[AR] Re: Successful liquid launch at FAR
- From: Robert Watzlavick <rocket@xxxxxxxxxxxxxx>
- To: arocket@xxxxxxxxxxxxx, Paul Breed <paul@xxxxxxxxxx>
- Date: Tue, 18 Jan 2022 21:49:37 -0600
Paul,
On 1/18/22 3:39 PM, Paul Breed wrote:
Robert,
A couple of questions....
If you were doing it again, what would you change?
Specific questions about the design as flown...
Vehicle-wise:
1. I would have put the GHe (tank pressurization) and GN2 (pyroless)
fill ports lower on the vehicle for easier access. I also would have
made the access panel larger and probably out of a thicker material than
the 0.016 inch thick skin. I already was going to place some aluminum
tape over the DZUS fasteners anyway since they were kind of loose but
when I broke off the access panel hinge rivets on the pad, I ended up
taping over the entire panel. The access panel for the avionics bay
worked much better since it was made out of the 0.1 inch thick
fiberglass reinforced phenolic body tube. I think keeping the avionics
switch panel fwd on the vehicle and away from the hot flamey stuff is
still a good idea since a burn through of those wires would prevent
moving the valves to a safe position in the event of a fire. A larger
pressurization access panel would have also helped because at 3000 psi,
the pin depressor in the strut valve was really hard to turn by hand.
2. I should have found a better place to put the telemetry antennas.
They were in the recovery bay right next to the GPS antenna which
required a special GPS antenna with a filter to reject the 900 MHz
primary TM signal. The GPS antenna was supposed to be in the avionics
bay (that's why it was fiberglass/phenolic) but the EMI from the flight
computer swamped the uBlox GPS receiver.
3. I would have made the vehicle with a smaller aspect ratio. The long
skinny design has stability issues with wind, which fortunately weren't
a problem on Saturday. The design though was driven from choice of
modified oxygen cylinders so it didn't make sense to use a larger
diameter body. I liked the oxygen cylinders though since they had a
thick heel to put a boss port into and had a neck to use for an inlet
fitting.
4. As you probably noticed, I had a lot of fasteners. In several
places there wasn't enough clearance to use a the floating nutplates so
I cut off one ear and used JB Weld to make a Click-Bond style nutplate.
The JB Weld cracked loose in many occasions during build so I should
have used MS212061 one-lug nutplates with two fastener holes on one
side. I really liked the floating nutplates though because they
compensate for misaligned holes between the skin and stringers. And
they allow you to take the skins on/off on the accessible bays without
worrying about stripping out threaded holes in the stringers. A lot of
people commented "wow, that's a lot of screws!" Yes, that's true but my
informal survey shows skin/stringer construction to be the lightest
option at 0.36 lbm/ft. The closest was PML CF ultralight with a 0.040
wall and that came in at 0.44 lbm/ft. I think the PSAS team had a
honeycomb sandwich design that was super light though. The main thing
you gain with a skin/stringer design is good access for tubes, wrenches,
etc.and lots of places to attach things to the flanges.
5. I should have made the flight computer smaller and/or more easily
removable. It consists of three boards, a power supply, a
microcontroller board, and an analog signal conditioner board. Luckily I
didn't have any failures but it would have been a real pain to get any
of those boards out for replacement, especially since there's a shared
heatsink for the TM transmitter and power supply board that ties into
the airframe.
6. I used miniature Endevco/Kulite analog sensors with millivolt output
but it would have been nice to have all digital sensors as I could have
eliminated the signal conditioner board. However, using all digital
sensors implies potentially using a common bus which could be a single
point failure.
7. I was worried about FOD or dripping water from the LOX tank getting
into the main valve servo. I made a makeshift rain cover out of some
aluminum tape but it would have been better to completely encapsulate
the gear assy and servo in some kind of 3D printed cover. There's not
much room down there though.
8. I should have used 0.028 wall aluminum tubing instead of 0.035
tubing to save weight and make it easier to bend.
9. If I had tried harder, I could probably have shortened the vehicle
by 1-2 feet by making some custom manifolds instead of the ratsnest of
tubing in the pressurization and engine bays.
10. I considered adding externally powered solenoid valves to
depressurize the tanks in the event of a flight computer failure but
they were big and heavy. Luckily I didn't have a problem. My argument
against it was that the helium tank was DOT rated and the LOX tank had a
burst disc which would have self-relieved. The fuel tank could have
remained pressurized though. If there had been an explosion on the pad,
backup valves might not have helped anyway so the Winchester
depressurization method might have had to be used (not sure if FAR has a
rifle on hand - haha).
11. A slightly longer recovery bay (by a few inches) would have made it
easier to pack the parachutes.
12. Stronger magnets on the umbilical or perhaps a different design
would have avoided that last minute trip to the pad. There was only
minimal clearance between the ground side of the umbilical disconnect
and the launch rail so when I was disconnecting the LOX fill hose, I
likely bumped the rocket enough for the umbilical to pop off. I wanted
to avoid the fins getting snagged on the umbilical so I had to add a
bungee to pull them away at launch but that reduced the hold force that
the magnets provided. The umbilical could have been placed near the
base of the vehicle but then you have a lot of wires running through the
engine bay which isn't ideal. There are other disconnect ideas out
there I've seen but I just though the magnetic was kind of different so
I thought I'd try it.
13. Software-wise, I wish I had completed the moving map (I ran out of
time) so I would have had better situational awareness of where the
rocket was during descent.
14. The threaded end cap for the the spring loaded SAFE/ARM switch
rattled off during the trip out there. I used blue threadlocker on all
non-locking fasteners but I missed that one. Luckily I noticed it prior
to launch.
Things that didn't work:
14. I tried to use a spring scale to measure the weight of the vehicle
to ensure I completely filled the LOX tank. During a static test one
time, I thought I had plenty of LOX loaded but apparently it was only
half full. The vehicle/rail combination had too much friction so I
didn't end up using the spring scale. I wish there were some reasonable
solutions for cryogenic level sensing.
15. I added a small needle valve to vent the LOX hose prior to
disconnect but it froze up and I couldn't turn it. I ended up
disconnecting the LOX fill hose under pressure which provided a little
excitement. I had purchased a Rego valve like they use on the dewar but
it was really heavy and I was worried it would put too much moment on
the LOX fill port without some sort of support.
Were all valves driven by AX12A servos?
Yes - there were 4 servos, one for the GHe supply, directly connected to
a Swagelok SS-4P4T valve:
http://www.watzlavick.com/robert/rocket/rocket1/photos/dsc_1174m.jpg
The Fuel and LOX vent valves were directly connected to my modified
Swagelok SS-62T6 ball valves:
http://www.watzlavick.com/robert/rocket/rocket1/photos/dsc_1191m.jpg
The main valves were ganged together with a 3:1 gear reduction:
http://www.watzlavick.com/robert/rocket/rocket1/photos/dsc_0081m.jpg,
http://www.watzlavick.com/robert/rocket/rocket1/photos/dsc_0080m.jpg
I liked the smart servos vs. traditional RC servos because you can read
back the position, torque (really just the current), temperature, etc.
And if you lose power, they stay in the last position.
What exactly did you use for the He pressure regulators?
Aqua Environment 1247-1, with the female NPT inlet modified to be a -6
boss port. I really hate NPT fittings so I wanted to get rid of as many
as possible. There are Aqua 955 check valves on the outlets.
What did you use for ignitor valves?
GEMS A2012-2M-OC-V-VO-C204 solenoid valves, same as I had on the test
stand. They are fairly heavy (~0.5 lb) and they were 24 VDC so I had to
power them externally via the umbilical. I could have got them in 12 VDC
to power from the vehicle but the current plus the AX-12A servos would
have required a voltage regulator more than the 2 A one I had selected.
Plus, I wanted to minimize the electrical noise from the solenoids -
they're controlled by an opto-isolated SSR.
Did you do anything special to keep your ball valves tight?
A lot of times people put Belville washers on 3 pc Lox valves to keep
the valve tight as temps change?
Any seal/thermal issues with what appears to be stainless Steel ball
and aluminum body? (I could have that wrong)
Any more details on the seals in the lox ball valve?
I kept the ball, PTFE seals, and Belleville washers from the Swagelok
SS-62T6 valves. I never had a problem with downstream leakage past the
ball. The main problem was with the stem seals and we had a discussion
on that a while back on aRocket. I wanted to reduce the actuator torque
when cold so I used spring energized PTFE seals. They sort of worked
with a small amount of leakage which I fixed by using two seals and
pressurizing the cavity between them with the helium supply. They still
leak a bit during tank pressurization but once LOX starts flowing, there
should be a slightly higher pressure at the gas supply vs. at the valve
because of the pressure drop through the plumbing. Once I did that, the
leakage out the valve was very small. The fuel valves are leak free of
course. I used #10-32 aluminum bolts to hold the valves together
instead of SS to save a lot of weight and to self-compensate for thermal
differences. It would be nice if some university group could take the
existing design and improve on it to make it leak free (hint hint) but
still being cheap to make and controllable by a small servo actuator.
You put purge bags around the valves/actuators..
If you were going to do it again would it be earlier to just make the
rocket body panels purge tight?
I put the purge bag around the valve because I have open bearings (so I
can clean them for LOX) and during multiple back-to-back LN2 tests,
water condensed in the bearings and froze. It would be really hard I
think to make a skin/stringer design air tight. Well, maybe not with
appropriate seals - airplane access panels are usually well sealed
against moisture. But between bays, there were big holes in the body
frames with wires and tubes running through them which would be hard to
seal.
Do you have a total weight breakdown for all the components?
SEe
http://www.watzlavick.com/robert/rocket/rocket1/design/rocket1_mass_properties_public_20220120a.xls
The as-flown configuration had about 16 sec of fuel and apparently 10
sec of LOX so the final weight was slightly less than the spreadsheet.
Again congratulations on a fantastic flight!
Thanks! And thanks for coming out to see it and helping to recover it.
-Bob
Paul
Other related posts: