Glasgow Kiss: Build

Thirty Pound Under Cutter “Glasgow Kiss” Build Report: Build

By: Andrew Smith

Photos: Andrew Smith, Pete Smith

Glasgow Kiss is Team Rolling Thunders latest creation; it is a 30 lb under cutter robot that had its first competition at the Franklin Institute in October 2015. I plan for this to be a three part article series covering the initial design, build, and post event analysis for future design changes. This second article covers the build process, and I will describe and show the major steps involved in building the robot.

In the first article, I explained the basic design principles behind the robot, most notably the circular tube frame design to resist the torsional forces that are to be expected in an undercutter robot. The finalized design can be seen in Figure 1.

The first stage of the build was constructing the aluminum tube frame. In order to do this, a semi-circular cut had to be completed in the bottom tube in order for the upper tube to “slot” inside of the lower tube. This semi-circular cut was achieved by using a 4 inch hole saw on our milling machine. This was a fairly slow process, and required a Dremel to cut out the excess material every inch or so, so the hole saw would have clearance to keep cutting. A picture of the cut in process can be seen in Figure 2, with the final cut shown in Figure 3.

With the circular cut complete, the next major cut is a vertical cut through the upper tube for the weapon axle tube to be placed in. Once again, this was completed with a hole saw in our milling machine. Thankfully, the hole saw was just long enough for us to complete the cut entirely from one side, which allowed as straight a cut as possible. This cut in progress can be seen in Figure 4.

An additional hole was cut on the underside of the upper tube to allow wiring to pass between both tubes. With the major cuts completed, final mounting holes for the drive pods, and a slot for wiring to access the weapon motor are added. With the tubes completed, the weapon motor attachment, hoop mounts, and weapon axle tube all have to be cut. The weapon motor attachment is completed by milling a square tube into a lopsided “U” shape that will align with the upper tube. When making the hoop mounts, I actually decided to alter the initial design. Originally I had planned on the angled bracket being directly welded on to the lower tube, however after reconsidering I decided to just create a mount that the angled bracket could be bolted to. This decision has two main advantages, first, it allows for different angle brackets to be tested out to ensure a proper roll over can be achieved, and second of all, it allows easy replacement of an angled bracket if it gets damaged. With all the parts cut out, a jig is made to allow the tubes to be welded to complete the chassis. The jig was made from scrap pieces of wood that are glued to a base board. The final result can be seen in Figure 5.

With the jig completed, we sent the frame to a local welder to finish up the frame. Unfortunately, when the chassis returned, the upper tube was rotated about five degrees from the proper orientation. This was partially our fault because, as we found out, the jig had a little bit too much slop in it, so it is possible that the tube had been moved to that orientation when it was dropped off at the welder. In hindsight, we realized it would have been a good idea to have the lower and upper tubes welded first, and then cut the axle tube hole out, in order to ensure a truly vertical fit. With the tube bent to that degree, it was decided that we would need to redo that section of the chassis. We decided to cut out the old axle tube, and cut a truly vertical hole with a larger diameter to ensure that our axle will be properly oriented. You can see this in progress in Figure 6.

Unfortunately, with the larger diameter hole, it also required that we have a larger diameter tube, which was quite a bit heavier. We sent the chassis back to the welder, and the final chassis can be seen in Figure 7, after it had been shot blasted clean.

With the chassis complete, we were now able to turn our attention to machining the blade. The blade was machined out of a 4” X 24” X 0.5” bar of 4140 steel from McMaster-Carr. The bar was bolted down onto a scrap aluminium bar on the table of our milling machine. The first step of the machining process was to find the exact center of the steel bar that was being machined; this was accomplished with an edge finder. With the center found, I calibrated the digital read out on our milling machine to set the exact center at 0,0. The center was then bored out with a boring head to a diameter of 1.688”, which is the required diameter for the keyless bushing to work. With the center hole milled out, a carbide tipped end-mill was used to mill-out the rest of the blade. This was a fairly tedious process as the milling had to be done at a slow feed rate, with lots of material having to be cut out. In the design phase, I intentionally left the pockets in the blade 1.5” from the end of the blade, to allow for quite large weight savings if necessary, by simply cutting off material at the ends of the blade.

With the blade finished we were able to start putting the weapon assembly together. We had to alter the weapon shaft (a 1” shoulder bolt) by cutting a keyway for the pulley, and rounding off the bottom of the bolt head to try and minimize the contact area with the ground. Two large spacer washers were machined on our lathe, one at the bottom to stop the blade from sliding off, and one in between the blade and the top of the keyless bushing, to prevent the blade from sliding up the shaft. The last section to machine was the flanges to go on either side of the pulley. These flanges were made on the lathe and should prevent the belt from ever popping off. The entire weapon assembly is shown in Figure 8; the bushings will be mounted in the weapon axle tube.

With the weapon assembly complete, most of the major machining of the bot was done, and we could start concentrating our efforts on completing the drive pods. Since most of drive pod parts were designed to be 3D printed, they went together fairly easily and quickly. We had some extra time while parts were being printed, so we made a set of UHMW gearbox mounts on our lathe. We figured it would be nice to have the option to use the UHMW mounts if the 3D printed mounts were not holding up well in combat. A picture of a completed drive pod can be seen in Figure 9.

The last mechanical part of the robot to build was the weapon motor mount. The mount was made from 0.25” thick 6061 aluminium plate. A rough cut was made using a circular saw, with all the mounting holes being made on a drill press. In order to save weight, and have better roll over characteristics, the mount was then rounded off on a belt sander. The final weapon motor mount can be seen in Figure 10.

With all the mechanical components of the bot complete, the electrical system could then be wired up. Unlike all the other bots that we have built, we decided to have two entirely separate systems. This was mostly due to the fact that the weapon motor was being run at 27.2 volts, while the drive system was being run at 13.6 volts. While it is possible to run multiple voltages on one wiring circuit, we decided that it would make the initial wiring more simple, and in an event, easier to trouble shoot any electrical problems within the bot. For the drive circuit, we made the drive pods be able to simply be replaced in a plug-n-go fashion by making the power go through a connector. This would allow a rapid repair in case of damage to one of the drive pods. In order to have two completely separate systems, we also had to have two separate receivers. We were able to bind both receivers to the same transmitter, with the drive receiver having inputs on channels one and two (elevator and aileron), and the weapon receiver having an input on channel 3 (throttle). Both systems had a Team Whyachi MS-05 switches that were then mounted to the upper tube end cap that was 3D printed. Separate LEDs were used to identify what system was live. The endcap can be seen in Figure 11, and while the design of it was honestly an afterthought, it came out to be one of the sleekest parts of the bot overall.

With the end cap completed, the robot was essentially done. At that point, I took the robot up to my local post office to use the postage scales to get an accurate weight of the bot. Much to my disappointment, we were tipping the scales at 31.5 lbs. Even though I had calculated all the weights through an excel spreadsheet, I realized that some of the choices that were made during the build of the bot had crept up on us. First of all, the thicker diameter weapon axle tube was required when we had to cut out the original axle tube, and second of all, we beefed up some of the weapon motor supports which added to the weight. I was annoyed at myself for allowing the bot to be that overweight, and for my future builds I will definitely be paying a lot closer attention to the weight estimates. With the event at the Franklin Institute only a week and a half away, Glasgow Kiss had to go on a major diet. First of all, we replaced the UHMW gearbox mounts back to the originally planned 3D printed mounts, which did save us around half a pound. The next area of focus was the blade. We ended up taking around an inch off of both ends of the blade, which also saved quite a bit of weight. The rest of the weight savings came from the holes cut in non-structurally-dependent areas of the chassis, and trimming down the width of the roll over hoop. With the event only a few days away, we managed to get the bot down to 30 lbs.

At this point we had a chance to briefly test the bot to ensure that it was fully functional. In testing, we discovered that the weapon gearing was a little too high, and I had to slowly spin up the blade, with it taking about 4 seconds to get up to full speed. In addition, we also found the Hobbyking X-Car ESCs for drive did not have a very good drag brake setting. Theoretically, if a brushless ESC has drag braking enabled, the ESC should be able to resist the wheels trying to turn when an outside force tries to turn them. In our case, that outside force was the weapon axle on the ground trying to spin the robot around. During testing, it seemed like the brushless ESCs did not even try and resist the blade spinning, and unless I was actively countering it with my controlling, the bot would move in about an 8 foot diameter circle. While we were aware of both of these problems, we didn’t have any time to try and fix them, as I was busy with school work and we had to leave in two days. I knew that I would have to focus on those issues during a fight at the competition. You can see Glasgow Kiss in its completed state in Figure 12, as it was waiting for its first fight.

While the event at The Franklin Institute is about half the size of that of Motorama, I knew it would be a great opportunity to test Glasgow Kiss out, as the 30 lb field consisted of some very tough bots. At the end of the event, Glasgow Kiss went 3-2 and had secured second place at its first competition. I was very pleased with the potential the bot showed, both in its durability and its ability to dish out some devastating hits. I ended up having to fight Zac O’Donnell’s “Triggo”, which is currently the #1 ranked 30 lber in the world, three times throughout the competition. Since most bots are lucky to survive one fight with Triggo, I was pleased in how Glasgow Kiss was able to put up a decent fight. If it wasn’t for a loose weapon motor wire, I believe we would have had a good chance at beating Triggo in the second final.

While there were many positives to Glasgow Kiss’ first event, there were also a few details about the bot that I was not satisfied with. In my next article, I will go into a bit more detail about how the Franklin Institute event went, and what modifications I am making in anticipation for Motorama. There are twenty-six 30 lb bots registered for Motorama, so I know that if I want to do well there I will have to make sure Glasgow Kiss has had all the kinks worked out.