Week of December 22, 2001 Meeting Notes

 

In attendance:

 

John Carmack

Phil Eaton

Russ Blink (Tuesday)

Bob Norwood (Saturday)

 

On Tuesday, we built new catalyst packs for the engines, and tried to make new anti-channel rings. Mark Henry had explained to us that the rings should be fairly thick, and of the same material as the engine casing, so that they would expand at the same rate and stay pressed against the walls at all times. I bought an assortment of different brass sheet stock, but we broke a few end mills on Russ’s pc-board CNC mill trying to cut the thicker sheets too fast. Russ is going to work out a better process.

 

We also made a change to the way we are controlling the main ball valve. When we first started working with the ball valves, we had two motor drives burn out on us (complete with smoke!), so when Russ built a custom board for us, I had the software include significant delays between motor reversals to add some extra protection.

 

We later found out that the likely source of our problems was that the connector on the motor had been internally twisted up pretty badly, likely causing an intermittent short circuit.  I later lowered the turnaround delay from 100 msec to 50 msec without any ill effects.

 

On Tuesday, I took the delay out completely, and everything still seems to be fine.

 

This will help with the manual throttle control, because the latency makes it harder to hover and land smoothly.  It will also be beneficial when I get an altimeter hooked up and have the computer running the throttle in closed loop mode.

 

Our TZM moly bar stock is on the way. A 2.75” by 12” bar is 26 pounds, at $44 / pound. That should be enough for at least four test nozzles. If we get lucky and have our first coating (probably going to try plain platinum first) work, we can hang on to the rest for potential biprop tests at some point.

 

Russ was away for the holidays, so on Saturday, Phil and I put the engines together with the old control rings, and prepared for some normal flight testing.

 

The water test went fine, and after loading the peroxide, the engines warmed quickly and ran perfectly clear. Then things went rather badly.

 

I started the guided run, and began throttling up. The engines were all firing, but it wouldn’t lift off. I was throttled all the way up to maximum, about to shut down and reassess the situation, when it jumped off the ground. It snapped off both tethers like they weren’t even there, and flew up out of my field of vision from inside the garage.

 

As soon as it left the ground, I throttled it back, but I was thinking that the motor drive was stuck in some way due to our changes on Tuesday. When I couldn’t see it, I completely killed the control. I was gratified to hear the engines cut off, which I was a bit afraid wouldn’t happen if the motor drive was smoked. When I could see it again, I reengaged the attitude engines, which straightened it back out, but it still plowed into the ground pretty hard. This definitely looked like a better way to crash than the one we had with the offset CG.

 

http://media.armadilloaerospace.com/misc/splat.mpg

(our professional camera guy wasn’t there yet, so it is messier than usual…)

 

When I looked at the telemetry, it showed the throttle ramping up to 25%, then staying there for several seconds and jumping all the way to 100%. The motor drive was doing exactly what it was told, but the entire throttle ramp up that I did was just missing. I was almost positive that that meant the flight computer wasn’t receiving packets from the laptop over the wireless link (even though the laptop was getting packets from the flight computer). We have had some signal quality problems (the 1000 mW Esteem 802.11b units just showed up on Friday, but I hadn’t integrated them yet), but the computer is supposed to just shut everything down if it doesn’t get any remote pilot packets for 200 msec.

 

I was considering that I might have broken that code when I added the local pilot joystick control, but when I went back and looked at the joystick sequence numbers in the telemetry (glad I saved those off!) I found that the computer had been getting constant updates from the laptop just like it should, but the joystick values just weren’t changing. Either there was a real hardware problem with the joystick, or WindowsME was fucking up. Considering all the problems I have had with this laptop, I am leaning towards WindowsME. I will probably put XP on it soon, but I am giving some strong consideration to making a stripped down linux laptop for the remote piloting. I don’t trust X windows any more than ME for real time applications (less, actually), but I might consider coding the remote pilot application directly to fbdev for display purposes.

 

 

What went well:

 

The frame collapsed just about how you would want, absorbing a whole lot of energy.

 

The attitude engine restart while airborne worked perfectly. After our last crash, I added code to let the computer continue tracking the gyros even when it isn’t active, and when it is reactivated, if it determines it is airborne by seeing that the gravity vector isn’t (roughly) unity, it leaves the axis where they are instead of reinitializing them.

 

We had added a manual ball valve to the fill plumbing on the vehicle, which allowed us to vent most of the pressure out of the tank before firing the remainder through the engines. Our fill-cart based switchbox for manually firing the engines also worked well. We might want to add another valve to the vehicle to allow us to vent peroxide without going through the engines.

 

Nothing broke loose in the electronics box.

 

 

What went poorly:

 

We lost the computer on impact again, and it wouldn’t come back on a power cycle. When I got it home and pulled everything apart, I found that a couple pins on the bottom PC104 connector had been bent together. After I straightened them out, everything came back to life. I’m still not exactly sure how the could have been bent, because there were standoffs on the bottom that should have prevented them from hitting anything. There was probably sufficient flex in the box at the time of impact to do it, so I have swapped the standoffs out for taller ones.

 

The utter failure of the tethers was a rude shock. The chains are pretty heavy, but this was a classic case of the “weakest link”. One side ripped off the thin bar of 4130 on the landing leg that we had it attached to, and the other side broke the aluminum quick link that we had the chain connected back to itself with. We will add reinforced mounting loops directly to the frame on the rebuild, and if we continue to use chains, use only steel quick links. We may move to nylon rope for the tethers, which would let them take up the shock much more gently. We couldn’t do that on the previous vehicle because the engines were right next to the attach points, but with them in the current locations it would probably be ok. There is still a worry about flying back over them on the ground.

 

The main motor bell hit the ground hard enough to break the bolts securing it to the frame, which pushed all the plumbing up high enough to crunch the burst disk holder through the bottom of the seat by an inch or so. We plan on modifying our main distribution manifold so we can get rid of a cross fitting and a union fitting above the manifold, which will clear up enough room so that it can’t hit the seat even if the frame is sitting completely on the ground.

 

 

The rest of the damage was more or less as expected. A few tubes broke on the frame. One attitude engine hit hard enough to break the expansion cone off. One solenoid had a wire ripped off. The ball valve motor broke off the valve. A couple connectors in the electronics box were bent.

 

The basic vehicle design seems pretty sound, and won’t change much. Bob wants to upsize two of the crossmembers, and we will add really good tether points and shorten the plumbing under the seat, but mostly it is going to come back together the same way it was. I have ordered some replacement valves and solenoids, and Bob thinks he can have the frame re-built within a week or two. We may wind up having the altimeter and computer control of the throttle before we fly again, which would allow the flight computer to auto-land if the remote pilot computer loses its mind in some way.

 

 

Master To-Do list (created before the crash…)

 

Misc stuff:

visible strobe on vehicle for syncing telemetry to video

ground penetrator drop tests

side seating drop tests

Hydraulic press gauge adapter

More shelves for shop

AST introduction letter

Representative introduction letter

New catalyst pack material

Get new small engine retaining plate material and water jet cut to size

combined spreading plate / gaskets for motors

Combined solid state relay / motor drive board

Combined 5v / 12v / 15v power supply board with current sense output

1000 mW 802.11b integration

Altimeter integration

 

Computer stuff:

Auto-land from altitude computer simulation

Logging of all inertial updates

New real-time graphs of all flight parameters

graph average duty cycle for each async engine

Optional real time smoothing of any data channels

Flight sequencer for automated flights

convert GPS lat/long to meters from start point

Investigate throttle dead band at 0.5 on linux joystick driver

investigate using a large disk-on-chip instead of the IDE-flash board

Get four port serial expansion board working on linux

 

Rotary rocket stuff:

Side engines

Tap other end of shaft

Ball bearing mount on old vehicle for initial spin tests

Adapt tachometer sensor for computer logging

Tapered roller bearing housings for prop

Order prop

Rotating hub seal

 

Hybrid motor testing stuff:

catalyst chamber

grain casing

aluminum nozzle for meltdown testing

molybdenum nozzles

moly nozzle coatings

 

test range stuff:

waiver

trench / berm

launch pad

water barrels

Transportation jigs

order more 5 gallon containers

electric or air pump for drum

 

composite fabrication stuff:

bulkhead fabrication

pressure testing

hatch fabrication

triple seat fabrication

tank bladder testing

 

ballistic parachute stuff:

shipping from CA

conversion to HPR motors

computer ignition

integrated continuity tester

vehicle mounting