September 18, 2001 Meeting Notes

 

In attendance:

 

John Carmack

Phil Eaton

Russ Blink

Bob Norwood

 

I wrote new graphing software that displays a lot more of our telemetry data, and organizes it better.

 

We started looking at GPS outputs in our logs today. The positional data drifts too much to be useful for any active control, but the lat / long velocity data seems to be stable and precise enough for dynamic use coupled with an inertial system. Unfortunately, the altitude velocity doesn’t seem to work as well, because it didn’t register much of any speed even on the fast launch we did today.

 

Russ painted the big lander tank so we won’t get carbon fiber itches as much while handling it.

 

Russ finished the layout for new motor drive board, but we don’t have it built yet.

 

Welded on new mounting brackets for the seat belt at a lower position.

 

Built new braces for the throttle and joystick platforms.

 

Added top braces between the tank supports.

 

We have two ten foot long by two foot diameter filament wound tubes on order for building our first streamlined (unmanned) vehicle. The total vehicle will weigh about 200 pounds, holding the same main propulsion system as the manned lander. The second tube is just for getting some experience machining filament wound structures. We will probably try and cut a hatch in it as practice for the next larger vehicle.

 

The big lander still needs to have one of the support legs re-tapped, so we weren’t able to fly it today. Instead, we got the little lander back out to see how it worked with the new asynchronous flight control logic. We had to wait for a big thunderstorm to blow past us, so the air was at 100% humidity, making for pretty cloudy runs.

 

Initially, we just left the engines pointing straight down, because we didn’t want to mess with canting them by jamming spacers into the engine mounts again (the big vehicle has smoothly turning roll adjustments), but it wouldn’t fly straight at all like that because it was trying hard to make a roll correction happen that never would. The async control logic doesn’t have an option to ignore roll control, so we got out some spacers and properly set the engines up.

 

The next flight went perfectly, with a couple quick maneuvers. The async control reacts amazingly well, tracking the joystick position within a quarter second, and that is only limited by the rate curve I set for adjustments. I suspect I could tighten that up significantly more if we wanted even quicker response. The small lander has the engines in closer to the centerline, which gives it less control authority, which reduces the overshoots with the async control method. The big lander should be much smoother when we try flying it in the short form again.

 

http://media.armadilloaerospace.com/2001_09_18/hop5.gif

 

We loaded three liters for the final flight. I intended to fly it up about thirty feet, higher than we have taken any of the vehicles. It popped right up there, and was easy to control coming back down. If I had known it was going to fly so well, I would have hovered it at the top for a little while.

 

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

http://media.armadilloaerospace.com/2001_09_18/hop6.gif

 

The async control is clearly superior on both vehicles, so I am going to remove the old PWM support completely. The back and forth roughness is the only drawback, but the perfect control more than makes up for it. I can probably make a predictive software model for the vehicle response and make it a lot smoother, but that would require some vehicle and load specific tuning. Removing hardware latency would make things better in every way, so we are looking at a couple more things that can be done there. I have inquired to KVH about getting gyros with greater than 100hz bandwidth. Ideally, we would want 1khz bandwidth, even if it was noisier. Russ and I briefly talked about making attitude engines specifically to reduce latency in thrust generation, even if they were less efficient or rougher. We could tap some extra holes and cut an O ring groove in the bottom of a solenoid and create an engine that bolted directly on the solenoid without a fitting in between, and it might be better to have catalyst right at the exit of the fitting, without any kind of a spreading plate.

 

We have a few things we need to get before this Saturday:

 

Bob needs to fix the support bar that galled.

Russ or Phil needs to finish the new motor drive board.

Our new peroxide shipment needs to arrive, and Phil and I need to go get ten gallons.

 

The plan is to do a ballasted hop with ten liters, then immediately load it with another ten liters and have our first passenger climb on. Each hop will only be about six seconds. If everything has gone well, we will probably load up a full five gallons for a longer manned flight.