May 12, 2005 notes
We made a new aluminum biprop chamber with reduced and
tapered cooling channels for higher velocity.
The chamber has 20 passages of 0.094 (3/32) width and 0.060 depth
along the chamber, tapering to 0.030 depth at the throat. At 6 gpm (23.1 cubic inches / second) alcohol
flow, this gives flow rates between 17 and 34 fps. Slop in the chamber
jacket will make the actual flow rates a bit lower than this.
The chamber thickness was also tapered slightly to make it
thinner at the throat, giving us a 1.40 throat diameter instead of 1.25. This also makes our L* even lower, which
hurts Isp, but also lowers chamber temperature.
We have previously has aluminum hardcoated at Dallas
Precious Metals, but it usually takes a few days. Russ was researching the hardcoating process and found that one
of the methods basically involves immersing the part in a sodium hydroxide /
water solution and hitting it with a high voltage / current AC power
supply. He experimented a bit with
using a welder as the power supply, and we seem to be getting pretty good
results from the homebrew hardcoating.
We coated the new engine.
We started off with 25% (by mass) water added to the
methanol for extra cooling.
DOT-5 brake fluid has been reported to work as an alcohol
additive to continuously form a silicon layer in the combustion chamber, but we
found that it didnt dissolve at all in methanol, and while it did seem to
dissolve in ethanol, a slightly visible puddle still separated out at the
bottom after a little while, even though the color from the brake fluid was
evenly dispersed. Ethyl Silicate does
seem to dissolve in methanol, but we havent tried firing it in an engine yet.
We made several long firings on the new engine, and it has
held up fine, even when we removed the water from the methanol. Isp has dropped a bit, down to 178 from 188,
almost certainly due to the reduced L* of the new engine. We tried a new injector plate that had the
fuel jets firing directly into the gox jets instead of at 90 degrees, but it
actually performed worse. The injector
also had the ability to puddle a bit of fuel, and we managed to get a preburner
hard-start, which popped the solenoid off and bent a flange. This injector design also had a
recirculation area at the top of the chamber that eroded some aluminum away.
Because of the eroded top and bent flange, we went ahead and
made another chamber. The latest engine
has a 1 longer chamber, which will bring our L* back up and a little higher,
so we expect Isp to exceed 190.
Theoretical calculations for various lox / alcohol / water
mixtures at 150 psi chamber pressure pressure:
O:F isp temp
ethanol 1:1 225 3105 tiny
bit better at slightly leaner ratio
75% ethanol 1:1 219 3058 peak
methanol 1:1 219 3096
methanol 5:6 222 3054 peak
75% methanol 1:1 208 2923
75% methanol 4.5:6 214 2903 peak
The watered mixtures dont drop the combustion temperature
all that much, but they can absorb a good deal more total heat as a coolant.
Our long test runs have all been at full throttle, because
the jetted solenoid as a preburner fuel control doesnt work well for
throttling. When the lox is throttled
down, preburner pressure drops, giving more delta-P across the jet and spray
nozzle, resulting in more fuel flow and excessive temperatures. We have modified a small ball valve to accept
one of our KZCO actuators to finely control the small amount of fuel going to
Our blast deflector problems are pretty much gone. After the extra-high temperature bricks (McMaster
9355K1) failed miserably, we poured a tray of ultra-high temperature cement
(McMaster 9876K1). This was sort of puffy
and we didnt expect it to do well, but it lasted a lot better than the
bricks. Still, it dug at least a half
inch pocket each run. Finally, James
talked to a local refractory supplier and they suggested Louisville fire
bricks. These have lasted very well
indeed, eroding less than a quarter inch each run, with no flying sparky
bits. We can also get them in 12 x 12
x 3 bricks, so we are just going to make a central slab that we replace every
ten runs or so.
When we initially set up the engine gimbal we had ball
joints at each pivot one at the top of the engine, and one at each end of the
linear actuators. This was under-constrained,
and the setup could flop around. We
replaced the engine pivot ball joint with a U-joint, which held things together
properly, but we also realized that the actuators only need a U-joint on the
base plate and a hinge at the engine side, which will tighten things up even
more. Even after all the linkages are
tightened, there is still a little bit of play in the linear actuators
themselves, but I dont think it will hurt us.
If we are forced to make the assembly completely rigid we will need to
go to a ball screw based actuator.
The closed loop control on our valves and jet vanes was
always simply full speed one way or the other, which gave them a jittery feel
that can be seen in the close-up videos and on the telemetry graphs. This was acceptable for those cases, but the
long arm on the gimbal magnifies it a lot.
I finally got around to adding software pulse width modulation to all
the motor drives and adjusting the control loops so the speed ramps down as it approaches
the desired position. It is very coarse,
running off the 186 hz IMU tick with only eight levels, but it still
dramatically smoothes out the motion, and I doubt the slower response will
effect anything. It is likely that we
wouldnt trip the thermal cutoffs on the motors now that they dont slam back
and forth continuously at full power, but we will continue to remove the
cutoffs, because we have found them to be extremely conservative, and tripping
one in flight would be disastrous.
Delivery times for aluminum tanks was excessively long, so
we are just going with a steel tank for the alcohol and a stainless steel tank
for the lox. For 10 gallon tanks, mass
ratios are a little better than the 7 gallon ones, at 37 pounds, but it is
still pretty heavy. We had to lay the
tanks sideways for everything to fit, but what first looked like a problem
turned out to be a benefit. We have a
manifold under each tank that drains the tank from two points, which should
make us basically immune to the slosh and swirl issues that can crop up when
draining a tank from a polar boss.
We are leaning towards using two gimbaled engines instead of
one engine plus roll control thrusters, because this is the likely
configuration of the next larger vehicle, and the current 350 pounds thrust
engines that we are firing would be the right size for the vehicle. However, I wouldnt be shocked if we flip
flopped again on this issue as the vehicle comes together. We are also still discussing the merits of a
single preburner with gox hoses to two engine chambers versus each chamber
having its own preburner. We are almost
certainly going to use just a single set of throttle valves and fork the flow
to the two engines, because there is no situation where we would want to
control them independently.
We are going to need to build new electronics again, because
a twin biprop engine needs more actuators and sensors than we currently have
available. Our Diamond A/D + DIO board has
plenty of bits 24 digital outputs and 32 analog inputs, but we dont have
connectors for all of them, and we need more H-bridges for the motor
drives. A single engine with roll
thrusters can just barely fit the existing boards if Russ makes a few trace
changes, which may yet be the strongest argument for it. On the other hand, we will certainly need
more capability for the next vehicle, so we might as well get on it now
A single engine needs the following bi-directional motor
drive / feedback lines:
The following actuators without feedback:
Spark box points signal
If we use roll control thrusters, we need two more piloted
The following sensors:
LOX tank pressure
Fuel tank pressure
Pressurant tank pressure
We also track five internal values: three isolated battery
voltages, regulated 12v and regulated 5v.