5 / 0 / -1.5 m/s rate of climb depending on the joystick hat
button. On throttle up, the vehicle
took off at a bit of an angle. When it
was cut off, it swung quite a bit under the lift, but it was harmless. We steadied it and did some more
experiments. Manually throttling up to
a level below liftoff while stabilized behaved reasonably, but when I went for
liftoff it again headed off to the side.
When we first looked at the video, we thought that it was
flying straight, just with an incorrect up vector due to zeroing the gyros
while it was swinging under the lift. However,
after I had time to analyze all the telemetry, I found that it was fighting an
increasing angle, but couldn’t quite correct it. The motors were able to follow the calculated differential gain without
any problem, but that amount of differencing isn’t enough to control the
vehicle. If the vehicle had started off
from level ground without any initial angular kick, it probably would have been
able to hold zero angle, but it couldn’t catch a higher angle with propellant
weight transfer.
I should be able to just increase the tipping gain and
correct for this.
For the same differential throttling, we would also get more
authority by running at a higher tank pressure. At 200 psi, it required 60% throttle even with the 1” valves to have
liftoff thrust. If we are a little more
patient with the venturi pump we can use less nitrogen during the filling
process and probably get another 10 psi or so, and we are probably going to
bypass the regulator when dumping cylinders into the tank so it goes faster at
lower pressures, allowing us to cascade fill more quickly. We could always move to using an additional
bottle, but it is very convenient to just hook up to a single six-pack of
nitrogen bottles for each test flight.
We are going to ask our suppliers if they are willing to fill the
bottles a little higher, to 3000 psi instead of 2500 psi. The burst disc is 4500 psi, so that shouldn’t
really be a problem. Obviously, it
would be easier to do all this testing on a vehicle with a smaller tank.
We are probably going to try gently resting the vehicle on 4’
sections of pipe after it has been lifted into the air, so it won’t pick up any
rotation before liftoff. The pipes should
just fall over as the vehicle lifts up.
Also from the telemetry, I was able to tell that the engines
have a markedly different ramp up rate, but they all make similar chamber pressure
for a given throttle after they get going.
Engine zero is still somewhat rough as the throttle goes up. These four engines have been cut apart and
modified so many times that it is little surprise that they aren’t
identical. We are having new 7” chambers
and nozzles machined for eventual replacement of these engines.
Both hops were aborted by the computer for exceeding 20
degrees angle before I commanded them to do anything.
The GPS lost lock when the vehicle swung 40 degrees from
vertical, and it took two seconds to regain it.
After I increase the gains, if the motors may have trouble reaching
the increased deltas fast enough, there are two things we can look at: Check to see if we are getting a voltage
drop on the actuator battery when all four valves are cycling, and change to a
larger battery if necessary. The
throttling, tipping, and rolling control are all time-division multiplexed on
the actual valve commands (50% for throttling, 38% for tippinig, 12% for roll),
so I may try adjusting these, or possibly even making them dynamically
allocated.
Post test inspection showed that all the vehicle insulation
held up well, with no signs of any heat damage. All the electronics stayed operations through everything, and all
of our ground systems behaved as expected.
We should have a nicer test next week.
http://media.armadilloaerospace.com/2004_04_11/firstBigHover.mpg
Jet Vanes
The continuing issues with engine warmup and variability are
boosting the value of a single engine, jet vane controlled vehicle to us.
We successfully tested a jet vane driven by one of our valve
actuators. Under a 400 lbf engine, the
2” by 3” vane didn’t have any problem being turned. I intentionally started the engine up without complete warming,
so you can see some of the shocks in the exhaust streaks.
http://media.armadilloaerospace.com/2004_04_11/jetVaneTest.mpg
This engine with the 4” tall ceramic monolith has been
getting worse and worse, and has always required the cavitating venturi to run
smoothly, so we decided to cut it apart and convert it to compressed ring
catalyst before using it in the new test vehicle. Inside, the catalyst monolith was shattered badly, so there is
little wonder about the poor behavior.
John Carr said that this type of fracture is usually due to thermal
stresses, which is understandable in our application.
http://media.armadilloaerospace.com/2004_04_11/shatteredCatalyst.jpg
We are building a full jet vane control system for the old
manned lander framework. To give us the
maximum control authority without allowing the vanes to collide, they are
triangular in shape. There is a pretty
good chance we will be able to fly this vehicle in about two weeks.
http://media.armadilloaerospace.com/2004_04_11/fourVanes.jpg
Big Engine V2.0
We assembled a new 12” diameter engine with the following
changes:
Water jet cut support plates instead of braced perforated
metal plates.
Single layer of 3000 grams of catalyst
Catalyst retention plate welded in under 8000 psi gauge
pressure (about 16000 lb force, or 141 psi across the catalyst)
Spark ignition, double plate flameholder
Single 1” thick 900 cpsi cold pack monolith
Total weight is 111 lb using one of the lighter nozzles, but
the flanges aren’t necessary in a welded configuration, so that gives us right
about 100 lb for the engine fabricated in all 316 stainless. I wouldn’t want to thin them out much more,
so if we want lighter weight, we will have to go to superalloy fabrication.
This engine should have less pressure drop across the cold
pack, but more across the hot pack. It
should warm up quickly and run smoothly, but we probably won’t get a chance to
test it until something holds us up on vehicle flight testing.
http://media.armadilloaerospace.com/2004_04_11/new12.jpg
http://media.armadilloaerospace.com/2004_04_11/filled12.jpg
News
