This summer, a whole bunch of ideas I've had on a number of rocket-related topics all came together in a single rocket, which I call my C40 Booster (so named because it launches on a cluster of A10-3Ts). So, this article ends up with:
- My review of the OpenRocket model rocket design/simulation package (with the design of the C40).
- A biaxial ejection baffle.
- Using a butane lighter "spy" camera on a model rocket.
- The World's Most Over-Engineered Igniter Whip.
- First Flight: August 7, 2010
So, on to the content!
I've always had an interest in computer-aided design (CAD) systems in general, and model rocket CAD in particular -- in fact, a long time ago when I was assigned to teach a software engineering course, the term project was a model rocket CAD system (we didn't quite get it done; I've still got the barnacle-encrusted code around somewhere). Up until recently, the only model rocket CAD packages I've seen have either been closed-source Windows programs for money, or very limited toy systems. Probably the best-known of the prior category is RockSim, available from Apogee Components (see http://www.apogeerockets.com/rocksim.asp); by all accounts an excellent package, but I'm not going to reboot into Windows to do something that's supposed to be fun! My previous designs have been made using xfig -- the single best general-purpose drawing package I've ever seen, but with precisely zero support specifically for rockets.
Early this summer I came across the OpenRocket project (see http://openrocket.sourceforge.net/). This is an open-source (and, in fact, Free as defined by the Free Software Foundation; it's released under Version 3 of the GPL) application written in Java, so it's cross-platform. The program started out as Sampo Niskanen's MS Thesis at the Helsinki University of Technology, and has since been supported by a community at SourceForge.
Download and Execution
The package can be downloaded from http://openrocket.sourceforge.net/download.html There is a link on this page to download the .jar file that contains the entire program (there are also links to download the source code, or the current project repository). There is no installation; you simply execute the .jar file. Alternatively, it has also been packaged for Debian GNU/Linux -- Debian users can simply install it using their favorite package manager. Curiously, it isn't available for Ubuntu. I didn't have any problem copying the .deb file over to my laptop and installing it, however.
Here's a screen shot showing the C40 design:
(I've rescaled the image to better fit web browsers)The screen is broken into three sections: an area to the upper left showing the logical structure of the design: an area to the upper right with components to insert; and a lower area showing the whole rocket. Let's take a look at the top left area.
This window shows the structure of the rocket design: the rocket has a nose cone, followed by three body tubes. The first body tube also has a tube coupler and a bulkhead (we'll talk about the two "launch lugs" in a moment); the second body tube contains a parachute and has a launch lug. The third section is pretty complicated: the two tube couplers (and their contents) are the biaxial ejection baffle; the centering ring and inner tubes are the motor mount.
You can see how this relates to the bottom area by looking back at the previous image. You can see it can handle quite a complex design: it handles tubes located entirely internal to a body tube, offset from the centerline of the rocket. It handles clusters of motors. While this rocket is roughly radially symmetric, OpenRocket would also be able to handle rockets with non-identical fins.
Something it can't really handle is multiple body tubes. The intent of the C40 is to tape cameras to the outside of the rocket; there is no good way to represent these. I represented the cameras as jumbo-size launch lugs roughly the same diameter as the greater diameter of the cameras; I have no idea how accurately this was reflected in the simulations. On the one hand the rocket flew (more about that later); on the other, my fins are vastly larger than the program's stability analysis estimates are necessary -- the program estimated the rocket as severely over-stable at just over three calibers.
To summarize designing a rocket with OpenRocket: I found it to be very easy to use, and it resulted in a rocket that flew.
The other feature provided by OpenRocket is simulations of the resulting design. The program has a large number of RockSim thrust curves, and will run simulations of the design. Here's a simulation of the C40, with a cluster of A10 motors, but no ejection charge:
(the results are also presented as a numerical table, and can be exported for manipulation in a spreadsheet). As can be seen, an optimal ejection charge would be approximately 4 seconds. It's unfortunate that Estes's' only A10 motor has a three second delay! We'll return to this later, when the rocket flies.
I haven't conducted any tests to verify or validate the simulation results. The program claims to use "extended Barrowman" equations, and a Runge-Kutte simulation.
OpenRocket is a very cool project. The current state of the program is very useable; I anticipate using it for my designs for the foreseeable future.
Oh, one last thing. Here's the C40 design, in OpenRocket's format (of course, to actually look at the design you'll need to try out OpenRocket!): http://flare-rocketry.com/downloads/c40.ork.
The Biaxial Baffle
Note: this is a simple enough idea that I'm sure it's been invented dozens if not hundreds of times before I did. If somebody knows who did it first, I'll be happy to give them appropriate credit.
I don't like ejection wadding. My experience has been that there's always some gasses and burning gunpowder fragments that make it past the wadding, and my parachutes always accumulate innumerable holes and singes. While "dog barf" works better than the standard treated toilet paper, neither seem to work terribly well. Besides, it just seems really inelegant.
So, I've always had an interest in better ejection systems. I've tried pistons with some success, but I've occasionally gotten the parachute caught between the piston and the body tube, or had the piston just get caught; in either case, the results are disastrous.
So, I've been looking in to baffle systems. This article is far too long already, so rather than provide a survey of prior work I'll just go right to what I call a "biaxial baffle".
Here's a cutaway drawing showing the baffle (no, this wasn't rendered by OpenRocket -- that program doesn't do 3D renderings. I drew it in xfig):
It's made out of a tubing coupler, two bulkheads, and two tubes slightly less than half the diameter of the tubing coupler. The tubes are cut to roughly 2/3 the length of the coupler, and holes to fit them are drilled in the bulkheads. One tube is glued in each bulkhead, and the whole thing is glued together. Then it can be inserted as a unit into the body tube of the rocket. An eyelet can be screwed into the upper bulkhead to serve as a shock cord mount.
In the case of the C40, the body tube is BT55, and the inner tubes are BT5. Because this rocket was going to have four ejection charges going off and filling the lower part of the body tube with heat, I lined the body tube from the motor mount to the baffle with another tubing coupler, and also spray-painted painted all the exposed surfaces with barbecue paint. I actually attached the forward bulkhead to the forward inner tube and forward tubing coupler, and the aft bulkhead to the aft inner tube and aft tubing coupler as shown in the following figure:
Both assemblies were then glued into the body tube. Also, since the rocket's payload wasn't terribly aerodynamic to start with, I cut the body tube half-way up the forward bulkhead, so I would be able to open it there and inspect the baffle. I put some shipping tape on the body tube after painting, and hold it together with masking tape for flight.
A Spy Camera in the Sky
I'd been thinking about a 4 x A10 cluster for years, but the catalyst for building it was seeing several participants at NSL2010 with various on-board video cameras. Talking to some of them, it turned out that many of them had been purchased for very little money from on-line vendors. Shortly after returning, I spent a few minutes wandering around Ebay and eventually found Item Number 160440491847. If you go to this listing, you will find it closed (of course), but it'll bring up a page of USB lighter cameras to choose from. I bought mine from an Ebay online store called welcometoauction2003 (note: my only association with this store is this single purchase)
The camera shipped from Hong Kong within a day or so, and arrived after a couple of weeks. It was exactly as described in the listing: a USB camera cleverly disguised as a disposable butane lighter. There were two limitations, which had both been clearly described in the listing: it comes without memory (you need to provide your own Micro-SD card), and it doesn't actually function as a lighter (awwww).
Update: I'm afraid the spycam is built about as well as you'd expect for the price. I got one successful launch out of it; then the main control switch broke (physically broke: I took the case off the camera and it was in fragments). I soldered a switch across the most obvious connections -- I should mention here that it's all surface-mount -- and it didn't fix it. For the price, it isn't worth continuing to mess with it! Going back to ebay this particular camera seems to have disappeared; there are a number of other tiny cameras out there, and I'll be trying one of them some time...
The documentation reads like someone who'd never actually seen the camera tried to describe its operation in Cantonese and then passed it through BabelFish to translate it into English. A representative sample is:
Long press power button, red indicator light begins to bright, now it enters into standby mode( red light flashing fast), D019 will automatic video recording When the voice more than 60 decibels, now red light is turn off (indicate videoing). Press “power “button save the file, red light bright ,now it enter to Video/photo standby mode.
Apparently it has a variety of operating modes, including various sound-activated and still-picture modes, almost none of which I've explored. The LED mentioned is completely invisible in anything but complete darkness, since it's inside the body of the camera (the photo of the camera above was taken from the item listing. The arrow pointing to the "LCD" is indicative of the quality of the documentation). I used a knife to scrape the plastic camera body thin enough to be able to see the LED in indoor light.
What I eventually determined through trial-and-error was that if I put it on my laptop's USB and then took it off again, it entered a mode in which if I press down on the "lighter" (the power switch) for ten or fifteen seconds, it would record from then until I pushed the button again. Good, that's what I want!
For flight, I taped it to one side of the C40, and taped a real butane lighter to the other side to counter its weight and drag. Now that I've seen that it works (see below!), I'll be buying another one in the near future so I can have cameras facing both directions.
The World's Most Over-Engineered Igniter Whip
There were also several cluster rockets at NSL2010, with a variety of approaches to igniting them. Most either twisted the igniters of several motors together, or had some sort of cat o' nine tails arrangement with wires connecting several micro-clips, and the launch system clips connected to this igniter whip.
While it worked (I don't remember seeing any failures caused by the ignition systems) it seemed to me like there were several flaws in the system:
- It wasn't practical to individually check the continuity of the igniters. Typically a participant would connect one lead to each of his igniters, and then try the other lead on each igniter in turn, and then connect the second leads on all the igniters. Of course, this meant that all the previously-checked igniters were disturbed to a greater or less extent. Occasionally one would visibly move, and the participant would go back and recheck that one.
- I was always a little concerned that one of the motors might start a little early, and when spitting out its igniter possibly dislodge one or more of its neighbors' igniters. It seemed like this could lead to an ignition failure.
- I was also concerned that one of the motors might start a little late, after the rocket had started to move. Again, it seemed like this could lead to the igniter being dislodged and the motor failing to light.
So, I set out to build an igniter whip that would not suffer from these deficiencies. While designing it, I added one more requirement::
- The whip should plug directly into our launch box, rather than being connected with the pad's micro-clips.
With that in mind, I designed the control box you can see in the following photo:
There's very little to say about it that isn't clear from the photo. The switches, plastic box, wires, and micro-clips are all from Radio Shack. The aluminum channel support rod and C-clamp are from Lowes.
The box is at roughly half the height of a launch rod, the idea being that with the rocket on the pad there should be little or no weight trying to pull the igniters out of the motors, but when the rocket is near the top of the rod the igniter will get pulled out of any still-unfired motors.
The C-clamp is drilled and tapped to screw the support to it. Note that this requires a somewhat "heftier" blast deflector than a standard Estes launch pad; the club rack has deflectors cut from surplus road signs, and my own deflector is also much heavier duty.
The C40 had its maiden voyage at the regular club launch on August 7, 2010 and was without incident. In spite of the simulation results reported above I elected to launch on the 4xA10 motors; I was concerned about the weight of the lighter and camera and wanted to be sure to have lots of velocity early. Connecting the igniter whip to the deflector, and to the launch control box, was no problem. The continuity check worked exactly as planned: I was able to turn on the power to each igniter and test its continuity individually, then turn on all four igniters and arm the control box. As could be expected from the simulation results, the ejection charge was very early. One disappointment was that the parachute showed some very minor scorching; enough to prevent full opening but not enough to cause damage on landing.
I've put the video up on youtube:
As you can see on the video, both video and sound quality were excellent. The video has been edited to remove dead time both before and after the flight; the beep you can hear at the start of the video is the continuity check at the range box. One surprise is that the four ejection charges were clearly audible from the ground, while only one and may two (I'm not quite sure whether I'm hearing a second one) can be heard on the video. There is supposed to be a way to set the time stamp on the camera to better reflect reality, but I didn't experiment with it.
A few of the frames are of particular interest. Here's a frame at ignition Unfortunately you can't really see that the motors all ignited (they did), but you can clearly see an igniter plug being ejected.
Also, here's one showing the ejection. Something that surprises me about this one is that the payload has separated from the booster by quite a bit, but the parachute is still inside. I wonder if a looser wrap would help it eject faster, and protect the chute from blowby better?
The rocket's next flight will be with 4xA3-4; openrocket's simulations indicate that this will result in parachute deployment very slightly after apogee.