May 22, 2004 notes
Bent jet vanes
Some shots I didnt have time to put up last week:
We pulled the engine back off the vehicle to rectify our
error of leaving the paper towel plug in the valve top last week. Unfortunately, only half of it was left
above the valves, and there were little pieces visible in the spreading plate
holes. We decided to go ahead and cut
the top plate off so we could clean everything out well, which turned out to be
absolutely mandatory, because there were shreds of towel in every single
spreading plate hole.
After cleaning everything out, we welded the engine back
together and reassembled the engine on the vehicle. We added an extra layer of Teflon sealant around the double
O-ring seal between the engine valves and the tank manway. This was the thing we were most worried
about sealing, because they are really tiny 1/16 thick O-rings, but it turned
out to work perfectly. We tightened up
the fittings on the drain valve and pressure gauge, so the plumbing is now leak
free except for a tiny drip on the loading connector. Unfortunately, there is a slight leak past both the master cutoff
valve and the throttle valve, so propellant trickles into the engine when the
tank is pressurized. This will
eventually auto-ignite and start heating the engine, which is good from a
warmup standpoint, but we think this is also how we cook the cold packs, by
having such a low flow rate that actual burning can take place above the
flameholder. We usually get these leaks
after we drill vent holes in the ball valves, either due to a metal chip or a
burr scuffing the seats. I have rebuild
kits on the way, but it will be a hassle to dismount everything.
All of the wiring was completed on the vehicle base, and
everything looked like it was ready to go, but the A/D board went out on us
again. We thought we had a pretty good
theory this time: it was working earlier in the day when I set up the vane
calibration, but between then and when we found the board to be bad again, we
had TIG welded a bracket onto the landing legs. The high frequency starting current on the welder may have found
its way into the wiring harness somehow.
We got yet another A/D board for the next day of tests.
The 1 foam cubes had worked so perfectly for supporting
tethered takeoffs of the little vehicle that we had 2 foam cubes made for the
big vehicle to replace our various attempts at fall-away supports.
We added a vacuum gauge to the new adjustable high flow
venturi pump, which turned out to be a very good idea, because the interaction
of the inlet pressure and the pump adjustment covered a very wide range of
efficiency, and when we were randomly adjusting it we were probably wasting
half the gas. I had started to use our
normal gear for the vacuum valve and gauge adapters, but I realized that I
could save a couple hundred dollars by using CPVC fittings and valves instead
of very expensive 316 SS 1 hardware.
The combination of the stronger vacuum pump and the hotter weather
causes the fiberglass reinforced PVC hose to get squashed pretty flat, but it
starts pulling propellant in much faster than the smaller two stage pump.
The leak past the valves was preheating the engine fairly
well, and when I brought the throttle up it was hot and clear almost
immediately. The throttle up burned
quite a bit away from the inside edges of the foam cubes, depositing lots of
sticky melted plastic goo on everything.
We need to put some insulation or coating on the inside edge to prevent
this next time.
Unfortunately, we lost the computer at throttle up. The master cutoff worked perfectly, shutting
off the propellant flow and avoiding what would have been a rather exciting
swing of the vehicle pulling on the tether.
I was able to telnet back in to the computer a minute later, which
showed that it had rebooted itself, but the Crossbow wasnt responding. We saw this exact behavior on a previous
test of the big vehicle a while ago, and a real power cycle did bring the
crossbow back to life. When this
happened before, I wasnt 100% sure if the computer had rebooted, or if I had
just lost the telnet connection, but this time I checked the system date, which
I have to manually reset after booting if I want to do any code building, and
it had indeed been reset to 1970, positively indicating a reboot.
A momentary voltage drop is probably the cause. The battery voltage drops from 12.5V to
11.7V when all the motors (vanes and throttle) are being driven, but the DC/DC
converters for the computer and other gear should work down to 9V source. There isnt any indication in the telemetry
that it is dropping below about 11.5V, but that is only a 20hz sampling, and it
seems reasonable that the jet vane motors will draw more current when they are
working against the exhaust stream, which is our current theory.
We set the vehicle back up on the foam blocks, but because
the Crossbow was still down, we had to take it back down and get the ladder out
so we could power cycle everything. All
this time, the leak past the valves was cooking propellant off in the engine,
and we were getting quite concerned that this trickle flow was going to burn up
the cold catalyst pack.
After the reboot, I throttled up the engine and found that
there was a stream of liquid coming out the center. We thought that we had probably burned out the engine, but I was
eventually able to get it running clean.
When I throttled up, it didnt crash, but it showed up another
problem: the lift truck was taking most
of the vehicle weight on the tether, so as soon as the throttle came up enough
to lighten it just a little bit, it blew the foam blocks away, but allowed the vehicle
to bounce on the tether. This up and
down acceleration confused the auto-hover software, causing it only increase
the throttle during parts of the negative G sections of the bouncing, resulting
in it only creeping the throttle up, and not getting high enough to actually
hover before it went into auto-land mode.
I did move the vanes around while it was sort of bouncing there, and the
directional control was basically working, but it didnt make it up in the air. Next time, we are going to let it rest most
of its weight on the foam blocks.
After pausing briefly to reset, I tried again, and it
rebooted the computer again. The master
cutoff again did its job properly. This
time, when it came back, the A/D board was hosed again. That was it for our testing. We didnt weld anything on the vehicle, and
I didnt even use the spark ignition system, so I am going to move to a
different brand of A/D board and see if we have better luck there.
When we laid the vehicle back down, we found that all the
jet vanes had been bent to various degrees, even though the engine had only
seen about 2000 lbf of thrust. The long
heat soak under the trickle burning engine before the high thrust run certainly
weakened them. We were more-or-less
expecting this, and we have ¾ shafts and 3/16 thick vanes ready to weld
together. ¾ bearings happen to fit in
exactly the same mounts as the ½ bearings, so we dont need to change any of
The actuator that was downwind of the engine was badly
burned. The protective cover box has a
slot that the shaft slides down, so hot exhaust and (during warmup) propellant
can still get through. We actually saw
a puddle of burning goo under the actuator at one point, and I wasnt sure if
it was melted foam pad that started burning up there, or melted actuator
casing. The entire mounting plate was
quite hot, but conduction through the shaft and mounting brackets didnt seem
to be hurting anything. As usual,
convective heating is the killer. We
are probably going to add some insulation boards around the shaft to block off
more of the flow. The burned motor may
have had something to do with some of our problems.
All the melted polyethylene foam came off the vehicle pretty
We were also prepared to finally do boosted hops with the
small vehicle, but we cant fly that vehicle either until we fix the A/D board.
We should have things back in shape to try again on Tuesday.
We will have the stouter vanes completed and installed.
We will get a different brand of A/D board and rewire the
breakout board for it.
We are going to go back to a separate battery for the actuators,
which should prevent anything from the actuators hurting the computer power. There isnt room on the electronics board
for another batter, so it is going to have to be externally mounted somewhere
in the cabin.
We are near the upper end of the current supply on our 5V
and 12V DC/DC converters, so we are going to build a new power supply board
with converters that are twice as large to give us plenty of margin.
Liquid catalyst testing
We have done experiments with liquid catalyst injection in
the past, and generally found it inferior to catalyst packs for our purposes,
but now that we only have a single engine on the vehicle and we are looking at
making 20,000 lbf class engines, we thought it might be worth a little more
experimentation. The upside would be
that there wouldnt be five week waits for custom catalyst monoliths, and we
would probably pick up a little chamber pressure and some engine thrust to
weight. The downside is that you need
to manage twice as many propellants, you get grit clogging up your plumbing, and
you wind up staining everything you work with.
Because we need deep throttling capability, a classing
liquid-liquid atomization injector scheme wont really work for us. We are experimenting with using static pipe
mixers to perform the mixing, which might be an interesting option with 50%
peroxide mixed-monoprop. The catalyst
and peroxide are flowed through the static mixer together, which then exhausts
into the main chamber with flameholders and a spark plug ignition.
We hacked together a test rig for this, and worked our way
up from just the static mixer, to open flameholder tests, to a welded on engine
Results were rather mixed.
We got a nice burn with the open flameholder, but when we added an
extension chamber below it, it chuffed very badly, and with the nozzle on, we
clearly werent getting full combustion.
We are going to add pressure and temperature instrumentation and try
adding additional layers of flameholders in the chamber.
One interesting point:
we are using potassium permanganate as the catalyst, and we were
surprised to find that while we could only dissolve 0.5% by weight in tap water
before condensate started to form, we could dissolve over 5% by weight in de-ionized
water. The tiny little impurities in
tap water make a huge difference.
We also have some sodium borohydride coming for various