February 18 and 22, 2003 notes
We refitted the main bulkhead to the cabin, using an angle
grinder to make much deeper gouges in the aluminum before bonding than the
sanding we had done previously. We also
welded four custom brackets above the bulkhead location, so we could put
backing plates on the pilot side as an additional precaution against popping
loose. Ideally, we would have had the
custom brackets on the bottom side, but we cant weld that close to the
bulkhead without cooking the epoxy bond.
The hatch sealing lip is almost finished being welded on,
and we are working on the pressure seal and hatch dogs.
Small Vehicle Engine Tests
We broke in all four new engines on the small vehicle, and
we saw something rather interesting.
Because we didnt want to disassemble everything, we just flipped the
bulkhead upside down and clamped it to the trailer, so the engines would fire
straight up. This is a Really Bad Idea
for biprop engines due to the possibility of propellant pooling, but for a
monoprop break in, we decided to do it anyway.
When fresh cat packs are first fired, the exhaust is very
cloudy (after the first liquid gush of peroxide has cooked off), and clearly
visible. On several of the break-in
runs, the exhaust jet shot off at a pretty significant angle instead of
straight up. This was a result of flow
separation on the highly overexpanded nozzle (the throttle was only cracked, so
chamber pressure was very low) causing the jet to attach to only one side of
the nozzle. If this nozzle was designed
to be minimum weight, it would have buckled, but all of our engines are way,
way overdesigned, so it doesnt hurt anything.
We had never seen this effect so dramatically before,
because we never had the right combination of firing direction and
non-broken-in engines. It was
interesting enough that it got me thinking again about alternate nozzle
designs. I think we are going to
fabricate and test an expansion-deflection nozzle for our bench test 5.5
engine. We have two sets of nozzles for
these engines, 1.25 D throat, and 2 D throat. An E-D insert can be easily made to give the 2 D throat the same
annular throat area as the smaller nozzles, which would make for a perfect
back-to-back set of test to contrast the purported pressure compensating
effects of an E-D nozzle with a normal nozzle.
In any case, the engines are ready for liftoff now.
Aborted Hover Test
We were hoping we were going to have everything ready for a
hover test of the new control software (this will be our first use of four
servo valved engines for attitude control, all previous vehicles used solenoids),
but we didnt quite make it.
We insulated all the engines with about a 1/8 thick layer
of Fastblock-800, which should dramatically reduce the temperatures that the
parachutes and vehicle see. We already
had some Nomex gores RTVd to the bottom of the tank for some insulation there,
but insulating the engines is a superior solution, because it keeps the heat
away from everything at once.
On the advice of a reader, I tried moving everything to a
SANDisk brand Compact Flash for the flight computer. That seems to have fixed all the stability issues. To recap:
Kingston CF cards I was never able to get Linux installed without
running into fatal errors. Syncrotech
CF cards were just reliable enough to be dangerous I could get things
installed, but there would be an occasional IO error during disk intensive
operations like compiling. The new card
hasnt shown any problems whatsoever.
The AMPro BIOS still has a quirk of some kind that causes it to not boot
the CF on initial power on, but it finds it after a quick power cycle. I am probably going to ask AMPro about this.
Another improvement to the electronics that I finally got
around to doing was to add a proper connector for battery charging / external
power. I had always been just clipping
battery chargers to each battery, but now we have a multiple charger setup that
can be connected with a single CPC connector.
We took measurements on the vehicle in basically the
condition that it will fly:
Dry weight: 280
Total height, ground to nose cap tip: 120
Leg length: 9
Tube length: 68
Nose length: 43 (15 degree half angle)
We measured the CG to be 29 from the bottom of the tube,
which is nearly one caliber of negative stability. Hopefully the control system is up to it
What actually killed the hover test this week was the quad
motor drive board. We had tested it at
50hz cycling rate, but the flight computer cycles at 200hz, which started
killing off the transistors. Russ built
this one slightly differently than the previous one, so he is going to change
back to the old design, and use higher spec transistors. The next board will have the full complement
of actuators necessary for the X-Prize vehicle.
Im out of town next Saturday, and we are moving to a
bigger, dedicated facility (yeah!) the first week of March, but we will
probably be able to combine the hover test and the repeat of the crush cone
drop test the week after that.