June 15, 2005 notes:
Work is proceeding well on the vehicle:
Since this picture was taken, we have basically finished all
the plumbing, and most of the wiring.
Our first leak check was on Tuesday, which turned up a couple things
that need to be fixed.
I have decided to go ahead and build a second copy of this
vehicle. Doing so will be a lot easier
with the original sitting next to it, rather than crumpled up in a post-crash
We did a top-to-bottom inventory of all the parts that go into it, and it
turned out to be a longer list than expected four pages. It will be interesting to have everything in
a big pile, then turn it into a rocket.
Building wiring harnesses always takes longer than you
expect, especially when working with shielded aircraft cables. We are still working with AMP metal shell
circular connectors for this vehicle, but I am actively working to move our
next iteration to higher end connectors that will be sealed as well as
shielded. I was originally looking at
MIL-C-5015 Amphenol / Cannon connectors, but I was strongly urged by a helpful
aRocket contributor to use the more modern MIL-C-38999 Series III connectors
instead, for sensible-sounding reasons.
These connectors are several times as expensive, but more annoying is
the fact that you cant just click and buy the connectors. I have been waiting a week for a distributor
to get back to me on my second quote.
We calibrated several fuel / lox throttle points, and did a
long run throttling the engine up and down several times. The throttled cases were intentionally very
rich for cooling safety, but the fuel flows were higher on the variably
throttled run than on the calibration runs, perhaps indicating that we have a
lash problem with our fuel valve actuator.
We were preparing to test a new style of injector that moved
the fuel injection point in from the side to the same radial distance as the
lox injection. The positioning should
help, but I also managed to get 4x as many 1/32 holes drilled as we used to
use 1/16 holes, which should positively be an improvement.
Unfortunately, we had another chamber burn through before we
could test it. We have a pretty good
idea what caused this one the weld at the base of the lox injector cracked,
leaking lox into the chamber from there and allowing it to burn and melt off
some aluminum. When molten aluminum
falls from the top of the chamber, it can pick up a lot of heat and smack into
the throat converging section, allowing a burn-through in the otherwise
steady-state cooled area. We have had
weld cracking problems in cases like this where we have had to face off a
welded seam to allow a flat plat of some sort to bolt directly below it. Better V-grooving of the weld area would
help, but for the new engine I made both the fuel injection ring and the lox
injection ring together from a single billet of aluminum, so there are no welds
to crack at all. We have also changed our
fuel and lox plumbing to tangent entry instead of perpendicular entry, which
should improve distribution. On water
testing, we could see that there is still a noticeable difference in flow out
of the channels around the circumference, so our next engine iteration will
provide much more manifold space.
We are getting a professional hardcoat put on this engine,
but it should be ready to test this Saturday.
We are also going to start out with a fuel additive. We have switched over to ethanol for future
work due to the slightly better handling and performance, and we tested both
DOT-5 silcone brake fluid and ethyl silicate.
The brake fluid has reportedly been used by others, but at least the
brand we tried (Russel performance) seemed to settle out somewhat after sitting
for a little while. The color stayed
evenly distributed, but there was a puddle of different refraction at the
bottom of the flask. Ethyl silicate
seems to be completely soluble up to very high concentrations, so we are going
to use that. We are going to start out
with a 2% solution, but we hope that we will be able to do without it. We want to make sure we get at least one
good run first.
We built and tested a third-generation free-piston
pump. As a development aid, we used 4
diameter commercial air cylinders for the chambers. This automatically gave us a convenient, well-sealed piston, and
the actuator rod is an excellent visualization aid. They are very heavy, with huge slabs of aluminum on each end, and
steel tie rods and actuator rods, but when we get around to making a
lightweight version we can still take the cylinder and piston, and just weld
caps on each end.
The big piloted solenoids on the last pump gave us a fair
amount of trouble, because the vents wouldnt always close reliably when the
piston was all the way at the top and all the pressure had completely
vented. For the new design, we use a
single 3-way ball valve on each cylinder that flips the chamber from pressure
to vent. By arranging the two cylinders
properly, we were able to use a single pneumatic actuator to cycle the valves
on both cylinders simultaneously. This
is a really neat arrangement.
We did vibrate off a nut on the linkage of our pneumatic
valve actuator linkage during testing oscillating motion brings in issues we
havent faced before. The actuator used
snap rings to contain its rotating parts, which is probably a good idea. Our 3-way ball valves are also not perfect,
because there is an overlap period where pressurant can travel directly to the
vent. Interestingly, this seems to be
more of an issue on the valve that we had to rotate the ball to get the
orientation correct. Some 3-way ball
valve seats arent designed for pressure in all directions. What we ideally would want is a ball valve
that had smaller plumbing ports than the ball, so there was no overlap period
at all. I think I have seen some valves
that claim this.
Overall, I think this free-piston design is just better than
the pistonless pump concept. We don't need any sensors, we waste less
pressurant, we have a single actuator instead of four, and two valves instead
of four. We still have the same four check valves. Warmer gas could
be used with a piston separating it from the working fluid.
The weight savings offered by a pump versus heavy propellant tanks is related
to the ratio of the pump cycle time to the total burn time. You get the
best benefit if you cycle fluid through a small pump really fast. XCOR is
driving their pump so fast they had to solve a lot of vibration problems.
We are using long cylinders and a long cycle time at the moment because it
makes understanding what is going on easier, but when we make a lightweight
version we are probably going to have the cylinders be half or a quarter the
length, and the cycle times proportionately shorter. This ceases to be a
benefit when your valve actuation times become a significant fraction of the
cycle time, when your trapped gas volume becomes significant compared to your
displaced volume, or when the gas used for actuating the valves becomes
significant to the displaced volume. I'm pretty sure you want cycle times
under a second, perhaps well under. If you are using off the shelf check
valves and vent valves, cylinder weight won't be worth optimizing away too
much, but if you make very lightweight integrated valves you would probably
wind up down at the speeds that XCOR is running.
Like a pistonless pump, a free-piston pump can't self prime, it requires some
head pressure to force liquid into the chambers. I considered connecting
the two chambers (especially since we already had actuator rods sticking out of
them...), but the packaging gets messier, and the pressure output then has huge
dips in it as the piston pulls away. Pistonless / free-piston pump chambers
each act as accumulators during the transition to the other chamber, so it is
easy to get smooth output pressure.
Even though we are having great success with this, I don't think we are going
to use it in any currently planned vehicle. We are going to be making 3'
spherical propellant tanks from spun 5086 aluminum hemispheres, and they just
aren't going to be that heavy for our operating pressures, even with a 2x
safety margin. We are going to be bursting a trial tank fairly soon to
see how closely it hits our expectations.
It only starts looking attractive if you can also significantly reduce the
weight of the pressurant bottle. Flowmetric's plan to use a separate pump
to pump liquid helium up to high pressure before vaporizing it sounds
troublesome to me, but with a piston you can deal with hotter gas, so using
cooled Tridyne might be easy and workable, or you could always go to the
conventional gas generator or chamber tapoff with a differential driven piston.
Our next big vehicle and engine tests are going to need a
much larger regulator to maintain pressure than anything we have used
before. Because of the nature of the
motors and tanks we were using, blowdown pressurization made good sense for the
monoprop vehicles, but we need tight pressure control for biprop work.
We built a manifold with a high pressure ball valve, a
pressure regulator, and a very large relief valve so we could experiment with
direct servo regulation from the high pressure bottles. This also has the advantage of allowing us
to start or stop the pressurization of the main tanks remotely, which we cant
do with a fixed regulator. I started
testing using regulated pressure at modest levels, but after getting the basic
proportional differential control loop coded, I was able to run it directly
from a high pressure bottle. It works
pretty well, and doesnt suffer from any droop at changing supply or demand,
but the ball valve seems to have significant control lash in it, cracking open
at 15%, but not shutting completely off until it goes back down to 5%. We havent determined yet if this is some
aspect of the seats in the valve, or if our actuator mounting has significant
lash. We are going to try making an
adapter using lash-free helical beam couplers.