July 11, 2004 notes

 

Electronics Failures

 

We set up for another captive hover test of the big vehicle on Tuesday.  Everything proceeded well, but as I started the flight control software up to warm the engine, the self-test failed because one of the jet vane actuators was not responding.  The automated self test on every startup now is a Very Good Thing, because I very likely would have missed this condition, as the vane position graphs aren’t on the first visible page of telemetry results.  On the driver board, two actuator bit LED’s seemed to be stuck on   We connected the ribbon cable to a brand new driver board, and found that they were still on, so the fault was in the AccessIO A/D + DIO board.  We had moved to a Diamond A/D + DIO board for the A/D, since we had fried four of the AccessIO boards (probably from the motor inductive kicks, in hindsight), but the digital IO pinout was completely different, so we had been continuing to use the AccessIO board to run the motor drives, because the DIO part seemed to still be functioning fine, even though the A/D was fried.  Apparently this wasn’t a great idea.  It is odd that it seemed to give up sometime late in the vehicle pressurization, because I had started it and let it pass self test several times during the loading process.

 

We decided to just quickly swap the board for another one (also with blown A/D) that we had sitting on a shelf, so we could use the already loaded propellant.  The replacement board seemed to operate fine, but as the engine throttled up to liftoff thrust, it started chugging badly.  We had increased the tank pressure to 300 psi to give it a little more thrust margin, which usually adds stability to an engine, but it may also have just been settling some from the test runs.  A heavily chugging engine confuses the auto throttle software, because vehicle acceleration bounces back and forth across the desired acceleration line no matter what the vehicle is doing.  This usually leads to it not continuing to change the throttle, so the vehicle wouldn’t throttle up past the 60% mark.  This left it bobbing at the end of the tether after blowing out the metal support stands (which worked perfectly after we added some T feet to keep them from toppling over too easily). 

 

http://media.armadilloaerospace.com/2004_07_11/newStands.jpg

 

While it was hanging there under thrust I shoved the joystick hard over to see how the attitude control was working at the almost-hovering point, and the vehicle rotated very rapidly.  This change in attitude seemed to give the control system just enough of a kick to let the auto throttle bring in a little more throttle and let it lift off.  It straightened back up very rapidly, and started rising.  The engine roughness also made it difficult to bring it to a hover, and I thought it might slowly bump into the lift forks, but it started descending a few inches shy.  What happened next was unexpected – the tether looped itself underneath an eye bolt above the hatch that we use for horizontal lifting of the vehicle.  We wound up with the vehicle stuck hanging there, six feet above where it started.  We lowered it back down with the lift and took stock of the situation.

 

http://media.armadilloaerospace.com/2004_07_11/snagged.mpg

 

The bolted down road plate worked well as a blast deflector.

 

The automatic self test was a very good thing to implement.

 

We needed to remove all the protruding things that the tether could hang up on.

 

The jet vanes provide lots of very rapid control authority.  The big vehicle seemed to change attitude as fast as the little one.

 

We need to reduce the open area in the engine spreading plate to bring the engine back to smooth operation.

 

We hung the vehicle from a scale for an up to date weight, and in dry hover test form, the big vehicle weighed 1650 pounds.  Add in the thin nose cone and the shock absorbers, and it is a bit over 1700 pounds, plus we pump nearly 100 pounds of nitrogen into it for pressurization (almost all of that can go away if we use helium, but that would get expensive to throw away during testing).  This is over our target weight.  Each drum of peroxide plus methanol is about 580 pounds. 

 

The test stand 12” engine had made 3700 lbf at 378 psi tank pressure, but the vehicle engine will be making less than that due to the compressed hot pack (for better warmup), the reduced spreading plate area (for better stability margin), and the jet vanes adding some drag.

 

Given that we missed both our target weight and target engine thrust, it looks like it isn’t going to be worth it to fly this vehicle configuration under the burn time waiver, because it would just be creeping off the pad with under a quarter G of acceleration.  The new plan was to modify the engine, go ahead and do the captive hover tests and the 15 second boosted hop locally as-is, then move everything over to the 48” diameter tank, which saves enough weight to let it do a decent long burn flight.  Conservative simulation shows it going to 10,000’ and lading 120 seconds after liftoff at the no-launch-license impulse limit, although it is still throwing away most of the impulse as gravity loss and landing (it will just barely hit terminal velocity in descent before powering up for landing).  We need to build bigger engines.

 

I ordered new 12” spreading plates, but they weren’t going to be delivered in time, so James took on the tedious task of manually closing up a quarter of the spreading plate holes with braze.

http://media.armadilloaerospace.com/2004_07_11/brazedUp.jpg

 

Russ decided to weld in a couple more aluminum tubes to stiffen up the landing support frame.

 

On Saturday, we got everything back together for another test.  The vehicle continued to pass the self test all day, but after warming the engine up and starting the rise-to-hover, we lost the computer.  Shit.  We were worried that the master cutoff hadn’t worked, because the engine was still making quite a bit of noise, but it turned out that the cutoff had worked fine, but it was controlling the leaking ball valve, so there was a decent trickle of propellant getting by the still-open throttle.  After the computer rebooted, I restarted the flight control software, but the self test failed with all the jet vanes in off-scale positions.  The throttle was still working fine, so the engine finally quieted down when that was shut again.

 

We decided to just set it down on the road plate, vent the tank pressure way down, and burn off the entire drum of propellant.  In hindsight, this was the wrong thing to do.  The road plate held up fine, but the exhaust backwash completely fried all of our vane actuators and caused a hot crack in one of the aluminum landing frame struts.

 

http://media.armadilloaerospace.com/2004_07_11/slagged.jpg

 

When we laid the vehicle on its side, we found that all the vanes were oriented almost perpendicular to the exhaust flow, and most of them were somewhat bent.  We may want to work out some kind of physical stop for the vanes to keep them from ever being pushed perpendicular, but it would have to cooperate with the internal limit switches.  The vanes are always fine through tests when they are in the normal operating range, but if the computer hangs and they get pushed perpendicular, they get bent.  We are probably also going to try clipping the top on the vane a bit to possibly make it self centering, but that will increase the load on the actuators.

 

When we opened the cabin, there was smoke from the motor drive board.  We had smoked transistors before when driving to a partially shorted motor, but much to Russ’s surprise, the diodes were burned out this time.  We think that the jet force on the vane pushed it hard enough to generate sufficient power to burn the diodes, and somehow upset the shared ground enough to reboot the computer.

 

We still have some snubbing components for the motors on order, but we are going to completely separate the motor controllers from the rest of the electronics anyway, both physically and electrically.  We have had to share a common ground due to the motor drive board also controlling the ignition, which has to ground into the motor and the rest of the chassis, but I am going to add some separate solid state relays to control the spark ignition from the main computer box so we can truly isolate the motors.

 

We are going to rebuild most of the electronics and do just about everything we can think of to improve reliability, because computer failures have been the biggest problem on the big vehicle.  I am going to replace the CPU board, because it seemed like it was getting easier and easier to kill it when we were investigating the inductive kick problem, and we may have permanently damaged it in a subtle way.

 

Since we have a lot of rebuilding to do, we are going to go ahead and move to the 48” tank.

 

We also worked on the layout for our new trailer, which we are going to have to outfit for the trip out to the southwest regional spaceport for our waivered flights.  We should be able to carry enough nitrogen and propellants for several waivered flights, but we also want it to be able to hold enough for a single space shot flight.  We had planned on using locally supplied nitrogen and lift trucks, but after Neil’s scouting expedition, we decided it was going to be worth the effort to be self contained.  We should be able to just pull up, unload and erect the vehicle with a truck carried A-frame, load it up and fly it.

http://media.armadilloaerospace.com/2004_07_11/trailerLayout.jpg

 

We also got in the big batch of mini-nozzles for our experimental multi-nozzle engine:

http://media.armadilloaerospace.com/2004_07_11/miniNozzle.jpg

http://media.armadilloaerospace.com/2004_07_11/manyNozzles.jpg

The support plates should be here soon.

 

A clarification – some people took a comment I made about not bothering to order something if Burt had a perfect flight the wrong way.  We are carrying on with our work no matter what happens with the X-Prize, we will just have a slightly different emphasis.