January 14 and 18, 2003 Meeting Notes
In the course of discussing robust electrically actuated
release mechanisms for the main chute deployment, the idea of using auto trunk
releases came up. Russ and Phil went
down to a local GM dealer and looked at a few different options, and brought
one to our Tuesday work session.
The latch itself looked pretty sturdy, but the casing was
just two pieces of stamped steel, so Russ welded the seams together for extra
strength. The actuator looked like a
solenoid, but it is actually a motor drive, so polarity on the electric input
matters. We made a mounting bracket for
the release, hung it from the chain hoist, and picked up one of the 250 lb
boilerplate tank ends for testing. We
just lifted it a couple inches off the ground with 2x4s underneath it, and hit
the actuator with one of our remote solenoid button boxes. It released perfectly, so we decided to load
it up and see how it held up. With Russ
standing on the tank end, bringing the weight up to 400 lb, it still operated
fine. Adding Phil, bringing it to 620
lb, the actuator would not open reliably with 12v, but putting two batteries in
series for 24v still did it. Adding
Neil, bringing it to 850 lb, the actuator would only intermittently operate
even with 24v. We sprayed some Teflon
lube in the actuator, and it seemed to work reliably with 24v after that. At 850 lb, it didnt show any signs of
structural flexing, so it seems stout enough.
The release latch did have a plastic coating on it that had been abraded
off by our overload tests, but it doesnt seem to effect the operation.
The non-operating load on the 2 vehicle drogue release will
be fairly high, possibly in the 800+ lb range for a supersonic deployment, but
the load at time of release will be under 200 lb. We are going to be using two releases for redundancy, so the load
on each will be halved. This release
seems to be perfect for the job, so we got another one, and welded the mounting
brackets to our bottom bulkhead.
After some discussion on aRocket about completely pyro-free
ejection systems, Rick Weber has graciously offered to build one of his CO2
ejection systems for us for initial drogue / stabilization chute
deployment. We could have lived with a
black-powder solution for this vehicle, but we have been told that being
completely pyro-free for our X-Prize vehicle will ease some of the regulatory
hurdles, so we might as well test it out on this vehicle.
Crushable Nose Caps
We tested our crushable nose caps on Saturday. We really had no idea how well they were
going to work, so instead of testing with the full vehicle, we bolted the nose cap bulkhead and a crush cone to the
bottom of the boilerplate tank end, which weighs about as much as the final
vehicle will weigh. We had a +/-50g
accelerometer mounted to the clamp plate inside the tank end, and I had paid
for the WinDaq-Pro update (ripoff price, but I dont have the time to write
something better right now) so we could get high sampling rates from our data
The crush cones are 0.050 thick aluminum, rolled and welded
to a 15 degree cone. The top has a 1
disc welded to it to close out the rolled cone, and the bottom slides and bolts
over a 6 diameter aluminum bulkhead with a matching slope on it at the top of
the main vehicle nosecone (which is made of 0.125 thick aluminum). There is about 6 of crushable space for deceleration.
We did the first drop from 3 above the ground, giving a 0.43
second drop time, and a 13.9 ft/s velocity at impact. The top of the cone crumpled up very nicely.
The second drop test was from 66 above the ground, giving
a 0.64 second drop time and a 20.4 ft/s velocity at impact, which is about what
we expect our impact speed under parachute to be. One of the weld seams did split, and the bolt holes elongated a
bit, but the entire cone crumpled
perfectly, bottoming one edge to the bulkhead.
Overall, these tests came out much better than we expected.
The accelerometer data didnt come out very well. The sensor is quoted as having 100hz
bandwidth, so I expected it to show 30 msec or so of sharp positive deceleration
on the main axis, but instead it had a vibration like pattern, bouncing above
and below 0g. I doubled the sampling
rate for the tall drop, and you can clearly see the ringing at a much higher
frequency than the sensor is supposed to output. I should have mounted the sensor on a little slab of rubber to damp
off high frequency oscillations. The
peak numbers look credible, at about 40 g maximum deceleration for the tall
drop. For equipment, that is completely
reasonable, and we are going to fly without any modifications. You can get a 40 g deceleration by dropping
something hard only a few inches onto a solid surface.
The X-Prize crushable cone will be over 7 tall, giving it
14x the deceleration stroke of the small cone, so it shouldnt be a problem to
get decelerations down to 10g-15g for the pilot.
We are probably going to try full-size drop tests for the
X-Prize vehicle in the relatively near future.
We bolted one of the tank ends to our mockup cabin cone, and our plan is
to add sandbags in the tank end until it is ballasted to the full landing mass
of the final vehicle (around 2500 lb).
We will just weld some chain link fence over the end to keep the
sandbags in place after the drop test.
We will need to rent a crane that can lift the weight 20+ into the air
for that test. Before we can do this,
we need to get the reinforced cabin hatch and the reinforced cabin-to-crush
cone section fabricated, as well as a couple full size crush cones.
I finally got the new Ampro PC104 board all working. I had to upgrade to a newer linux kernel to
use the on-board Ethernet, and I had some grief duplicating the entire
filesystem from our previous computer, but it is all working now.
I bought a 512mb compact flash from Synchromesh, and I was
able to finally use that for the filesystem.
The Kingston CF cards I had previously tried would always flake out with
ide timeouts after they had been used with Linux for a while. The synchromesh has given one non-fatal
dropped interrupt message, but everything has been working well so far. No filesystem access is performed during
flight, so it cant be a cause of an in-flight failure in any case.
One thing Ampro does which is still causing me grief is that
they actually plug the sockets on their connectors that are supposed to be key
pins. None of my other hardware does
this, and I cant figure out a way to pull the plugs out, so they are basically
forcing me to try ripping pins off of my other hardware. Another gripe is that the CF socket on the
bottom of the CPU board is too tall to allow another PC104 board underneath,
but they also have top-mounted (as opposed to edge mounted) ribbon cable
connectors for the floppy drive, so when you stack boards on top, you cant get
at it. I would have preferred to have
the CPU board on top, and just plug in rarely-used ribbon cables like floppy
and external ide when needed, rather than leaving the cables on all the time.
My standard advice for people thinking about using PC104 for
rocket applications: Get one of the Parvus
PC104 separating tools so you dont bend a lot of pins changing things around,
and make sure you use four metal standoffs between each board. Most vendors only ship two, often plastic,
standoffs with each board, but we have broken plastic standoffs and flexed the
boards enough to pop the bus connectors off in crashes if all four corners arent