There are a pair of additional clips on Discovery's website. One shows another fight with Dan - this time blowing up at Joe with Andrew attempting to mediate. The other shows Mark's flawed analysis of what went wrong.
I was captain of the red team with Corey, Andrew, Dan, and Joe. We had 5 days and $20,000.
Anybody that wants to think about their own ideas for the challenge might be interested in knowing the detailed rules, which were given to us after the blueprint challenge and team selection. Here's the ones that I remember:
The judges selected my net-gun concept as runner up in the blue print challenge, so I was team captain with second pick. For the most part, both me and Gui (my counterpart) selected the same teams we had for the first challenge. The exception to this is that Gui chose Amy instead of Joe, who had already been eliminated. Keep in mind that we did not see what went on in the other team's shop so it made sense to go with the people we knew.
People keep asking me why I picked Dan to be on my team. I had no problem with Dan after the first challenge. He got excited and loud sometimes, but it didn't bother me. I think what you see in episode one was exaggerated in editing to play more smoothly into what happened on the second challenge. When I said he was the weakest link during eliminations, I meant exactly what I said: I thought everybody on my team was great, but if I was forced to pick somebody as the weakest, I felt he was just slightly less awesome than the rest. Since I thought Dan was the weakest member of a strong team, I picked him after the other members of the team were already selected.
What I didn't realize was that he thought this was part of some sort of conspiracy to get him eliminated rather than a genuine assessment. This is extra ridiculous because it was already obvious to everybody that Joe was going to get eliminated, so what we said on the red team didn't matter. Anyway, he was pissed at me and Corey and maybe Andrew too, which is probably where the trouble started.
While he was probably exaggerated in the first episode, if anything it was downplayed in the second. By the end of that challenge, Dan was the least professional person I have worked with in any context for as long as I can remember by a wide margin. In addition to fighting with everybody on the team (except maybe Andrew?), he was constantly attempting to claim credit for anything that worked and figuring out who he could blame for anything that didn't. All that being said, during other challenges, while it was too late for us to be friends, there were no major problems. I believe I did a particularly bad job of handling him as team leader.
Joe was the only team member that I hadn't worked with on the previous challenge, so he was something of a wild card. The only thing I knew was that the other team did not like his leadership style. I can't comment on his leadership, but he was a great teammate. He worked tirelessly and effectively despite having just been eliminated.
After two builds, Corey, Andrew, and I were pretty tight. We worked well together and generally had fun. It may not be too obvious from watching yet, but hopefully future episodes will show why we were called The Three Amigos.
Hollow foam bomb shape with paintball paint inside. We were given exactly one (without paint) to examine meaning we had to carefully ration our destructive testing. We found it to be surprisingly durable. Many of our tests would put a dent in it or knock a piece off the wall into the center cavity without fully removing any material. It's possible that the paint filling would transfer the shock to the far wall, but without testing, we couldn't rely on that.
The team decided (with my reluctant agreement) after examining the missile that a net, even made of razor wire, would not destroy it. Unfortunately, swatting it away wouldn't count as success. We tried to come up with a secondary system: passive barbs or activated blades that would finish it once contained in the net, but we couldn't come up with anything we were confident would work and was buildable.
Our device had to be in front of the bunker and could not face upwards more than 60 degrees. Given ballistics and air drag, the projectile would be coming down at a bit steeper than 50 degrees. With less than a 10 degree crossing angle, for a shot that was going long (which we were required to defend against as well), we actually had to hit it quite high up. Without the detailed trajectory simulations (done in Excel of all things) in front of me, I don't remember exactly, but I think 60 or 100 feet up in the worst case. Given the limited time, limited budget, and most importantly zero testing opportunities, we decided early on that any sort of computer aiming was impossible. Relying on human aiming meant covering as large an area of the sky as possible.
This meant launching either a single large object or many small objects. Most large objects would have too much drag to go very far without being too massive to launch in the first place. We could overcome this by making the object lift generating or spinning (covers a large area without having to be that large). The blue team did both. If we were launching multiple smaller objects there arises the question of how small. Note that it is only twice as easy to hit a 12 inch circle with another 12 inch circle as it is to hit the original 12 inch circle with a point. This meant that if we were going to try to launch things of significant size, they had to be at least 2 feet in diameter before they would be all that much better than a tiny bullet. To this end, we thought about a central rod with spring-loaded blades that folded out as they left the barrel, but figured it'd have to be very massive for this configuration to actually destroy the missile.
My favorite large-projectile idea was a spinning shape similar to what the blue team built, but instead of having the projectile itself generate lift, we would launch it with a ballista. Unfortunately, there was too much that could go wrong that we wouldn't be able to test adequately, so we decided on a field of smaller launchers.
Lots of things could work, but a pneumatic cannon is very simple and reliable. Enough so that we could make lots of them. This approach also had the advantage of scalability, so we could more easily adjust the size of our array bigger or smaller as we were ahead or behind schedule.
Once we decided on a field of pneumatic cannons, we had to work the details to build as big a field as possible within the budget and then figure out how to aim it. Valves were the problem. They're expensive. Only so much air can flow through a small orifice before it becomes "choked" and no more will flow regardless of how much pressure you have. To minimize our flow requirements, we went with close to the smallest legal barrel diameter, somewhere around 1.75 inches. With dozens of barrels, that still adds up to a lot of air. Electric valves in the size and quantity we needed were prohibitively expensive, so we were stuck with manual ball valves.
In the episode, you see us arguing over tests (this is when I try with minimal success to break up a fight between Dan and Corey). An ordering deadline was coming up and we had to get our order in or lose most of a day. Corey's test involved a single valve (maybe 3/4") for a single barrel and worked. We had already had other successful tests, but this was the first successful test that we could afford to mass produce. Dan's test used a single, larger valve (1.5") leading into a manifold with 4 barrels coming off of it. Despite the whole argument, we were able to do this test as well, which was also successful, and get our order in close enough to on time. We ended up going with a configuration that had four 1.5" valves leading into a manifold with 10 barrels attached to it. Then we built as many of these (turned out to be 6) as we had time and money for.
We could handle long vs. short incoming missile shots simply by firing earlier or later, so we didn't have to build any hardware for that dimension of aiming. For left vs. right aiming, we put the entire thing on a carriage that sat on a pair of 20 foot rails. We could then slide the whole thing from side to side with a large electric motor as we saw which way the incoming missile was going. The other choice was to put it on a well-balanced see-saw and pivot it rather than sliding it, but this would require tracking farther to the side as the missile got closer, making it more difficult to aim.
We had 24 manual ball valves that all needed to have their handles rotated 90 degress simultaneously. After working them in, this required about 5 pounds of force at the end of the handle (which had to move about 6 inches). We built a framework linking all 24 valve handles and actuated it with 4 pneumatic cylinders mounted near the corners.
We were told we would be defending against 5 missile shots and needed a 6th set of ammunition for one test shot. To get moving on construction, we fixed our barrel size at around 1.75" before we really knew what we were going to be firing out of them. The best option was steel ball bearings, but they are expensive and we could not afford 6*65=390 of them. We found cheaper metal spheres, but they had a 1-week lead time. We thought of lots and tried a few cheaper options and settled on short lengths (about 8") of heavy chain as our cheaper option. The plan was to wait until we knew our final budget then do the simple linear algebra to get 390 shots with as many as possible being the more effective ammunition. At some point, I suggested getting some steel round stock and slicing it into short cylinders. I figured we'd have time to get maybe 20 more good shots this way, but Joe put forth a heroic effort and somehow made 200 of them, slicing them on the bandsaw and sanding the burrs off. When we go to order the ball bearings, there are only 15 in stock.
We plan to use the ball bearings for the Voodoo gun, put Joe's bullets in the middle of our grid, and chain around the perimeter. However, we are running behind schedule and production tells us we are only going to do 3 shots instead of five. This means we can forego the chain altogether and use bullets in all 60 barrels. However, after 3 misses for each team, they give us one more shot and we have almost no bullets left. We had 3 people reloading while the other two frantically ran out into the field gathering up the already-fired bullets. The shot in one of the trailer videos of me running around in a jumpsuit came from that. Then they told us we were going to take a 5th shot and we did it again, but this time we had to go farther since we'd already gotten most of the close bullets.
We did these tests in our indoor firing range, which we spent about $1500 of our budget to create. It was a big chunk of money, but the alternative was guessing, and lack of proper testing was often our biggest challenge.
We had an awful lot of welding to do on this challenge. Many of the welds had to be airtight, which meant they had to be done carefully with MIG, then gone over with TIG. Then we did soapy water tests (pour soapy water on it and look for bubbles) to find the leaks and patched them. This meant carefully managing the workflow to maximize uptime on the MIG machine. We would commonly have 2 (or even 3) people trading off the torch, one fixturing while the other welded. In retrospect, we maybe should have used some of our budget to get a second MIG machine, but we didn't think of this until too late for it to be worth the hassle and expense.
We also built a backup system, which we called the Voodoo Gun. It's in the background of several shots but never mentioned during the episode. This was mostly my responsibility as the guy with the most electrical/software/controls experience and not a particularly skilled welder.
Five more barrels in a cluster attached to a shoulder-high tripod by a universal joint. Two linear actuators turn this into a 2 degree-of-freedom turret. What makes this interesting is the control system. Andrew, our team skeet shooter, held a plywood mockup of a rifle. The mockup was mounted to a pivot and its orientation was measured by two string-potentiometers. It was pointed at a screen wired to a webcam mounted to the barrel of the gun. The result, when you point the little wooden gun to the left, the five 5-foot barrels that can each fire a 1.75" steel ball 100 meters also point to the left. Since there were only five barrels on this device, we could afford to make individual accumulators for each of them with electric valves.
I never really thought we had much chance of hitting with this (though Andrew, the operator, had more hope). On the other hand our progress on the main device was limited more by welding machines than by manpower, so it made sense to find something to do that required little welding. Additionally, we knew the judges were going to be unexcited by our massive, but fundamentally simple primary solution, so we wanted something a bit more flashy to impress them. I thought of it as something of a tie-breaker. Turned out the other team did something even flashier but less likely to work, but we didn't know that at the time.
The slew rate (how fast it could turn) was very slow because the linear actuators we got, making it even harder to aim. Unfortunately, we couldn't find any faster ones (of sufficient strength) that we could source quickly enough and fit within the portion of our budget we were willing to allocate to this secondary priority.
The original plan was to fire the 5 shots in sequence at about 0.25 second intervals. Unfortunately, my help became required for something on the primary device and we had to save time by simplifying the firing circuit to launch all at once. We were reluctant to buy a proper soldering iron, so I was doing everything with a Harbor Freight soldering gun, which was totally inappropriate for precise work like attaching wires to relays.
It was originally planned to be welded to the main frame so it had a rather narrow tripod base. However, it wouldn't have fit out the shop door if we did this, so it ended up freestanding, weighted down by excess ammunition. When we ran out of ball bearings for it and switched to the heavier bullets, it fell down from the recoil (looked silly, but didn't hurt it).
Bottom line, it didn't hit anything. It did look cool, but obviously not cooler than engineers arguing out of context or you would have seen it on TV.
We did fire metal into the air, so to some extent our device worked, but we just missed. However, a few things went wrong that would have made hitting a lot easier.
We tested our carriage (the thing that moves the entire array from side to side) in the shop and had no problems. We rented a generator to use in the field. We ended up with a smaller generator than the one we ordered (we think production went to the wrong Home Depot, where it was out of stock). When Corey tried to move the carriage to track the missile, the starting current from the motor popped the breaker on the generator. Without power, we couldn't trigger the system. If we had had the generator we ordered, the carriage and trigger would have been on separate circuits, so we still would have been able to fire. It might have even been able to handle moving the carriage. At any rate, we reset the breaker and were able to fire on subsequent tests but lost the ability to slide side to side.
Our trigger was too slow. Our bullets had already left the barrels before the valves were all the way open, which drastically decreased our muzzle velocity, making it more difficult to aim and preventing our shots from going high enough for the longer missile trajectories.
What we should have done was go with a system that Joe proposed, which would use a spring to pull the valves open quickly and would only require an actuator to pull the pin. We went with the system described above because:
Despite all these reasons in its favor, this trigger system was the mistake that hurt us. As team leader, the final say was mine, and I chose incorrectly.
In short: No.
In long: pneumatic cylinders have a very flat force-speed curve. This means the speed at which they move is nearly independent of the force on them. So long as the cylinder has enough force to move easily, further increasing the size of the cylinder will only increase the speed slightly. The limiting factor in speed is choked flow in the air entering and exiting the cylinder. For the same port size, bigger cylinders will move more slowly because it takes more air to fill them. We therefore selected the cylinder with the largest ratio of port cross section to cylinder cross section that had sufficient force to move our mechanism. This was still not enough. I can state with certainty that no cylinder in our supplier's (McMaster) catalog would have worked, and I'm reasonably confident that no other supplier had reasonably priced cylinders that we could source in 1 day that would have worked either. Commonly available pneumatics simply do not move as fast as we needed because it's not commonly required (or desired).
While faster cylinders were not available, more forceful ones were. We could have built a linkage to turn that extra force into extra speed (much like gearing a rotary system up). To move fast enough, we were probably looking at somewhere around 1-2 horsepower, which is also attainable with electric motors, but this would have required a gearbox and probably a winch to work with this configuration. Either of these solutions add enough complexity that it's no longer simpler than the alternative spring-powered setup.
Our guns had to be propped up 30 degrees from vertical. Corey led the construction of a stand to hold them up. Dan pointed out that it would be better if the whole thing was on hinges so it could come down for easier loading and safety during transport. I agreed that this would be an advantage but didn't think it was worth the extra complexity or that we would be able to lift it. After much arguing I told Dan he could make it hinged, but he was responsible for making sure it got done. While I was out of the shop for an interview, Dan cut off the stand Corey had made and installed his own. In the frenzy to get last minute details done, nobody noticed that it was completely inadequate. That's a reason, not an excuse - every one of us should have noticed, especially myself as leader.
We get to the field and try to set up. Turns out it's too heavy and we need the rigging equipment to lift it, so we won't be able to use it to speed reloading after all. Then it is deemed unsafe, probably rightly so, and we spent an hour with outside help to prop it up with some 2x6's to get it safe. The concern was that recoil would break the inadequate stand, causing the device to fall, possibly causing catastrophic (and dangerous) failure of our homemade pressure vessels. This event didn't hurt our performance, but it nearly got us disqualified and certainly hurt us in the eyes of the judges.
This challenge, both teams were afraid of the other team's devices. You can see us literally cowering in the back of our bunker when their propellor started wobbling and looked like it was going to come loose, hit the ground, and spin towards us. They, on the other hand, were afraid of our homemade pressure vessels. We had them at relatively low pressure, and Corey did a quick Finite Element Analysis to make us comfortable with them.
In a clip on Discovery's website, Mark suggests that we should have put multiple rounds in each barrel. We considered this and discarded it as a bad idea. Our main problem was muzzle velocity. Adding more mass would have decreased it further. He suggests we should have put different size projectiles in so they would spread out. They would all have the same muzzle velocity because only the lowest is getting pushed on. Drag is pretty small for dense objects at low speeds, so they wouldn't spread much. Additionally, if you put flatter discs in with length less than diameter, you have to worry about them jamming.
We made sure we met the maximum size allowable for transport from the shop to the test site, but we never thought about getting it out the door. We fit out with maybe 2 inches margin. Then it took over 3 hours to load it onto the truck - it was heavy, had no convenient lift points, and barely fit out the door. Also, we were to exhausted to be much help to the rigging crew who were in charge of transport. I think I ended up napping under a workbench on some padding leftover from our indoor firing range.
Back (Eric's TBBT Home)