19 days ago •
Thrust measurement and weight reduction has been the focus of the work on Goliath over the last few weeks. Prior to the last test, Goliath weighed 238 lbs, about 40 lbs more than the target weight when the project started. With the vehicle design still in flux it's not time to move away from the slotted angle yet, but there were a few areas that could provide some potential weight reduction.
Bolts and Washers
When Goliath being assembled for the first time, the final design was somewhat nebulous. To make things easier, the same sized bolts where used for the frame assembly. Most of the bolts don't need to be the full length, so wherever possible, the original 3/4" long bolts have been replaced with 1/2" long bolts. The size difference is shown below. About 75 of these bolts have been changed out for smaller bolts
While the bolts were being swapped out, the traditional lock washers where replace with external toothed lock washers. The primary reason for replacing the washers is that the traditional lock washers aren't very vibration resistant and every test would see at least one or two come loose. A bonus to replacing the washers is that the external tooth washers are lighter than the traditional lock washers. The picture above also shows the external toothed and traditional lock washers.
Cross Beams and Side Beams
The side and cross beams previously used the wide slotted angle. It was felt that they could be swapped out for the narrow slotted angle with a minimal reduction in stiffness. Below on the right is the wide cross beam and on the left is the narrow cross beam replacement installed.
The cross members were another area improved. Previously the two cross members on each side of Goliath spanned the whole length of the side and angle wasn't ideal for stiffness. Each of the cross members where removed and trimmed into two smaller pieces, with a little bit of excess. These pieces where then used to form two smaller crosses on each side. Below is a shot of the side beams and cross members being reconfigured on one side of the vehicle.
After completing the structure modifications, the end result is that Goliath looks a bit leaner, and the weight was reduced by 5 lbs to a total weight of 233 lbs. The slotted angle structure is probably as lean as it's going to get.
Another issue that needed to be addressed was the axles for the two propellers powered by the double sided belt. After inspecting the vehicle it was found that the axles were slightly deformed (see picture below). The double sided belt needs additional tension, probably because of the extra length and the difference in the angles. Eventually the all-thread axles will be replaced with thicker axles, but for now an easy fix was to replace the 5/8" zinc plated all thread rod with 5/8" stainless all thread rod. The stainless has a higher strength and has more resistance to being deformed.
The thrust measurement portion of the # Drone Test Stand has been completed. You can read the details there, but the test stand has been setup to measure the thrust while Goliath is running. Below is a picture of the remote display with Goliath in the background.
In the above picture the reading says 220.92 lbs. This was because some of the hardware had been removed from Goliath. Note that there is no data logging capabilities built it. The measurements are being recorded visually for now.
Three more tests have been conducted since the last posted test. There hasn't been time to compile the video yet, it'll probably get posted over the weekend.
Test 22 tested the latest changes in the frame. The test went smoothly and Goliath was brought up to full speed without the ignition switch coming loose. There was no flexing in the frame visually, so the new design appears to be working.
Test 23 attempted to get full thrust measurement data. However the double sided belt tensioner was not performing satisfactorily...
Read more »
a month ago •
The last few tests have shown that under full power Goliath was unable to produce enough thrust to lift itself. Analysis shows that Goliath should have sufficient power to lift itself, but it's difficult to say why Goliath is not hovering without additional data on the actual thrust and engine power being produced. The electronics for measuring the thrust produced is almost complete. Measuring the engine power requires measuring the engine RPM and a hall effect sensor is being installed that will be triggered by the magnet on the flywheel.
Meanwhile the design analysis has been revisited again to make sure there wasn't something's that been missed. The power required calculations were shown to be sound, so a closer look was at the propeller design was next.
Originally when designing the shape of the propellers, analysis was performed using simplified blade element theory. This consisted of some code borrowed from a Matlab code that I got from a University of Cambridge website. An attempt to make a spreadsheet out of this was unsatisfactory and Prop Designer was tried instead. Aircraft propellers and helicopter rotors work on the same principle, but operate in different regimes. Since the hover condition (static thrust) was the design condition for Goliath, Prop Designer gave an answer, but the software warned it may not be a good answer. With the deadline for the 2014 Hackaday Prize looming, it was decided to push forward and design the propeller with an angle of attack at the maximum L/D of the airfoil to get an efficient design.
Taking a closer look at the propeller design required having a more accurate analysis. This time a python code was written to analyze the design and the issues encountered previously were resolved. While the new python code is giving the best results for the propeller design so far, it still needs to be validated with data. However if it is right, then the new analysis shows a reason why Goliath isn't flying and more importantly how I can change the design to get Goliath flying.
Below is a plot showing the predicted power curves for different propeller designs using the new code.
The current design has a angle varying from 20 degrees at the root to 12 degrees at the tip. The problem with the design could be that it produces too much thrust at lower RPMs. The predicted power is too much for the engine to get above 2400 RPM, limiting the power to 21 Hp. The predicted thrust is at this condition is 220 lbs. Goliath weighed 238 lbs when last tested, so this could be why Goliath isn't flying.
If this is indeed the case, then the fix is to lower the tip angle. The plot above shows 3 additional designs where the tip angle is decreased in increments of 2 degrees. The last design in the list has a tip angle of six degrees and looks to be the most promising. This design could allow the engine to run up to 3300 RPM generating 27 Hp and the predicted total thrust is 290 lbs which should provide more than sufficient thrust.
The next test will provide thrust data and possibly engine RPM data. This will help validate the software and provide a good path forward to get Goliath flying.
2 months ago •
Goliath has now been completely reassembled and is running again. It took awhile to double check everything and get the tension just right on the belts. Below is the video showing Goliath being tested after reassembly.
Obviously, the ducting didn't work well. I had assembled the ducting using screw clamps at the joints. In hindsight, I should have used sheet metal screws, in addition to the screw clamps, to keep it from vibrating loose. I had protected against the possibility of the loose ends falling down, but didn't think about the joints. When the ducting failed during the test, the vehicle was vibrating more than usual because the tension on the belts wasn't quite right. That was later fixed, so if I decide to use the ducting again, the vibrations shouldn't be as bad. Goliath didn't suffer any damage because the aluminum ducting is pretty insubstantial.
The second run was good, with the exception of the ignition connector coming loose after the engine came up to full speed (and I need to move my Styrofoam sheet stock somewhere else).
For those who haven't read about the ignition system previously, here's a quick refresher. The engine has two ignition coils, one for each cylinder. The coils don't require a power source and are triggered as the engine rotates. To start the engine, the ground connection to the coils is opened and then the starter is used to start up the motor. To shut off the engine, the connection is closed again.
On Goliath, relays are used to operate the starter and ignition and the relays are setup for the ignition and starter to go to a fail-safe condition if they lose power or signal. During the full throttle test, the signal wire coming from the remote switch vibrated loose and the relay closed, grounding the coils and shutting off the engine. On the bright side, I know the fail-safe works, I just need better connections. Below is a picture I'd posted previously of the Parallax relay board being used on Goliath.
The relay signals come through the standard headers next to the parallax logo, labeled RLY1 & RLY2, connected with Futaba style servo connectors. The only thing holding the connectors to the headers is the contact friction on the pins themselves. Eventually, something more vibration-resistant, like the Hirose DF13 connectors that the Pixhawk uses, will be needed.
This was also the first test where I had good weight measurement immediately prior to the test. A bathroom scale was used to take the measurement. First, some wood blocks were placed on the scale to distribute the weight evenly and then Goliath was placed on top of the blocks. Here is a picture of the scale with wood blocks supporting the vehicle.
Subtracting out the weight of the blocks, Goliath weighed 238 lbs at the start of the test. This is higher than was expected as an empty vehicle weight of 200 lbs is desired, but Goliath weighs less than the theoretical total thrust of 255 lbs. It is evident that Goliath is putting out less than 238 lbs of thrust. The thrust measurement electronics will be finished up soon, which will give a better idea of the total thrust. The fix is likely going to be re-designing the propellers to get additionally thrust, but additional data is required before starting that effort.
Efforts will continue to determine where weight can be eliminated without compromising the stiffness of the frame. Once the configuration is nailed down a bit more, a Mark II version with a lighter frame will be built, but for now, the design flexibility that the slotted angle provides is necessary. During initial testing, the frame tended to flex while the engine was running, particularly at startup (see Test 2 video below).
Later cross members and some all thread running vertically to stiffen the frame were added. It's sufficiently stiff now, but efforts have already been started to make the prototype lighter while maintaining the stiffness. Below is a shot of removing one side of the vehicle and replacing the wide side beams with...
Read more »