February 1, 2006 notes
Igniter / Engine work
We have built a new test stand arrangement that gives us two
benefits increased tank size (2 x 13 gallon spherical tanks we will use the
2 x 100 gallon tanks on the vehicle for longer tests), and the ability to use
the tanks either connected to our vertical stand, or a horizontal stand. We have also been experimenting with sound suppression
methods, because we know we are sort of pushing our luck firing > 1000 lbf biprop engines behind our
While testing the new arrangement, we had a little hard
start on our test engine that blew the top closure off.
The ignition sequence on our engines has been:
Activating the vehicle (main trigger) opens the lox ball
valve to 30% to allow lox to chill down the plumbing and start filling the
engine with gox.
After a few seconds, I press the ignition button, which
opens a solenoid that runs fuel to a spray nozzle at the top of the engine, and
starts triggering the spark plug.
I have two options at that point if the igniter is working
properly, I release the ignition button and the main fuel valve also opens to
30%, putting the engine at idle. If the igniter isnt working, I can release the main trigger and
everything shuts down and auto-purges.
We have done over a hundred engine firings like this, but
ignition of the igniter spray fuel is sensitive to exact chamber details, and
we have had some experimental engines that wouldnt light 100% reliably, and as
the engines have gotten larger, it isnt as clear when the igniter is working. Small throatless
engines had an obvious jet of fire coming out the bottom, but on larger engines
we only really got to see the lox clouds turn clear.
What specifically happened this time was that we changed
valves for the new test stand, and at low throttles the flow through a ball valve
is very sensitive to exact positioning, so the 30% lox throttle on the previous
valves turned out to be slightly more open, and the little 0.030 jetted fuel spray
nozzle was getting overwhelmed with oxidizer.
The mixture by the spark plug didnt become ignitable until the main
fuel valve started to open, then it went bang.
We got lucky with our first firing, then we had some problems, then we
had it go bang.
We have been aware of the lox valve calibration issue for a
while, and had been discussing adding a dedicated solenoid to meter in a
specific amount of lox / gox, but I had been a bit
reticent to add another actuator to the engine setup. We had also known that ignition sensing was
getting to be an issue, and I was looking into some flame sensing
possibilities, but I also didnt really want to add another class of sensor. Nothing like a little bang
to get me over my reservations.
We decided to build an independent torch igniter for the
engines, in the style of XCORs, where it has
separate fuel and oxidizer solenoids, a spark plug, and a throat constriction,
so it builds igniter chamber pressure that can be reliably sensed. I realized that if I just put our existing
chamber pressure sensor in the igniter chamber, the same sensor could be used
for both ignition detection, and flight chamber pressure monitoring, because
the igniter shuts down after one second of operation, allowing the main chamber
pressure to fill back into the igniter chamber.
We are currently using a 0.018 fuel jet, a 0.060 oxidizer jet
(assuming it will be mostly gox, not lox), a 0.25
throat, and an L* of about 20 for the igniter.
I changed the control software to automate the ignition
process, so as soon as I squeeze the trigger, the engine lights up to idle all
by itself. There are two shutdown
checks, the first to see if there is igniter pressure, then the second to see
if there is chamber pressure after the igniter has shut down. We tested all the cases by alternately
disconnecting igniter and main fuel feeds.
The engine had one cracked weld and had broken the graphite
nozzle, but we fixed it back up and got it on the test stand with the new
igniter. The zirconium oxide coatings we
had put on the previous graphite nozzles didnt seem to last long at all on the
inside of the nozzle, so we just left the graphite bare on the new nozzle.
Our strategy for sound suppression on engine testing has
been to move the vertical test stand down into our truck loading dock well so
the plume is below ground level, and flood the well with several inches of
water. This has worked very well. The only downside is that so much water and
steam get kicked up during firing that camera positioning for viewing can be
(this is with the old ignition
style, not the torch igniter)
One issue with the new igniter is that there is a huge
difference in behavior based on if the igniter is getting gox
or lox. The test stand hoses leave a big
warm gox bubble in the line when they are filled, so
the igniter lights with a very low chamber pressure due to running extremely
rich, and the main chamber also runs extremely rich for the first couple
seconds. I started manually flowing some lox through the engine to get everything
chilled, but I want to avoid this if possible, because we are considering
linking our fuel and lox valves together with a single actuator, and I dont
want to add another valve just for chilldown.
The engine continues to work well, reliably making 1050 lbf at 350 psi tank pressrure. After the
first fifteen second run, we didnt measure any appreciable erosion of the graphite
throat, but when we took it apart after the second run and measured it more
carefully, we saw 0.05 wall erosion on the throat, and one area of slight
pitting. It is possible that we didnt
get an accurate measurement after the first run, since sticking calipers up a
hot nozzle on the test stand isnt super-precise, but it might also have
suffered some when I flowed some lox over it to chill
the plumbing while it was still fairly hot.
We are trying the silicon carbide based coating on the nozzle for the
next firing. It appears much harder and
better adhered that the zirconium oxide coating after it cured, well see how
it does with the firing. We also want to
investigate metallic plating to protect graphite.
Interestingly, we also saw some apparent reaction of the
graphite and the woven silica gasket that we used to seal the graphite nozzle to
the chamber. The frilled edges of the gasket that were exposed to the
combustion gasses did slightly melt, and the graphite had shallow indentations
under the gasket. We are going to try
using a graphite sheet gasket on the next test.
It will increase the thermal conductivity from the nozzle to the
chamber, but we dont think that will be an issue.
While repeatedly testing just the igniter, we did manage to
freeze the lox solenoid once. This just results in an engine that fails to start in a
completely benign way, and it shouldnt be an issue with realistic vehicle
operations, but we may try and do something about it.
The other issue we had was that the retaining plate that
clamped the graphite nozzle to the chamber warped after the run was over and
all the heat soaked out of the graphite.
This was expected. A thick
stainless steel plate would probably work, but we are going to try a ceramic
insulating spacer on top of an aluminum clamp plate.
I was originally hoping that we could just make the clamp
ring out of machinable ceramic (McMaster 8479K4
Alumina-silicate), but it turned out to be far too brittle to put any bending
load at all on. In fact, the test flange
with all the bolt holes literally came apart in Russs hands when he was
looking at it. There is a fairly
involved firing process you can go through that is supposed to increase the
strength by 40%, but it involves a programmable temperature controlled furnace
that can reach 2000 F, and I don't think I would be all that impressed with
even a 40% toughness increase. As a
spacer in compression, it seems to work fine, and it really is easy to machine.
We received our new motors for the gimbal
linear actuators, which was the last remaining piece for the new vehicle. The stock motors were 4 amp
stall current and, more critically, had very conservative internal thermal
cutouts that would stop the motor from moving if it was jiggled back and forth
rapidly by the computer for ten seconds.
We replaced them with 12 amp stall current motors to give us a little
more speed, a lot more force, and run-until-they-die capability. They have the same mounting pattern, but we did
need to grind a flat on the new motor shaft.
We moved the engine over to the vehicle, got everything
hooked up, and went through a full system test on the vehicle. We found a few issues, but nothing big:
Something was wrong with the fuel tank pressure gauge,
causing it to respond extremely slowly, as if it had a clogged snubber in front of it.
We pulled it off, but the aluminum attach point was galled, so we are
going to have to weld a new one on.
The gox igniter solenoid isnt
working now, something must have happened to the wiring.
When we loaded lox into the tank, we realized that we should
stand off and insulate a lot of our wiring, because all the aluminum around the
base of the vehicle gets real cold, and I worry about some of the wire
insulation and tie wraps.
We still have a hot gox bubble in
the lox feed line to the engine, we will try and rotate around some of the
sanitary clamps so it flows continuously downhill.
We need to use real shafts and bearings on the gimbal to engine hinge mount, because the current
hole-and-bolt arrangement contributes half the slop in the assembly. The other half is from the lash in the linear
actuator acme screw, which we cant do anything about. We have some ball screw actuators, but they
are much larger than what we want to mount here.
Vehicle dry weight is 540 pounds. With the current engine, we would only be
able to lift off with 400 pounds of propellant, but we will eventually upsize to
a bigger engine that would allow a flight with the full 1500 pound propellant