Obligatory First Featherweight Build Log (FeatherDozer)

[Sam_Gad]

Hi all who have taken time out of their busy lives to read another build log,

I am a long time lurker of this forum, but this is the first time I will contribute in an attempt to give back what this forum has helped me learn over the past few years. This, as you probably have worked out from the title is the build log of my first combat robot.

For the first entry, I will do a quick first introduction to me etc:

• Name: Sam Gad
  
• Current occupation: R&D Test Engineer for a Tier 1 supplier to the automotive industry
  
• Location: Essex
  
• Team Name: Just Send it Robotics
  
• Robot: FeatherDozer
  

As a first time builder but being already within my chosen field of work, I wanted to build a reliable first robot but also push my engineering skills. However, I will quickly mention that I am not great at coming up with names. As the robot is a featherweight and kind of looks like a bulldozer, I put two and two together ...

The robot was designed and built in 2019 ready for the 2020 events, but due to the current difficulties we are all facing at the time of writing this , I only went to 2 events in 2020. We will go into these events in the not-to-distant future but so far, I would say that progress has been much better than expected!

In the next entry I will go over some of the design philosophy and CAD. Below is a quick CAD teaser of FeatherDozer Mk1.0 (Note that I am a very lazy CAD designer, therefore some elements and not modelled).

DF MK1.0 CAD 1.PNG

[Sam_Gad]

Entry 2 - Design Philosophy & CAD

Design Philosophy & CAD

As was mentioned my in previous post, I wanted to build a reliable first build but apply the engineering knowledge gained at both university and in my day-to-day working environment. So, in its most basic form, FeatherDozer in a compact 2WD robot, armed with an electronic, rear hinged lifter.

Before this final design, there were many concept that ended up being shelved due to an aspect that I wanted to keep to: manufacturability. Due to my limited shed and tool capacity as well as my desire to make as much of the robot in house, I wanted to be sure I can achieve the design before starting the build.

Drive
Due to the compact design, it became clear that I did not have the space or the weight for conventional brushed motors. Therefore, the drive comes from 2 brushless outrunners connected to a live driveshaft via a single gear reduction. This was to limit the number of moving parts for each side of the drive as well as the inherent lack of space (maybe I should have just made it a bit bigger). The chosen wheels come from an industrial belt sander as they were an off the shelf part with have good grip qualities and a solid aluminium inner hub.

Front Scoop and Weapon Mechanism
The front “scoop” consists of 4 main sections:

• 1 main upper Hardox section
• 1 HDPE Front sacrificial panel
• 2 Lower HDPE “forks”/ side scoops

The “forks” are designed to not only get underneath opponents, but due to being made from HDPE, they can flex with both impacts and the floor to always be as low as possible. I decided to make the whole scoop lift as one as it seemed simpler to me at the time (not sure why now I think of it) as well as be able to act as a raised shield when required. This causes a number of issues which we will get to in later posts. When fully lifted, the front is protected by an inner HDPE impact panel. This is shown below.

The lifting mechanism is powered by a linear actuator, however after initial testing of off the shelf systems, I found a number of limitations such as weight, size and quality of materials used (due to being a mass-produced part).

Therefore, I chose build my own system using parts from a car scissor jack, a brushed motor with a planetary gearbox and a decent amount of hope. The leadscrew from the jack is retained in the centre of the design by the weapon motor and gearbox housing and the front plate that separates the fuse module from the rest of the machine. This was the first aspect of the design modelled which gives the robot its final structure characteristics. For now, this system will remain brushed as the electrical system can be basic (2 high power relays to create a H-bridge motor controller). The image below is in the early build phase (hence the mock up of the arm in wood) which shows the full final actuator mechanism in the robot.

IMG_20190915_203435 - Reduced.jpgDF MK1.0 CAD 5 Lifted.PNG

Chassis
With the weapon mechanism determined, the main chassis can be split into a left and right module. Each module consists of a main longitudinal bulkhead, inner bulkhead to support the weapon mechanism and a transverse bulkhead that acts as the battery separation from the wheels, driveshaft outer bearing carrier and retains the rear armour/ wheel guards. The two modules are then connected together by the rear bulkhead as well as a number of internal spacers (as shown in red in the picture below).

DF MK1.0 CAD 2 Inner Structure.PNG

Electronics and battery positioning
As the central part of the inner structure is taken up by the weapon system, the main electronics had to go either side of the module to keep the design balanced as well as be accessible between fights. Each side of the main inner structure houses a battery and drive speed controller. The electronics are protected by a HDPE curved cover that is shock mounted to dissipate the energy of blows from an overhead weapon (e.g. an axe).

DF MK1.0 CAD 4 Top View.PNG

Other aspects
From the images it might be apparent that the self-righting mechanism was a bit of an afterthought. This is because it was! From initial testing it became clear that without an additional structure at the rear of the lifting arm, the robot did not have the articulation to self-right. This system shown was limited due to time constraints and was added just before my first event. As a result it is a poor design. This has been changed in the latest version (Mk1.5) but is untested and seems to be a major limitation to this design which will be fixed in the next robot.


In the next entry I will go over some of the build process leading up to my first event in late July 2020.

[Ocracoke]

The chassis design looks to be quite neat, better than the implementation of my left and right sections in my Middleweight at any rate, well done How thick are the front and rear bulkhead in that design?

[Sam_Gad]

Hi Lucy,

Thanks for your kind words . The bulkhead sizes are as follows:

• The rear aluminium bulkhead is 15mm thick (6082 T6 grade).
• The main longitudinal HDPE bulkheads are 15mm thick.
• The front bulkhead/ Impact panel (when the arm is lifted) is only 6mm thick HDPE but the front of the 2 longitudinal bulkheads is retained in several different directions so is very stiff.

I should have some time in the coming days to detail the build, which may give better perspective of how it all goes together as well as how little space is in the robot when fully assembled.

[McMullet]

Great machine, and a nice write up. I saw the videos of Featherdozer at Robodojo on youtube, really nice looking machine.

How did you get on with the sander wheels, were they grippy enough? Could you post a link if they're good, nice to file away these options that have been tried and tested!

[Ocracoke]

Construction sounds fine to me The 6mm HDPE, as you say, won't be too much of an issue as it is anchored down quite well by the diagrams, Azriel has a 6mm front plate for deflection which I figured would be enough because of that and being well bolted down. Looking forwards to the construction posts.

[Sam_Gad]

Hi Sam,

Thanks for your comments. The wheels great for the surfaces usually used in robot combat (steel, wood etc.). I will be using these wheels again in my next featherweight which I am designing now.

These wheels specifically are 86mm outer diameter x 50mm width. There are many other options (both diameter and width) on either Banggood/ Aliexpress or ebay directly from China (Link to the ones I use listed below):

https://www.aliexpress.com/item/3294...B&gclsrc=aw.ds

Similar wheels have been used before in featherweights to great affect on both shrapnel and the latest version of Get Shrekt.

There is an option for either the serrated or slick tyre options. I have both, but so far haven't used the slicks yet as in the next event (whenever it will be), I will be using new drive speed controllers and don't want to make too many changes at once in case of an issue.

I did have to do a lot to them to suit my needs as they are quite heavy due to the amount of rubber. On FeatherDozer, the width is reduced to 45mm (still very wide) and bored out to fit the axle-to-wheel interface. Before an event I usually cover the wheels in gaffer tape to help clean the tyre surface before it is needed.

[McMullet]

Nice one, cheers! That's not a bad price and much easier than making them from scratch from bike tyre and HDPE.

[Mr Manuva27]

Hey Sam,

Was great to meet you and see Featherdozer in action at it's first event back in July. You've built a fantastic machine and I'll be following this thread with interest!

[Sam_Gad]

Entry 3 - Initial Build Part 1

Before we start I would like to thank all who have taken the time to read the log so far and the complements made .

Time to Build Part 1

As there are a vast number of aspects to cover during the build process, I will split this into a few posts.

The decision to start making combat robots came at a transition period for me as I gave up one hobby (restoring classic cars) to start this one. Therefore, I sold my track car to fund tools, the build etc. I chose to buy workshop tools tactically for this build and buy additional tools on demand. As a result some of the more complicated/ hard to make parts were water jet cut by our friends at K-Cut. These were:

• Front Hardox wedge/scoop section
• Outer drive bearing carriers (also made from Hardox)
• Rear aluminium bulkhead
• Centre transverse aluminium bulkheads (replaced later in the build by HDPE due to weight constrictions)
• Hardox arm scoop backplate
  

This meant that some main aspects of the chassis were built very quickly.
IMG_20190623_170550.jpg IMG_20190614_075958.jpg

Mock-up of a side module with the drive motor, gears and live axle.
IMG_20190518_164303.jpg

Chassis mock-up with wheels and weapon mechanism front plate.
IMG_20190703_200345.jpg

Weapon Arm Mock-up
Before investing the time in building the final arm mechanism out of a number of different materials, I wanted to check that the mechanism works as intended. Therefore the components were initially made out of MDF. These images also show how the front end of the robot is stiffened up by weapon mechanism front plate.

IMG_20191006_141246.jpg IMG_20191124_221135.jpg

Final Arm installed with the leadscrew mechanism to the side.
IMG_20200208_105652.jpg

The front impact plate & electronics covers both made from HDPE in 6mm and 8mm thickness respectively. Note that the electronics covers are shock mounted in 2 directions to be able to flex down during an impact but not too far to damage the electronics inside.
IMG_20191223_181951.jpg IMG_20200103_173845.jpg

So this is where I will leave this entry with most of the chassis, drive system and weapon arm made and semi assembled.
IMG_20200103_173627.jpg

The next entry will (hopefully) complete the Mk1.0 build leading up to the first event.

[Sam_Gad]

Entry 4 - Initial Build Part 2

Time to Build Part 2

Scoop Build
As the scoop design is a combination of materials and manufacturing techniques, this aspect was a little more complicated than first expected. The most difficult aspect was that the lower scoop sections had to be at the same angle and height. After a couple of adjustments I was happy with the result once it was mocked up on the robot.

IMG_20200314_171926.jpgIMG_20200315_182011.jpg

Really starting to come towards the end of the mechanical assembly.
IMG_20200315_184106.jpgIMG_20200315_184132.jpg

Only main aspect left to do was the install of electronics ready for final assembly (with rear armour) and paint.

Electronics
This I knew was going to be the real challenge. Not because of a limited knowledge of electronics, but due to a lack of any space inside (as usual). Despite this, I wanted to have some form of order to the main loom so I can access most of the vital components (batteries, link etc.) without taking the whole machine apart.
To help simplify the electronics as much as I could, the robot has separate low and high power circuits. The low power loom feeds the battery eliminator circuit to the RX, power lights and weapon limit switches/ low power side of the relays. The high power circuit, feeds the drive ESC's and weapon high power relay circuit directly from the battery.
The drive speed controllers are Hobbyking F80 brushless speed controllers programed for bidirectional control via SimonK. I found SimonK to be limited way of flashing ESC's, but for my first play with brushless drive it was more of a set-and-forget job that was an already tried and tested method throughout the community.
The sides of the main arm assembly are dedicated to the two 4s (1500mAh) lipo batteries and drive speed controllers. The front section houses the main fuse, main power distribution and weapon motor wires. As is seen by the image below, this meant both internal and external wire routing around the main bulkhead as the "high power" wires of the weapon drive relays and link would not fit (note that the main fuse is not in the image below). The external wires are then protected (somewhat) by the front impact panel. Changes to Mk1.5 have been applied to help better protect these wires.

IMG_20200419_212129.jpg IMG_20200322_001035.jpg

Lifting Arm limit switches
As the arm electronics are "simple" it meant that the endpoints could be controlled in the same way that a conventional linear actuator works - limit switches at either end of the desired travel. As I had made my own system, I could position these limits wherever I wanted. This however had to be completed last to test the limits in real-time and adjust accordingly.
The front limit switch is positioned under the arm scoop spacer block and moves with the weapon mechanism. When the arm is lowered, the switch contacts the front inner impact panel to stop operation in that direction. The rear limit switch is positioned above the weapon gearbox and is switched by the arm link to the leadscrew. Both systems are shown below.

PXL_20201229_163653531.jpg PXL_20201229_163713088.jpg

Final Assembly
After a couple of setbacks which included robot weight reduction and self-righter last minute testing, FeatherDozer was ready for it's first event - Robodojo on 26/07/2020. I will go into minor details and analysis of the event in the coming entry. For now here is the completed build of FeatherDozer Mk1.0

IMG_20200724_172232.jpg IMG_20200725_142730.jpg IMG_20200724_174739.jpg

[Ocracoke]

Dang, I love the neatness of this, you should be pleased with the outcome. What drive motors is FeatherDozer running? I see they are brushless, presumably outrunners?

[Sam_Gad]

Thanks , I think it came out better than expected.

The motors are indeed brushless outrunners. They are propdrive 4248 motors that are currently limited to about 40% throttle on the transmitter (need to gear the drive down more) to be acceptable for the Sportsman Robodojo class as well as being too fast for the current arena size.

Details of the motors are in the link below. They also have seemed to be out of stock everytime I check the website which is expected in the current world situation.

https://hobbyking.com/en_us/propdriv...___store=en_us

[Sam_Gad]

Entry 5 - First Event 26/07/20

Happy New Year! Lets hope that more events are given the go ahead in 2021.

Event 1 - Robodojo 26/07/20

I decided to start in the Sportsman class as I wanted to learn how the robot performs before it gets annihilated in the spinner heavy full combat.

Rather than going through the fights which can be found on Joe Brown's youtube channel, I will go over the event listing the good and bad parts of the robot. This will lead into the subsequent changes made for the second event on 26/09/20. Obviously the events last year were put on under difficult circumstances, but it was only due to the Robodojo team as well as competitor's obedience to the additional regulations that made them possible so thank you all.
Before the event, there was little to no time to practice driving the robot other than a quick functionality test. This was a definite issue as I wanted to be as prepared as possible before the event but did not prepare for one of the most important aspects - driving practice. Despite this, I packed the car and headed to Leeds not sure what to expect.

IMG_20200725_145652.jpg IMG_20200726_095556.jpg

My aims for the first event were pretty high based on the competitors that were there:

• Pass tech check
• Function in the arena
• Not break down from my own doing
• Win 1 fight

The robot passed the main tech check apart from initial weight check where the robot was 2 grams overweight! A quick removal of a bolt from the rear armour fixed this . As I was busy concentrating on maintenance throughout the event, I didn't get much of a chance to take many images post fights.

Below is a quick list of good and bad points of FeatherDozer Mk1.0 on its first outing:

• It works!
  
• Won a few fights - Better than expected
  
• Driving practice required!
• Rear ground clearance is a little bit too low as it did get stuck on the raised floor sections due to poor driving
  
• The robot is way too fast for the arena size but reasonably controllable when trimming the rates on the transmitter
  
• Last minute self-righter doesn't work - New one to be designed.
  
• When lifting, the robot tends to lift the rear (depending on the opponents weight distribution) meaning that it cannot move when lifted. This was a major flaw in the design that could only be found in real world testing.

So, to conclude, Featherdozer ended the day with 2 wins and 4 losses as well as still being fully functional. This is far better than I expected based on the level of the competition. Therefore I would give the robot's first performance as way above my expectations and with a few modifications, it could be much more competitive.

The images below are post event which showed very little damage caused. Note that the red paint used tended to peel off very easily, so if anyone from the event wondered where random red paint chunks came from......

IMG_20200727_120843.jpg IMG_20200727_120857.jpg IMG_20200727_120906.jpg

In the next entry I will go over the changes made for Mk1.5 to fix some of the issues listed above.

[Sam_Gad]

Entry 6 - Mk1.5 Design & build

Time to Modify - Mk1.5

Despite a lot of successes from the first event, a number of changes were required. Apart from the obvious driving practice, I needed to focus on two of the issues:

1. Unable to lift opponents without become unstable (lifting the rear wheels)
2. Unable to self-right

Unfortunately the main reason for these limitations was the fact that a lot of the mass of the robot is in the lifting mechanism. So when the weapon operates, the centre of gravity changes significantly. However, as the robot baseline is a good, solid platform, only minor changes were required.

Lifting Issue Fix
The simplest way that I found to amend the limitation to Mk1, was to extend that front of the chassis with fixed forks. These forks are an extension of the front inner impact panel/ bulkhead brackets. When the robot is sitting on the floor, the fixed forks are lifted to reduce front contact patches (only the scoop is in contact). When the arm is lifted, the fixed forks stabilise the front to help prevent the rear from lifting. As the forks have to be very long to be effective and clear the scoop, lower bracing was added to the fork assembly. This is shown in the CAD images below.
Mk1.5 CAD 1.PNG Mk1.5 CAD 2 (No Scoop).PNG
Fixed forks mocked up in the robot.
IMG_20200816_145913.jpg IMG_20200816_145918.jpg
Cut-outs were made to the inner impact panel to accommodate the forks. The image below and to the right shows the final assembly (without the front scoop).
IMG_20200818_214944.jpg /IMG_20200822_143253.jpg

Self-Righting Issue Fix
A complete redesign was required from Mk1.0 as it was clear that the design was very flimsy and didn't work well. This task however proved to be a very difficult and lengthy process. I think I went through 5 different design concepts ranging from a completely separate arm linkage running from the weapon actuator, to a fixed anti tipping beam. But in the end the only design that came close to the aims of functionality and not look as if I had lost my mind , was to revisit the MK1.0 concept with better geometry. After 10 prototypes of various geometries, the design is shown below. The design is much sturdier as the main polycarbonate sections are now part of the main weapon arm assembly. The self-righter inner bracing is now much more optimised to stop flex when upside-down.
When testing, results were promising but once the robot self-rights in the arena I will deem this design as successful.
Mk1.5 CAD 3.PNG PXL_20200924_191944424.jpg

As additional items were added, some additional weight reduction was required elsewhere. Most notable areas of weight reduction were:

• Front scoop mount (made from hardox) was replaced with an aluminium alternative of the same thickness.
• Main arm inner aluminium mounts reduced in length to use less fasteners.
• Cutouts in the self righting arm without compromising functionality.

Final assembly went smoothly ready for my second event on 26/09/20 (Robodojo again) which I will go through in the next post. For now, here are some images of the final assembly (FeatherDozer Mk1.5).

PXL_20200924_151323389.jpg PXL_20200924_151336540.jpg PXL_20200924_192009391.jpg