DragonFire2
(Modified PML Endeavour)

DragonFire2 - (2007, 2008)
I got tired of chasing this rocket for miles, so I decided to modify it to use flight computer controlled dual parachute deployment. I then volunteered it as a test vehicle for some of the new technologies that our design team is developing for the Super-ARLISS project.

Specifications:

Diameter: 3.9"
Length: 80"
Mass: 3.66 kg (8 lbs) without motor

Flight Computers: Two G-Wiz MC2s
Status: Built, but not yet flown in its latest configuration.

The primary modifications to the rocket are:

• Flight computer controlled dual parachute deployment.
• Uses non-pyro mechanisms to deploy the drogue parachute (Rouse-Tech CD3) and the main parachute (using a device discussed below).
• The lower airframe was redesigned to be very difficult to zipper.
• The rocket still has a good size payload compartment.
• The rocket is a test vehicle for our use of Li-Ion Polymer batteries. Redundant flight computers have been installed for this purpose. The primary flight computer uses 7.4V Li-Po batteries for both electronics and pyro power. The back-up flight computer uses 9V Alkaline batteries.
• The rocket will be a test vehicle for the Main Parachute Deployment Device being designed for the Super-ARLISS project.
• The parachute compartment was lengthened by about 9" to support he dual deployment tests.

The Endeavour comes with a very large payload compartment. So I divided the payload compartment into a 6.625" payload compartment and a 10" avionics compartment. There are now three sections to the rocket, the nosecone, the upper airframe, and the lower airframe.

The upper airframe houses the payload compartment, avionics compartment, and parachute compartment. A 3.5" band was cut from the bottom of the original payload compartment and glued to the middle of the avionics coupler. The avionics power on/off switches and the continuity test polling LEDs are mounted on this band. The avionics unit is designed to support four power sources, an electronics 9V Alkaline battery, two pyro 9V Alkaline batteries (wired in parallel), a 7.4V Li-Po electronics battery, and a 7.4V Li-Po pyro battery. The batteries are connected through two multi-pole switches, one switch controls the Alkaline batteries, the other controls the LiPo batteries. The G-Wiz MC2 constantly continuity checks the attached pyro elements. The continuity check of the Pyro0 port will flash the LED associated with its MC2. The way that the LED is wired, both electronics and pyro power must be supplied to the computer for it to flash.

The avionics compartment was adapted to hold two G-Wiz MC2 flight computers. We wanted both computers to be of the recording type so that we could analyze the flight logs and determine which flight computer triggered specific pyro events.The avionics compartment consists of a 10.5" coupler tube with stepped bulkheads on each side. Two 1/4 x 20 threaded rods are run through the two bulkheads to secure the three components into a single unit. Inside the avionics coupler, the electronics are mounted on a platform (a 1/8" piece of G-10) that is attached to the threaded rods using nylon cable ties.

We decided to try eliminate the use of black powder for parachute deployment. The bottom avionics bulkhead is fitted with a Rouse-Tech CD3 CO2 ejection device. the CD3 is be used to separate the upper and lower airframes and deploy the drogue parachute. The Main Parachute Release Device is also mounted on this bulkhead. This device and recovery scenario is discussed in detail below.

The lower airframe is dedicated to housing the motor. The original Endeavour body tube was cut just at the height needed to support a 54/2560 motor, the Quick-Switch adapter ring on the forward part of the motor tube was drilled out so that the this motor could slide all the way into the airframe, and the parachute ejection piston was removed. A coupler with bulkhead was then added to become the upper part of the lower airframe. When the 54/2560 motor is installed, the top of its extended foreword closure resides just below the coupler bulkhead. Mounting the coupler (with bulkhead) at the top of the lower airframe makes the design zipper resistant.

The Recovery Scenario:

When DragonFire2 reaches the apogee of its flight, one of the redundant flight computers will fire the Rouse-Tech CD3 ejection device. The CD3 will pressurize the parachute compartment, break the shear pins that hold the booster section (lower airframe) to the payload section (upper airframe plus nosecone), separate the sections, and deploy the drogue parachute. As the drogue inflates, it will begin to retard the velocity of the payload compartment without having to reverse its orientation; therefore there should be no danger of zippering the payload section. The braking force the drogue applies to the booster will make it reverse its orientation, putting it at risk for zippering.

The drogue parachute is attached to a knotted loop on the tether between the booster and payload sections. The loop is positioned such that the booster section will hang below the payload section as they descend. The expected descent rate using the drogue parachute is expected to be about 125 feet/sec.

At a programmed altitude, the flight computers will trigger deployment of the main parachute. During main parachute deployment, the drogue parachute will pull the main parachute bag out of the payload compartment and then pull the main parachute out of its bag. Once the main parachute inflates, the booster will hang further down below the payload compartment. The expected descent rate using the main parachute is expected to be about 20 feet/sec.

Several problems needed to be solved to execute this scenario:

• The booster needed to be made zipper resistant.
• The CD3 gasses needed to separate the rocket and deploy the drogue parachute without deploying the main parachute.
• The main parachute needed to be deployed at a specified altitude out of an open chamber.
• The connection sequence and positions of items attached to the parachute tether needed to support this recovery scenario.

Making the Booster Zipper Resistant

To prevent zippering, the parachute tether runs through a small hole in the center of a 1/2" plywood bulkhead that us mounted at the top of the booster section. The bulkhead not only prevents the booster from zippering, but also acts as a gas seal at the bottom of the payload compartment that prevents ejection gasses from escaping into the booster section. The small hole through which the tether passes is sealed with masking tape from inside the parachute compartment after the tether is run through it.

Deploying the Drogue Without Deploying the Main Parachute

To solve this problem we placed the main parachute inside a deployment bag and provided a path for the ejection gasses to bypass it. A one-foot long phenolic tube is attached to the knurled nut at the base of the CD3 unit, with a hose clamp. This tube runs along the side of the parachute compartment and the main parachute bag is positioned adjacent to it. The drogue parachute and most of the parachute tether cord bundle is positioned just beyond the end of this tube. This tube provides a low resistance path for the ejection gasses to bypass the main parachute and deploy the drogue. The top of the main parachute bag is also tied to a shackle on the avionics bulkhead that prevents it from being deployed.

Pictures showing the CD3 gas bypass tube. This phenolic tube provides a low resistance path for ejection gasses to bypass the main parachute. The main parachute is packed in a deployment bag and positioned to the side of the bypass tube. (Click on a picture to enlarge it.)

Deploying the Main Parachute From an Open Chamber

The shackle mentioned above prevents the main parachute bag from deploying. So, at a programmed altitude, the shackle must release the bag and permit the drogue parachute to deploy the main parachute. Devices exist that can accomplish this using a black powder charge. Since we didn't want to use black powder, we needed to invent something new. We invented the Non-Pyro Main Parachute Release Device to solve this problem. (Click on this link to see the detailed description of the device.)

Once triggered, the Main Parachute Release Device is expected to deploy the main parachute in about 5 seconds. This means that the flight computers need to trigger the main parachute release about 600 feet above the desired deployment altitude.

Routing the Parachute Tether

• One end of the tether is tied to an eyebolt that is screwed into the booster motor's foreword closure.
• A long length of tether is run from the eyebolt to the first knotted loop. The drogue parachute is attached to this loop. Ideally, this tether segment is as long as all the other segments combined, so that the booster section will hang below everything else on the tether.
• A short length of tether is run from the drogue loop to a second knotted loop that is attached to the top of the main parachute's deployment bag via a threaded quick link. This segment must be long enough, so that the drogue parachute is pulled completely out of the parachute compartment and inflates a safe distance from the payload section.
• About 8 inches (the length of the deployment bag) of tether then run from the bag loop to the third knotted loop, which attaches to the shackle of the Non-Pyro Main Parachute Release Device.
• About 5-feet of tether is run from the shackle loop to a forth knotted loop. This loop is attached via a threaded quick link to the main parachute shroud lines. The distance must be long enough so that when it is stretched out, the main parachute is pulled completely out of its parachute bag.
• A short length of tether is run from the main parachute loop to the bottom avionics bulkhead eyebolt. The end of the tether is tied to this eyebolt. The length of this tether segment must be long enough, so that the main parachute is pulled completely out of the parachute compartment and inflates a safe distance from the payload section.

The above routing scheme results in a small bundle of tether near the avionics bulkhead and a large bundle of tether near the bottom of the parachute compartment. The tether bundles are held together using very weak thin rubber bands. The rubber bands help reduce parachute deployment shock and prevent the tether from touching the heating elements of the main parachute release device.

Deployment System Ground Testing

We ground tested the deployment system to try to expose any issues that could interfere with a safe and soft recovery.

Testing Drogue Deployment

Goal: Deploy drogue chute without deploying main parachute.

I assembled the rocket from the avionics compartment on down. Although the pieces went together easily, preparation took quite a long time. It probably took me an hour to prep the rocket for this test. There were just a lot of steps and I double and triple checked each step.

I had planned to use a MC2 fight computer to fire the apogee charge to split the rocket and deploy the drogue, but I could not fit the G-Wiz USB adapter into the closed avionics compartment. The compartment had to be closed to seal the CD-3 gas in. So instead, I wired the CD3 e-match to the battery through one of switches. I turned on the switch. Pop, the e-match fired, the rocket separated, the drogue deployed, the main parachute did not.

Drogue deployment video (Quicktime) (Click here for Windows Media Video)

A perfect result. After the drogue was deployed, I tried pulling hard several times in the parachute tether. The main parachute bag moved a bit beyond the end of the gas bypass tube, but was still well within the parachute compartment.

Ten minutes passed as I cleaned up and reviewed the film of the previous test. As I was about to prepare for the next test, I noticed that the Li-Po batteries were very warm. During our battery characterization tests they never got warm, not even for the high current test. Then, I noticed that the drogue parachute ejection switch was still on. Apparently, when the e-match fired, the e-match circuit remained closed. I had been discharging the Li-Po battery through a short circuit for 10 minutes. Upon opening the avionics compartment, it was obvious that the circuit used to ignite the CD-3 was fried. I had used a piece of e-match twin conductor wire to make the temporary circuit to the CD-3. Just about all of the insulation from these wires were gone. They had gotten so hot, that they damaged some of the wires that were nearby. The Li-Po battery connector fused to the feed through pins to which it was attached. The permanent wires (twisted pair of stranded wire) that fed to the switch had also gotten hot, the insulation is all stuck together. The battery's paper label is now gray instead of white.

Toasted wires near the CD3	Short in e-match holder	Insulation melted off temporary wires
Insulation damage on wiring	Toasted Li-Po battery	Click on an image to enlarge it.

This test showed that:

• The CD3 easily overcame the shear pins and pressure loss from vent holes to provide good separation of the booster and payload sections.
• The drogue parachute was deployed about half way between the booster and payload sections.
• The main parachute deployment is not likely to caused by these events.
• E-match terminals can form a short circuit after the match fires.
• More care must be taken when powering experimental devices from high energy batteries.

Testing Main Parachute Deployment

The damage from the meltdown was repaired and the release of the main parachute was tested. Two separate tests were conducted, one tested the flight computer's ability to fire the Main Parachute Release Device, and the other tested the ease of deployment of the bag and parachute from the bag.

To test the ability of the Dragonfire2 flight computers to fire the release device, the avionics compartment was opened and a flight computer was attached to a notebook computer via a USB cable. In this configuration, a user interface can be used on the notebook to turn on and off flight computer pyro events. Power was turned on to the Li-Po powered MC2 flight computer. About a pound of force was exerted, pulling on the shackle tether segment, and the flight computer was instructed to activate the release device.

Main parachute release video (Quicktime) (Click here for Windows Media Video)

This test showed that:

• The shackle was freed within 5 seconds of the device being turned on.
• The shackle was easily separated from the mounting plate.
• The device remained on for 11 seconds without bursting into flame. The smoke was generated by decomposition of the enamel insulation on the resistance wire.
• None of the flight computer components significantly changed temperature. (The components were not stressed by firing the device.)
• The Li-Po batteries did not have a detectable change in temperature. (The batteries were not stressed by firing the device.)
• The resistances of the heating elements were in a range that the flight computer detected them and reported "good circuit continuity" on the pyro ports to which the elements were attached.

To test the ease, with which the main parachute is deployed after the shackle is released, the parachute compartment of rocket was prepared for flight except that the shackle was not tied in place. The tether was pulled and the force needed to pull the deployment bag out of the rocket and to pull the main parachute out of the bag was examined. Both of these goals were accomplished without even applying a pound of force. They slid out so easily, that the test was judged to be successful without the requirement of making a precise force measurement. Note that the parachute compartment was lengthened several inches so that the packed parachute bag could fit into the compartment beside the vent tube without being compressed in the vertical direction.

This test showed that:

• The drogue parachute should easily pull out and deploy the main parachute once shackle is released.

Flight Computer Programming

Redundant G-Wiz MC2 flight computers are being flown. The primary flight computer is powered by 7.4V 800mAh (10C) Li-Ion/Polymer batteries; one battery for electronics power, one battery for pyro power. The back-up flight computer electronics is powered by a 9V Duracell Alkaline battery. The back-up flight computer's pyro events are powered by two 9V Duracell Alkaline batteries wired in parallel.

The primary computers events have been set to occur ahead of the back-up computers events. We hope that this will make it easy to determine which computer triggered specific events. Note that we expect the main parachute to deploy about 5 seconds after the Pyro 3 event is triggered.

The flight is planned to use a J315-R motor and reach an altitude of about 3000 ft. We hope to be able to see and film all events.

Flight day, the April 19th 2008, TCC launch

I prepared the rocket for launch and was about bring it to the RSO for inspection. But, first I wanted to activate the electronics and verify that the flight computers were both receiving power and that they were detecting connectivity to the e-matches and heating elements. First, I turned on the back-up (alkaline battery powered) flight computer. POP! The CD3 immediately fired.

Deployment was perfect. The rocket sections separated well. The drogue deployed. The main parachute remained inside the parachute compartment. However, the CD3 was not supposed to fire. Something was wrong with the electronics. I carefully packed up the rocket, disturbing things as little as possible so that the failure analysis that I would conduct back in my lab had a good chance of uncovering the problem.
That failure analysis showed that an N-channel power FET on a pyro port 1 had shorted. The most likely cause for this was my handling of the electronics without proper static protection. The flight computers were installed in the avionics compartment through much of the design, assembly, and test process. It also exposed a hole in our test cases. None of the tests that we performed would have detected damage to these transistors.

A replacement part has been ordered and we now expect to be ready to try again at the Dairy-Aire launch.

Flight day, the May 17th 2008, Dairy-Aire TCC launch

The launch (and descent) went well. I launched DragonFire2 using a J315R that should have apogeed at between 2500 and 3000 feet. I selected this motor because it would allow us to see the entire flight. Just after apogee, the drogue parachute deployed. About five seconds later, the main parachute deployed.

<---- DragonFire2 taking off The recovered avionics section with ----> one fishing line segment. (Click on a picture to enlarge it.) Primary and Backup flight computer graphs.

Analysis of the flight computer data showed that both flight computers fired the Main Parachute Release Device at apogee rather than the desired 1500 and 1000 feet. We believe that this is due to a mismatch between the flight computer's firmware and software. This problem resulted in the main parachute deploying 1000 feet higher then we planned. The net effect was that the rocket drifted into an adjacent field and was difficult to find. The release device however worked exactly as it was designed to work. The delay between firing the pyro port and the measured effect of the main parachute slowing the descent rate was 9 seconds.

From the time the flight computers were turned on until they were turned off, a little more than 2 hours elapsed. The final voltage readings on the Li-Po batteries were Pyro Battery 7.7 volts and CPU Battery 7.8 volts. These values reasonably match the characterized data.

Conclusions:

Both the release device and the Lithium-Ion/Poly batteries worked exactly as they were supposed to work. SUCCESS!

Detailed Drawings:

Airframe Drawing	Main Parachute Deployment Device	Avionics Schematic
Bottom Bulkhead	Top Bulkhead

Gallery:

Assembled	Upper & Lower Airframes	Sections
The above pictures are taken with the original 15" parachute compartment.
Li-Po batteries mounted on the top avionics bulkhead	Assembled avionics unit	Bottom avionics bulkhead
Severed fishing line after main chute release device test	Heating element after using it to test main chute release device	Ready to fly at 4/19/08 TCC launch