January 31, 2004 notes
Analyzing the data from last weeks vehicle warm up test
showed that while all engines came up to pressure when a big burst was made, with
the throttle just cracked open, two of the engines werent making any
pressure. I added code to let us
individually calibrate the cracking fraction of each valve, so we can now
expect all of them to open within 1% throttle of each other. Still, micro throttle cracking isnt very
even, because 1% in the opening fraction of a 1 ball valve is a lot of flow. If we want to use a tiny idle flow, we may
need to add extra valves specifically for that.
PVC cementing the drum adapter directly to the 1 PVC feed
hose worked better than trying to RTV it, but by the end of the days testing
it was leaking a bit again. I need to
find a 2 buttress thread fitting in PVC that I can cement both the hose and a
support pipe directly to.
Our floor scale for weighing propellant inside drums gave us
a problem, where it basically stopped reading past about 250 lb, even though it
was a 1000 lb scale. We were able to make
an adjustment internally to fix it.
We have always had fairly noisy A/D signals on our flight
computer, but last week I saw a very strange behavior with several of the
signals dropping about 10% after the flight control program had run for a half
second or so and was polling them rapidly.
I decided to replace the WinSystems A/D board with a board from
AccessIO: http://www.accesio.com/go.cgi?p=../104/104-aio16-16.html This board also has 16 channels of digital
IO, so we were able to replace our dedicated digital IO board (also from
AccessIO) as well, reducing our total PC104 card count. Rewiring everything to a different breakout
board was a bit of a pain, but we got a few extra bits of precision out of it,
and the odd drop is gone. I also have
an AccessIO serial board that I am going to use to replace the WinSystems
serial board that doesnt seem to work with the AmPro CPU module.
Our Ashtech G12 HDMA GPS was still locking up on us, so it
went back to Thales for a second time.
They are going to replace the board this time instead of just flashing
the firmware and sending it back.
I finally figured out two little non-fatal but glitchy
things that have been with our flight control system for at least a year and a
half. One of them was pretty silly I often
had to hit enter a couple times to get the flight computer to acknowledge that
I wanted to exit the program. It didnt
seem to have any other effect, so I never chased after it, and just got used to
banging enter a few times rapidly when I wanted to quit. Many old sensor subsystems which are not
initialized are still being called to check for updates. Most of the systems are designed for this,
so putting them back in just means uncommenting the initialization line, rather
than having to hunt around for multiple places to fix code. However, in this case (the on-board pilots serial
joystick), the update wasnt checking if it had a valid serial file descriptor
before polling for input. Uninitialized
descriptor = 0 = stdin. Makes perfect
sense now. The other glitch was on the
laptop side, where I found that occasionally (every few thousand times or so
during throttle changes) the USB joystick reading would return a bad
value. No error, just a bad value. It is a consistent bad value, so I just
ignore it now and use the previous value.
This avoids a spurious throttle burp and occasional logging reset.
We managed to kill one channel of our lead-acid charger
system, so we replaced it with a new one.
Having a single hook up for charging all the batteries is very
important. I need to upsize the main
computer battery, because we had to stick the charger on a couple times during
testing to keep the voltage from dropping too low. Having to climb up a ladder and unbolt the hatch to connect the
charger each time was a bit of a pain.
Cabin-at-the-bottom on the next vehicle will make several operational
The propellant loading test we did today had one of the
engines start cooking off as soon as propellant was pumped into the
vehicle. It turned out that part of the
actuator to valve linkage was held together with just two little spot welds,
and they had broken, allowing the valve to slip 30 degrees or so before the
plate wedged up against something else.
Russ welded the plates together properly, and we checked out all the
other valves. The 2 valves had a nice
weld along the sides already, just like Russ had done, but the 1 valves were
all done with two spot welds, which were actually ground smooth, reducing their
strength. We now have three things to
do on the 1 KZCO valves: vent the ball, bridge the thermal cutoff, and seam weld
the actuator plate. We use the fastest
actuator gearing, which KZCO doesnt usually sell because it is difficult to
manually control, so this may not show up much with their normal
customers. This was a good thing to
catch on the ground
If we had good success with the vehicle tests today, we were
planning on heading out for hover tests next week, but we didnt get a
The first test had the valve problem. The remaining tests functioned properly in
all regards except that after removing the preheat plugs, we couldnt get clear
running main engines. We either got
huge clouds or they quenched out on throttle up. We let one test preheat until the warmup plugs were glowing red
hot on removal, so it was certainly quite hot.
We did get the vehicle to hop off the ground into the lift truck forks
(we set them up specifically to hold the vehicle down) with only 100 psi in the
tank, which means we have plenty of thrust.
I have a few theories: The combination of the smooth turn
elbows and the much shorter flex hose gives a higher flowing combination than
our test stand, so the engines might be getting drowned. We have never done testing with only 100 psi
tank pressure, it is possible that the engines need a higher pressure to wake
up. It is possible that the preheat we
did, which was a moderately high flow continuously with the plugs in, heated
the bottom of the pack quite well, but didnt heat the top, which would cause
We will do a bunch of test stand work with the flight
engines next week, and probably repeat the vehicle warming tests next
Saturday. On the bright side, the
electronics were flawless all day, and we are building a lot of experience
moving the vehicle around and working with drum scale quantities of propellant.
In other work, we have upsized all the plumbing on the 12
engine to 2, which should drastically improve the performance based on our
results with the 5.5 engine. If we run
the tank up to 450 psi, we should see 5000 lbf. We decided to skip the load cell and hard mount the engine
directly to the stand to avoid the shaking problems. We can calculate thrust from chamber pressure, although we will
likely underestimate it somewhat when scaling from the previous run, because
the nozzle factor will be better at the higher chamber pressures. I am getting to be a fan of the cam-lock
connectors we are using on the larger hoses, because they have a gasket seal
instead of relying on a metal to metal seal like the AN fittings. However, we have been seeing a lot of corrosion
on the aluminum pipes and fittings, even when we havent had anything but water
in them. I had hoped to save some money
and weight with aluminum in the larger sizes, but it looks like it may be
better to just stick with 316.