[Marobot: c'est un projet à fort potentiel et reproductible]
Il s’agit d’un projet qui secoue en se moment la communauté DIY du monde, et qui risque de bouleverser la technologie des Quadcopters et les multi rotors en général.
Les Quadcopters qu’on connait fonctionnent avec des batteries dites LiPo (Litium Polymère) qui ont des capacités qui vont de quelques centaines de mAh à plusieurs milliers de mAh (5000mAh et plus). Les Quadcopters avec ces batteries ne dépassent pas l’autonomie en vole d’une vingtaine de minutes.
Avec la possibilité d’utiliser des moteurs thermiques, la donne change. Ils auront plus d’autonomie.
Le projet Goliath a été réalisé par Peter McCloud. Il utilise un seule moteur thermique 30HP (je ne sais pas à quoi ça correspond exactement) et un système de transfert d’énergie cinétique vers les quatre hélices qu’il a conçu.
Dans le cas d’un Quadcopter on fait varier la vitesse de rotation des moteurs de chaque hélice pour l’équilibrer. Dans le cas de Goliath, il y a un seul moteur et les hélices tournent à la même vitesse. La poussée de chaque hélices est contrôlée par des vannes, même principe des avions imaginés dans le film “The Avengers”. Le débit de d’air de chaque hélice est contrôlé par des vannes, et par conséquent sa poussée.
Les hélices tournent à la même vitesse, via un mécanisme de courroies comme indiquée dans l’image après.
Le contrôl du vol et la stabilisation repose sur l’autopilote Pixhawk, modifé. Les modifications nécessaires pour fonctionner avec Goliath sont disponible en open source. Le reste de l’architecture reste pareil qu’un Quadcopter électrique. Il compte ajouter une interface WiFi et un Streaming vidéo amélioré.
Le projet est prévu être open source, à reproduire. Il n’est pas autorisé d’importer des Quadcopter, mais il est autorisé de les fabriquer. Il faut toute fois avoir l’autorisation de les faire voler. Les composants utilisés peuvent sont disponibles dans le commerce, ils sont décrit ici: http://hackaday.io/project/1230/components
Voici quelques étapes du projet (en anglais), pour plus d’info vous pouvez vous rendre sur la page du projet sur: [Source] http://hackaday.io/project/1230/logs
- 3 months ago •
Before documenting the previous steps, it seemed like a good idea to show where the project is at:
Currently, the frame is assembled from galvanized slotted steel and bolts. This method was chosen over steel tubing or composites to speed up the prototyping process. It will likely need a couple more cross members to increase the stiffness, but it’s close to being complete. The vertical shaft engine has been installed and the drive system is assembled with all of the belts, pulleys and tensioners in place. A custom exhaust system was built using go kart hardware. Also installed is a simple electrical system consisting of a small motorcycle battery just big enough for starting since the engine comes with an alternator.
The drive system has been tested by turning the engine over with the starter for a few seconds at a time. What’s not complete yet is the fuel system, propellers, ducting and controls and the electrical system will need to be completed for the controls.
The next steps are completing the propellers and adding controls for the throttle. After those items are complete, The quadcopter will be tethered a small distance off the ground and the gas engine run for the first time!
In the future we’ll add more details about what it took to get to this point in addition to updates as Goliath progresses.
3 months ago •
The Goliath design started back in November of 2013. The project goals were:
1) Build a gas powered quadcopter
2) Use parts that are easy to obtain (as much as possible)
3) Use parts that don’t require specialized skills to build (as much as possible)
4) Create a design that could be scaled up.
Prior to starting this project a number of paper design iterations were done, researching various methods of building a gas powered quadcopter. After looking at a number of power configurations (single engine, multiple engines, hybrid gas/electric and a hybrid gas pneumatic) as well as drive train options (belt, chain, gearboxes with drive shafts, etc). For the size of vehicle chosen the best power to weight option was a single vertical shaft gas engine with a belt drive. By using a vertical shaft engine, the shaft is already aligned with the propeller axes and the need for complex gearing goes away. The belt drive has the advantage that it’s lightweight, shock resistant and can be easily scaled up.
The chosen engine was simply the biggest vertical shaft gas engine I can easily order off the Internet. This ended up being a 810cc 30HP riding lawn mower engine (note: 30 HP is without air filter and exhaust). It comes with an electric starter, and an alternator. Now this engine doesn’t have as great as a power to weight ratio as a motorcycle engine (or other engines) of similar size, but it’s made to run in the vertical orientation and it doesn’t have an integrated transmission/clutch that would make incorporating it into a quadcopter more complicated.
After doing some part selections and creating an initial design in December and January. The engine was ordered and arrived in early February.
The engine documentation leaves something to be desired. It’s marketed as a direct replacement and there is no notes on the required fuel, electrical or exhaust connections. The fuel was simple enough since it’s labeled on the tubing that it’s 1/4″ and the exhaust comes with a gasket that could be measured to find the right parts. The electrical took a bit more research. There were three wires on the engine that all come to the harness. Power from the alternator coming off the regulator, the ignition coils and a fuel shutoff solenoid. After doing some searching I came across two electrical diagrams, both with different wire colors and neither matching the wire colors on this particular engine. It was time to remove the top cover and take a look inside.
Getting the cover off was a simple process. First the air filter was removed and the oil cooler was unbolted. Next the screen for the engine cooling air was removed and then there was just a few bolts on the cover. Once inside the wires were easy enough to trace back to various parts. Red was the power from the regulator, black was the ignition coils and gray went to the fuel shutoff solenoid. This will be noted on the wiring diagram that will be posted later.
With the engine in hand it was time to start building the frame around it. That’ll be covered in a later post.
3 months ago •
After picking the engine, some preliminary design work was done to estimate the total weight of Goliath and what type of structure might be required. After using some conservative estimates for the various components and assuming a steel tube frame, the total weight of the quadcopter was estimated to be 240 lbs. Heavy, but still within the capabilities of the motor. An early CAD drawing of the quadcopter with the steel tube frame is shown below.
The layout was designed to not only support the components, but to also encompass the belt system. This will protect the belts from external objects and help to keep a broken belt from flying outside of the frame.
However, after getting the engine and starting to lay out the structure, it was clear that adjustments were going to have to be made along the way. A steel tube frame wouldn’t be the easiest to modify, so a switch was made to slotted angle steel assembled using bolts, both of which can be obtained at most any hardware store and allows the frame to be adjusted as needed.
The layout started with the center beams beneath the engine. The weight of the engine required the larger 2 1/2″ x 1 1/2″ angle pieces versus the smaller 1 1/2″ x 1 1/2″ angles. The center rails are slightly skewed from front to back, (see the following picture) due to the locations of the engine mounting points.
With the top Center Beams in place, one of the Prop Supports was added. However this configuration allowed the frame to flex significantly when loaded.
More supports were needed. After trying different configurations, the layout below was used. All of the pieces around the engine are 2 1/2″ x 1 1/2″ and the Prop Support is 1 1/2″ x 1 1/2″. After applying some test loads, the flexing was found to be much less.
With the layout around the engine decided, a duplicate was built and placed underneath it to build a cage for the drive belts. Some short pieces of 1 1/2″ x 1 1/2″ angle was used to connect to top deck to the lower deck.
Finally all of the Prop Supports were added and a piece of 2 1/2″ x 1 1/2″ was added to the end of each Prop Support to mount the prop shafts to. At this point the frame the quadcopter is 7 feet long from corner to corner. The stiffness of the frame is probably less than a steel tube frame would be, but the weight of the assembled frame is about 52 lbs, about what was estimated for a steel tube frame.
More detailed list of measurements and instructions on assembling the frame will be added later.
3 months ago •
Most of this weekend was spent working on the propellers. Goliath was designed with four, two bladed propellers, each three feet in diameter. The were designed using Blade Element Theory for the specific RPM and loading Goliath is expected to experience.
The propellers will be made with foam cores covered with composites. The foam cores are created using a CNC Router which isn’t a straight forward process because the part needs to be machined on both sides. To do this, supports are needed to hold the part in place while machining the backside. This could be done using the CAM software, but more detailed supports can be built using the CAD software.
With the supports added, the CAD file was loaded in the CAM software (MeshCAM) and the tool paths were created. With the software side completed, the next step was creating the foam blanks. The foam blanks are made out of styrofoam sheets available at most any hardware store.
Three 38″ x 10″ pieces are cut and the protective coating is peeled off of both sides. It’s important to do this because the coating doesn’t cut well and can get wrapped around the router bits and ruin your part.
The three sheets are sprayed with adhesive (For this block, Loctite medium hold was used) and glued together to make one larger block. Note the label on the original sheet that says 1/2″ thick. These sheets are NOT 1/2″ thick. Even with the coating removed, the three sheets have a total height of 1.75″. This is the thickness the propeller was designed to. With the block glued together, it was mounted onto the table with 6 wood screws and was ready to cut.
A roughing pass was done using a 1/2″ straight bit with 1/4″ deep passes. Probably could have been more aggressive since the foam was easy to cut, but the process is still being refined. the results of the roughing is shown below.
With the roughing complete, a few finishing passes were made. Both an X and Y parallel pass were made and a pencil operation to clean up some of the surfaces. The last operation was to drill the mounting holes and some extra alignment holes at the ends of the propellers. All of this holes will be used to align the part when it’s flipped over to machine the other side.
The part was then removed and the alignment holes were drilled into the spoil board. The part was then flipped overand wood dowels were then used to align the part and to help hold it in place.
That was all the progress made this weekend. The next steps will be machining the backside and starting the process of adding composite reinforcements to the core.
3 months ago •
After completing the frame, the next step was building the drive system. Quadcopters traditionally use 4 individual motors with two spinning clockwise and two spinning counter clockwise to balance out the torque. Since Goliath uses a single engine, the drive system not only has to distribute power to each of the props, it also has to balance out the torque.
A belt system was chosen early on after considering the horsepower and speed requirements. In particular, the belts are HTD, with an 8mm pitch and 20 mm width. One single sided belt turns two of the propellers in the same direction as the engine and a double sided belt is used to turn the other two propellers in the opposite direction. Two idler pulleys, one for each belt are used to tension the two belts.
The first step in building the drive system was to attach the main pulley to the engine. This is a 50 mm wide pulley directly mounted to the engine shaft with a shaft key. A 50 mm wide pulley was chosen to fit the two 20 mm belts with a 10 mm gap between them.
The next step was to assembling the shafts and pulleys for each rotor. The pulleys are mounted to a drive shaft using a ball bearing bushing. The drive shaft is simply 3/4″ all thread and is bolted to the upper and lower support arms. Later, each propeller will be bolted to the pulleys.
Once the engine and propeller pulleys were mounted, a piece of string was used to measure the required belt sizes and lay out the location of the idler pulleys.
When researching HTD belts, it was surprising to find out how much power they can handle. The strength is due to the continuous length of cord inside the neoprene rubber. The downside to this design is that the belts come in set lengths and splicing them will severely reduce the load the belts can handle. So after measuring the belts, the next biggest size length was selected and the locations of the idlers were chosen to take up the slack. Shown below is the double sided belt.
The idlers are mounted onto 3/4″ square steel tube and rotate about a vertical piece of 3/8″ all thread. This allows the pulley to rotate about the all thread and apply tension to the belt.
On the opposite side of pulley, another length of all thread is attached horizontally. This piece is used to set the tension.
After assembling the idlers, tightening the belts and the making sure the pulleys were aligned, the drive system was complete! If you look closely, you’ll see part of the exhaust system being test fit. We’ll cover that in one of the upcoming project logs.
3 months ago •
Progress has been somewhat slow the last two weeks. Machining the other side of the propeller, went well and it was exciting to have a full scale propeller in hand!
However, having it in hand and thinking in detail about the next step, applying composites over the foam, I realized it would be rather difficult to fit the cloth material over the sharp edges of the hub and that the hub needs to be redesigned. At that point the computer I use for CAD and CAM went bad which brought the project to a standstill. I just got that fixed and now I have a new hub design with a much smoother transition that should be easier to apply the composites over. Hopefully the new design will be machined in the next few days.
Meanwhile, I’ve been researching the controls, and what controller to use. For the first iteration of the design I’m going to use 2 vanes under each propeller for directional control. This will require a total of 8 servos, plus one more for the engine throttle.
As for the controller, the Hackaday article on controllers, was an excellent starting point. After looking at all the choices, I decided to go to the Pixhawk controller which will handle all the servos needed. If anyone has experience with the Pixhawk and if there’s anything I should know about it before using it for Goliath, I’d welcome any inputs. Hopefully I’ll start ordering controller parts this week.
3 months ago •
If you haven’t seen it yet on the blog, Hackaday wrote a great article on Goliath! Thanks to everyone so far who’s followed, skulled or posted some of the great feedback.
This weekend I’m cutting the latest version of the propellers (see below). If all goes well I hope to have all four cut by the end of the weekend. I decided to stick with fiberglass for the first set of props since it’s about 4 times cheaper than the Kevlar I was going to use. It’ll be heavier, but weight isn’t an issue yet. The fiberglass should arrive on Wednesday, so doing the layups will be next weekend’s project. After those are done, it won’t be too long until the first test flight!
Also I’ve setup a github account to start sharing the project files at https://github.com/mccloudaero/goliath-quadcopter. I’ll start uploading the design files as I complete them.
2 months ago •
With the frame completed the next step was the exhaust system. Since Goliath is a unique vehicle, there wasn’t any off the shelf products that would work, so a custom exhaust was needed. The engine’s 30HP rating is with the air filter removed and no exhaust, so keeping the back pressure as low as possible was important. The engine has two exhaust ports so a dual exhaust system was built. The exhaust system was started from a exhaust kit and muffler intended for racing go-karts. The most difficult part to find was the right exhaust flange to fit the engine. After spending some time looking around, I decided to try making the exhaust flange out of steel on the CNC Router.
The CNC router used was a CNC Router Parts 2448. It’s a very beefy router, but it’s not made to be cutting steel, same as most CNC Routers. After doing a bit of research, it seemed that it would be safe enough to attempt cutting the two flanges I needed, as long as the machine went really slow so I don’t put too much stress on the system and used a coated end mill to help keep the temperatures down.
It worked out ok. As you can see from the video, I didn’t set the depth deep enough on the first pass. The biggest problem with the final piece was that the bar stock I used wasn’t very flat, which meant that it wouldn’t create a good seal. In the meantime I found a universal flange that would work with some extra welding and decided to go that route.
Below are the exhaust parts kit used to build one side of the exhaust.
After test fitting the parts together on the vehicle, they were taken back off and welded together. Below are most of the elements welded together.
With the exhaust assembled, it was test fit one last time and a bracket was made to support the exhaust. The bracket was made by twisting a piece of flat slotted steel and a U-bolt was used to mount the tubing.
A coat of high temperature primer and paint was added to keep the steel from rusting.
The exhaust was then bolted back onto Goliath for the final time. You may notice that the exhaust isn’t symmetrical. When I was fitting the exhaust on the starboard side, I accidentally placed the pipe on the wrong side of the bracket and welded the flange at the wrong angle. It’d be a lot of effort to rework it, so I’m going to leave it for now and fix it later if it becomes an issue.
At this point the only things keeping us from testing out the engine is getting the fuel and electrical hooked up to it. I’ll talk about the initial electrical system next time.
2 months ago •
The last few weeks have involved a lot of work, but not very much progress. Previously I had shown the latest propeller shape cut out on the router and I cut two initial rotors. Most of the supports were removed (with exception to the tips) and the foam was lightly sanded to remove the tool marks. Last Wednesday I received the fiberglass fabric and last weekend I started the layup process.
It didn’t take very long to have my first lesson learned. I’m using West System epoxy and I had gone out and purchased slow hardener (206) knowing that it was too hot (Houston during the summer) for the fast hardener (205). I really underestimated the working time I would have with this and it was gelling before I even got halfway through the first layer of fabric.
I then went and got extra slow hardener (209) and started again with the second foam blank. That worked much better with the heat and I had enough work time to get all three layers of fiberglass over the entire blade. Shown below is the propeller with 2 layers.
The second lesson learned was the the foam is not stiff enough to support the weight of the uncured epoxy and fiberglass when working with it. As I would turn it back and forth to get the fiberglass on it the blade would flex and the previous fiberglass layers would slide and start air bubbles (I kept working the air pockets out, but the final product has a lot of air pockets in it). After getting all the layers on it, it was wrapped up and placed in a vacuum bag.
After getting the cured product out of the vacuum bag the next day ( I guess I forgot to take pictures at this point), the final product showed the previously mentioned air pockets and one of the blades was twisted strangely, ruining the propeller. So the third lesson was that the supports at the tips are insufficient for keeping the blade in place in the vacuum bag. I’ll have to build some sort of support to hold it in place during the next attempt.
To fix the stiffness issue, I borrowed a technique from surfboard building, where they use wood strips along the center of the foam to add stiffness. The wood I choose was 1/16″ thick birch plywood. I cut the 10″ wide foam block into three pieces and used epoxy to glue the three sections and two wood inserts back together.
After trimming off the excess, I started cutting a new propeller. The wood strips didn’t seem to affect the routing and everything seemed to be going alright. Then for some reason I haven’t figured out yet, halfway through the finish pass on the top side the CNC Router lost sync with the computer. Now I’ve had this happen on one or two occasions and I just restarted the g-code from the beginning and it’ll be fine. This time I decided to restart the cut in the middle of the program, otherwise I didn’t have enough time to let it finish before I went to work.
I really thought I had done everything right, but the results show that was not the case.
It’s lucky I’m working with foam because if this was anything harder, it would have messed up the machine.
In the meantime I’ve ordered and received a PixHawk flight controller and I’ll get started on learning that soon. I’m also getting the frame ready to do some load testing to make sure it’s sufficiently strong enough before running the engine. I’ve got another propeller blank curing and now that I’ve got some confidence in the tool paths I’ve been using I’ll start posting those to the github shortly.
2 months ago •
I’m starting to get the Goliath design files place on Github . This evening the files necessary for creating the propellers that rotate CCW were committed. Please check them out and let me know if there’s anything else that should be there. The instructions need additional detail, but I’ve tried to label everything to be straight forward. The G-code files are included, but aren’t optimized. The roughing pass is fairly efficient, but the finish passes waste some time going over areas that are already cut. If anyone has the inclination and expertise to improve on them, please feel free.
I’ll be adding information to the repository as the project continues so be sure to periodically check it out!
2 months ago •
One thing I hadn’t addressed yet in the project logs is the connectivity requirement of The Hackaday Prize. I’ve had some concepts that I want to incorporate into Goliath, and over the past few weeks I’ve had a chance to refine some of those ideas.
There are a lot of discussions about the pros and cons of drones. The safety of others, is one of the concerns that people have, either due to a drone hitting a person or colliding with an aircraft. How do we address this valid concern? While Goliath is bigger and the consequences of a collision could be more serious, it does have the advantage that it has the ability to carry more hardware than a typical drone.
The FAA has addressed the potential for mid-air collisions with ADS-B. Using this system, individual aircraft transmits a standard signal to nearby aircraft with location and airspeed data. While the system is intended to be 2-way (I don’t think the FAA wants Goliath broadcasting the signal), the system can be used in a 1-way mode to locate other aircraft. John Wiseman (@lemonoder) has a great page on his blog on adding Cheap ADS-B on amateur drones. He got the information on Hacking a $20 USB TV Tuner to receive the ADS-B data from Clayton Smith. This could be integrated into the PixHawk to notify the Pilot of nearby aircraft and help prevent collisions. Not all aircraft are equipped yet, but having this capability on Goliath would help. All aircraft are supposed to be equipped with ADS-B by 2020.
So what if we could have something similar to ADS-B for people and drones? Almost everyone has a smart phone these days. What if Goliath was equipped so that people could interface with it and provide their location data? What if they wanted to know more about Goliath? What if they really don’t want it around?
To address this Goliath is going to be equipped with a Raspberry Pi setup up as a Wireless access point. An observer notices Goliath and sees that there is a new network available (say DRONE-FEEDBACK or something like that). The connect to it and it brings up a webpage that provides information about Goliath. The page would allow the user to share their location with Goliath (if they desire) to help maintain safe distances, submit feedback data to the operator and maybe even access some of the data from Goliath. The Raspberry Pi would be connected to the PixHawk to receive some of it’s data, but hopefully in a manner that prevents someone from maliciously accessing the PixHawk directly. See diagram below:
For The Hackaday Prize, the emphasis would be on the Wi-Fi / Web Interface and if time allows the ADS-B would be incorporated. This is because it’s unlikely Goliath will be operating near other aircraft anytime soon.
Meanwhile I’ve been working with the CNC router and had two hacks I wanted to post. The first was to address the amount of foam shavings that get everywhere. The CNC router has had a dust collection system for sometime, and the end of the 4″ ducting was brought up to the router, but nothing was done to route the vacuum closer to the spindle. This week I made a shroud out of cardboard that seems to be working well.
The other hack I did this week was to address the CNC hardware resetting in the the middle of cutting. After doing some research, the culprit was likely the Gecko G540 overheating. There is some good documentation on adding heat sinks and/or fans to improve the heat rejection. I stopped by the local electronics store and got some heat sinks. However no one seems to carry the thermal adhesive locally. So in the mean time I used some zip ties to attach the heat sinks. It seems to be working because I haven’t had an issues since.
2 months ago •
For those of you in the Houston area, next Tuesday July 15th at 6PM, I’ll be presenting at the local maker group, Creatorspace. I’ll be talking about Goliath as well as showing the hardware I’ve built thus far. Creatorspace is a Houston Hacker space near NASA Johnson Space Center. Tuesday’s meeting will be at Odyssey Hobbies, (100 E Nasa Road One #45 Webster, TX 77598 ) Hope to see you there!
a month ago •
Lately I’ve been spending most of my time outside of my day job working on the propellers. The picture above shows the current progress of the 4 propellers for Goliath. From left to right, there is the foam core with the birch plywood stiffeners inserted. Next is the first clockwise (CW) propeller done being milled. Second from the right is the second counter clockwise (CCW) propeller with a few layers of fiberglass tack in placed and almost ready to have resin applied. On the far right, is the first CCW propeller with the resin cured and currently being sanded and trimmed. It’ll get a few coats of primer and a finish coat before it’s done.
The birch stiffeners seem to work well, and the propeller core is stiff enough now to handle the weight of the fiberglass and resin during the layup process. Another improvement to the layup process was tacking the fiberglass layers in place with spray adhesive, just the minimum amount to hold them in place. This made dealing with the multiple layers and overlaps a lot easier.
Previously I had done a propeller layup that ended up twisting on side of the propeller, but the other side was still good. This was used to do some rudimentary load testing.
I incrementally placed weights at the 75% span location to test out the loading. While this doesn’t match exactly what the blade will see in flight, it’s close. Shown above is 30 lbs loaded on the single blade which translates to 240 lbs for the entire vehicle and the currently the target hover thrust. There was about an 1.25″ deflection at the tip. I did test the blade to 35 lbs, still without any failure, but I couldn’t keep the weights from tipping over and take a picture at the same time.
Hopefully in a few more weeks the the set of propellers will be complete and some initial hover testing can take place.
a month ago •
It’s been two weeks since I presented at Creatorspace and I’m just now getting the pictures together. Thanks to Patrick for taking the great pictures and videos (see below). I have to admit that I really hadn’t processed how big Goliath really is until I was faced with the prospect of actually having to move it somewhere. I used an engine hoist to move it on my neighbors flat bed trailer.
Thanks to everyone who showed up to check out the hardware and provided great inputs for finishing Goliath. Even the trailer ride over was a learning experience, because the zip ties holding the battery in place broke due to the vibrations. Definitely need to upgrade the battery mounting hardware. While Goliath isn’t ready to fly, I did turn the engine over using the starter to show the how the belt system works. Below is the video of turning the engine over.
Meanwhile I’m still working on the propellers, the second one is now cured after having epoxy applied and the first one is currently is the process of being sanded smooth. I’ve also been researching how to build the Ardupilot code and started a fork for Goliath
a month ago •
I haven’t been posting build updates as frequently because the build activities have mostly consisted of sanding….lots and lots of sanding. The good news is that I’m nearly complete with the first flight quality propeller! I’ve talked previously about everything up to adding resin to the propeller so now I’ll talk about the finishing process.
My early attempts at vacuum bagging used saran wrap over the fiberglass/resin and breather cloth over that before everything is placed inside the bag. That was the propeller where the blade ended up twisting. The other thing that went wrong with that earlier propeller was that the breather cloth left an uneven imprint on the surface. For the first flight quality propeller, I omitted the breather cloth since there wasn’t too much excess resin.
The results were significantly better, but I should have worked the wrinkles out of the bag better. The wrinkles fill with excess resin that eventually needed to be sanded off. After curing the next step was to start finishing the propeller. I placed it back onto the CNC router using the alignment holes and re-drilled the mounting holes, as well as an alignment hole in the center. The center alignment hole was used to drill out the center with an hole saw. The center hole allows the shaft and nut to pass through. I didn’t cut the center hole with the router because it had a hard time cutting through fiberglass. I probably could have just cut it a lot slower, but it’s easier to just use the hole saw.
After that the the excess at the tips was trimmed off and the trailing edge was marked off and cut off with some shears. You can see the excess resin in both these pictures. A Dremel was used to start the smoothing process by remove the bigger pieces. Then it was time to sand (being sure not to sand into the fiberglass) and then add primer.
The sand/primer process was repeated several times. The second propeller didn’t need as much since I did a better job of removing the wrinkles. Here are the first two props in various stages of sanding…
While the first one needs just a little more sanding and a final coat of paint, I wanted to see how much it weighed thus far:
I was quite surprised that the propeller only weighs 1 lb, 3 oz. I had assumed a much more conservative weight for the propellers, so that’ll help get the vehicle weight down. If I make another set I’ll probably add more reinforcement.
I’ve also started on getting the props ready to mount. I removed one of the shafts from the vehicle and replaced the bushing bolts with longer 4″ bolts that the props will mount to.
Then I testing out the hole alignment and bolted the prop to the bushing. The only thing left it to make a crush plate to help distribute the load over the prop.
I’ll be posting more often over the coming weeks as I get closer to finishing the props, getting ready for a hover test and completing the entry requirements for The Hackaday Prize deadline!
19 days ago •
Over the last week I completed doing the fiberglass layup for the 3rd and 4th propeller. Cutting the cores and doing the fiberglass layup for the propellers was one of the biggest parts of the project and it’s good to have it out of the way. The propellers all just need some sanding primer and paint and they’ll be ready to start hover testing. From left to right are propellers 1-4.
The last part needed for mounting the propellers are the crush plates. These will distribute the force from the bolts around the propeller hub. I created the CAD, machine code and cut four copies out of 1/8″ thick birch plywood.
All the files for the crush plates have been uploaded to the github repository. I’ll be uploading my entry video for The Hackaday Prize in the next day or two and getting everything else ready for the first hover test.
11 days ago •
Connecting a servo to throttle and choke is one of the last things to be done before the hover test. I’ve been taking a look at the hardware on the engine to see what’s required.
The engine user manual doesn’t cover anything regarding the mechanical connections. A bit of research on the internet hasn’t shed any light on the details on the throttle plate either. I know that the fuel solenoid is what has the electrical connection. The lever arm attached to the fuel solenoid is likely the throttle. That means the lower hardware where is likely the choke. There’s a few springs and connections, which don’t seem straight forward. The one thing that’s clear is the the vertical push rod is the only part that interacts with the rest of the engine. If I’m going to just connect a servo, this seems like the best place and the rest of the hardware could just be removed.
If any one has better knowledge or access to a schematic on this, any inputs would be helpful.
7 days ago •
So things are close to being ready to do a tethered hover test this weekend! As things progress over the weekend I’ll add a Hover Testing post and update it as things progress.
The tentative plan ( trying to allow for extra time for the inevitable issues that allows seem to pop up) for the weekend is:
Friday – Mount the remaining propeller and test the propellers and belt system using only the starter
Saturday – Run the engine on gas at low RPM.
Sunday – HOVER!
6 days ago •
UPDATE – Great news, got the engine started! Bad news, I broke lots of stuff! See bottom for details
I’m taking advantage of a four day weekend by doing a series of tests, eventually building up to a hover test. I’ll be updating this post over the weekend as I make progress towards the hover test. I plan on sharing whatever happens good or bad, and since these things seem to never go according to plan, please be patient if it seems like it’s taking too long or nothing happens at all.
As I laid out in the last post, Today (Friday) the plan is:
- Friday – Attach the rotors and test everything out using only the starter
- Saturday – Run the engine on gas for the first time, but only at low speed
- Sunday – Run the engine at higher speeds, building up to a hover test
- Monday – Not cleaning up debris from a failed test
So the weather this weekend does not look good. Today (Friday) is an 80% chance of rain, Saturday 90 %, Sunday 80% and Monday is 40%. Things might slip a bit. I was running out of room in the workshop to actually work on the vehicle with the rotors attached so I moved it outside under a pop-up canopy to get the last two rotors attached. I have them attached, but I’m still adjusting all of the pulleys to make sure they are all aligned. It started raining so I lowered the shade to cover Goliath better and waiting for a break in the rain. I did get one other important item installed (below).
I got my T-Shirt this week from the “Astronaut Or Not” Challenge, specifically the “Most Outrageous Component” round. They also sent included a few stickers, one of which is now placed on the engine.
So the rain let up for a while and I was able to finish adjusting the pulleys and belts. Goliath is finally starting to look like a quadcopter.
Next step is getting ready to remotely start the engine. Previously I had connected an riding lawn mower ignition switch to Goliath for testing out the starter and other hardware. Later the ignition will be controlled with the Pixhawk controller, but for doing some preliminary testing. I need to start the vehicle remotely for safety reasons, so I rigged up an extension for the ignition.
Last step is making sure Goliath isn’t going to go anywhere while just running the starter. I’ll get more serious with ties down for the hover test.
With all those things done, it was time to try it out.
Everything seemed to work as expected, so everything looks good for running the engine on gas next.
So after meeting all the goals on Friday, Saturday did not go as well. The first part of the day was doing a bit more research into the Pixhawk controller. I’d like to have the Pixhawk included as part of the hover test and I’m working on making sure I have everything I need. Turns out I misunderstood some of the documentation on the Pixhawk. While the Pixhawk has 8 main and 6 aux PWM outputs, the Pixhawk does NOT provide power for servos.
To power servos off the Pixhawk, a BEC needs to be connected to one of the servo inputs to provide power to the rest. So this morning I called around and got a Castle 10AMP BEC that will convert the voltage from the 12 V Battery currently on Goliath to what the servos need. This will be in addition to the Voltage Converter that powers the rest of the Pixhawk.
After making a trip to the Hobby Shop, I got ready for today’s testing. First thing was getting the fuel solenoid tied in to the ignition switch. After that I connected a temporary gas setup consisting of nothing more than a 1 gallon gas can with 1/4″ tubing shoved in it and taped off.
I had wanted to keep the gas separate from the rest of the vehicle, and it was setup underneath at first, knowing that the change in height might be too much for the fuel pump. However after turning the engine over several times it was obvious the fuel wasn’t reaching the engine. The engine comes equipped with a clear fuel filter case and the gas got to the fuel filter and stopped. Once the gas can was place on top of the vehicle, the fuel flowed freely though the filter.
Several attempts were made to start the engine, but the engine never fired. The connections were checked, but nothing seemed to work. Eventually the battery got low, so the battery charger was connected to recharge it. It could be that the long wires for the ignition extension are the problem. Sunday I’ll have to get the voltage meter out and do some testing.
While waiting for Goliath to recharge I did get some time to start prototyping the throttle servo bracket out of some scrap aluminum sheet. It’s an OK start but I need to find a spot on the frame or engine to connect to the top of the bracket to make it more secure.
Hopefully it won’t take too much more work on Sunday to get the engine figured out and still get the hover test completed this weekend.
I had to do some office work this morning, so I got a late start today. It’s raining so I’m working adding some of the electrical needed for the final design, starting with having a master switch and fuse, along with a mini fuse box for the different systems.
The good news is that we haven’t had to use the fire extinguisher so far, but the engine still isn’t starting. I took advantage of the rain to do some rewiring. After yesterday’s testing, I suspected that the long wires used for the ignition extension was the culprit so I worked on rewiring so that some of the components weren’t wired through the ignition extension.
The first thing I did was add a master electrical switch and a power box. This lets me know that I can turn everything off in the system and easily isolate different components. Currently the alternator and ignition are wired trough the box and they are connected when the master switch is enabled.
The fuel solenoid and starter are still wired through the remote ignition, which can only be used if the master switch is on. To make sure the fuel solenoid was working I removed the solenoid and visually checked to see if it actuated when the switch was enabled. It did activate and when I removed it, gas poured out of the carburetor, so I’m confident fuel is getting to the engine.
The rain cleared out for a little bit to try starting it up a few times again, but it didn’t start. There isn’t any sputtering, it just turns over without making any other sounds.
Right now I suspect that the ignition is not working correctly. There seems to be a slight acrid smell after the electronics have been powered on for a minute or two. It could be the ignition coils, but I’m not sure. If anyone has any experience with starting a brand new engine like this or if you have any ideas, I’d love to hear from you.
So the afternoon was rather eventful. The problem with the engine appears to have been fouled spark plugs (though I don’t have an idea how new plugs could be fouled). The engine started up strong and came up to speed quickly. After a few seconds I shut it off, but one of the belts appeared to lose tension and the blades got caught and two of the propellers broke. Attached is some video. Note, the startup seems to have been cut off, and while in the video the propellers seem to spin backwards, it’s likely an artifact of the video capture speed. I’ll post more details later when I get things sorted out.
2 days ago •
If you hadn’t seen the previous post yet, the gas engine was started for the first time, but during the process the vehicle was damaged.
Afterwards, I took an assessment of the damage to Goliath and tried to figure out where things went wrong.
After watching the video a few times and looking at the damage to the vehicle, I think I have a good idea of what went wrong and what I can do to prevent it from happening. Things were running good until the engine was shutoff. At this point one of the belt started losing tension. You can see this in the video at the lower right hand belt starts to flap. This was likely do the the engine spinning down faster than the belt. At some point the belt gets so much slack, that the belt bounced up and the propeller went under it and the belt got wrapped around the prop. Once it was tangled the belt cinched up really tight and bent two of the propeller shafts and the belt tensioner support. The other propeller attached to the belt was sheared off when it’s axle was bent and the propeller hit the angle iron support.
The changes I need to make to the vehicle to keep this from occurring again are:
- Add a one-way (overrunning) clutch to the engine pulley
- Add belt guards to prevent the belt from flying up into the path of the propellers
I may also need to add some auto tensioners, I need to do a bit more research into it.
Otherwise the test went well. We could really feel the wind coming off the vehicle. I’m really amazed that the belts are as strong as they are. I would have thought that the belt would have snapped in this situation. I’ve already started on making two new propellers and hopefully the process will go faster now that I’ve done it a few times.