The Great Brushless Drive Debate 2023

Our amazing and brilliant Brian Adamson had a wonderful solution for 250lb Malice’s Weight Issue, Brushless Drive.

Immediately, our Top David, Liaw, shot it down. “Our drive problem is solved.” Indeed, David is right, Malice’s drive is very infrequently the problem with why we loose matches. However, I, Bunny, know that if we don’t keep pushing the boundaries, we will get left behind. So when Brian approached me again, I knew we needed to have him get the real data on what it would take to make a brushless Malice drive equal to the brushed we currently use.

 

He did not disappoint, and came up with the following proposal:

Malice brushless drive proposal by Brian Adamson

29 April 2023



Introduction and Summaries



Introduction

 

Malice is currently facing complications related to weight management and a teamwide desire for additional features not included with the original design. These features include more robust top frame structure and potential permanent additions of titanium sheet armor. Of prime importance, however, is the desire to retain and improve the anti-wedge front-facing fork design, which has demonstrated tremendous usefulness across two competitions.

 

Malice was not initially designed for the fork system, and sacrifices are currently being made to create the weight allowances for these retrofits. Those limitations have included restrictions on the overall fork design (as they must stay within an extremely tight weight limit), limits on materials and fasteners used, and limitations on thickness (and therefore, strength) of frame materials. In addition, to create more ‘free weight,’ Malice has begun making use of custom-manufactured titanium fasteners in the robot, which provide adequate strength for their intended purposes, but represent a substantial monetary investment.

 

Finally, an additional issue has come to light across the last two competitions (Robogames 2023 and Battlebots 2022). Malice has lost drive power on at least one side in several fights, due to the drive motor shearing off of the gearbox. The motor is held to the gearbox face by four 8-32 screws, which are intended only for mechanical connection, rather than axial or lateral load. The gearbox itself, and the motor, are both held in relatively compliant UHMW structures which appear to, in certain circumstances, allow the motor and gearbox to move independently of one another. This movement results in the mounting screws shearing, which allows the motor to back out of the gearbox, allows hardened screw heads to fall into the gearbox, and can cause other mechanical issues.




Proposed Solution

 

A solution proposed in this document is to convert Malice’s current drive system, which uses high-quality, high-power brushed servomotors from Magmotor, to brushless motors (with hall-effect position sensors) and controllers used in personal electric vehicles (primarily, electric skateboards). These have been used with success by other teams at Battlebots and other competitions, and represent a substantial weight savings (in the most conservative estimate, over eight pounds). In addition to being lighter, brushless motors are physically smaller, which can make mechanical hardening of the drive system much easier. This hardening typically takes the form of traditional magnet-epoxy hardening (which is currently already performed on the Magmotors) and additional mechanical constraint of the motor to the gearbox. This additional constraint will reduce the likelihood of the motor shearing off the gearbox.



Requirements

 

Combat robotics is a time-constrained activity in a chaotic environment. Any proposed solution must be quick to install, easily maintainable, reliable, and rely on existing bodies of knowledge, with resources available outside of the immediate team. Therefore, this document treats the following as requirements for a successful conversion:

 

1.) As few parts as possible must be replaced. Gearboxes, especially, will be considered critical components, as much of the current frame design is informed by the shape and location of the gearboxes.

 

2.) The solution must maintain an adequate ground speed. Weight savings mean little if the robot becomes meaningfully slower on the field.

 

3.) The solution must use as many quickly-available, off-the-shelf parts as practical. 

 

4.) Motors and controllers must be quickly available, with a wide body of knowledge within the community. 

 

 

 

Potential Issues With Proposed Solution

 

Brushless drive (Here, meaning the combination of brushless motors, brushless speed controllers, and firmware configuration on those speed controllers) is, inescapably, more complex than the brushed alternative. While well-established “recipes” for motor, controller, and firmware configurations exist within the community (for example, MadCatter and 6374 motors, HUGE and 8085 motors, Claw Viper and Revolt RV-100/RV-120 motors), there is not a standardized ‘best practice’ process or document for brushless motor performance. These “recipes” are frequently most-applicable to the exact environment they were derived from, and additional tuning is frequently required when deviating from that plan. For example, the configuration and hardware combination used by the HUGE team with their motor/controller combination may not be directly applicable to Malice, and will require additional tuning. This tuning will consume time and testing cycles. The proposed speed controller hardware in this document (Vedder Electronic Speed Controller, or VESC) is intended for ‘inherently safe’ vehicles (bicycles, electric skateboards, etc) and as such, has many features that are actively detrimental to combat robotics. The process of ensuring these are disabled, and creating the ‘golden master’ configuration, will take time. Finally, brushless drive can be victim to intermittent, non-reproducible artifacts, which can either disable drive entirely until the system is reset, or cause undesirable driving mechanics. Brushed drive can also be subject to the same types of intermittent behavior issue, but anecdotal information suggests that they may be more common in brushless systems. 





Technical Discussion



Current configuration



The current configuration of Malice is as follows:



Drive motor: Magmotor S28-D4-400X (“Combat” Magmotor)

Drive motor controller: Castle XL2 (Brushed Mode configuration)

Drive motor voltage: 6S (22.2v nominal)

Drive motor speed: 4617RPM @ 22.2v (roughly 208 RPM per volt [Kv]). Information derived from motor specification of “5000RPM @ 24v”. 

Drive power: Approximately 4-5 horsepower (3000-3750 watts) of drive motor power per side

Wheel size: Approximately 12 inches per wheel

Gear reduction: Whyachi P3 12:1 planetary gearboxes

Weight per motor: 6.8 pounds

 

Approximate Ground speed: 20 feet per second (13.6 miles per hour)




Proposed configurations:

 

The proposed options primarily, are to replace the Magmotors with 8085-sized brushless motors from Maytech, and controllers with VESC from Trampa Boards. Trampa Boards works directly with Benjamin Vedder (the designer of the VESC) and is as close to a “first party” VESC supplier as exists. Additionally, these are well-understood within the community, and vast amounts of information are available from (for example) Rory Mangles, Ellis Ware, Jonathan la Plain, and Aren Hill. 

 

These motors have a large rotor, dissipate the desired power well, and fit well within the robot. They are well-understood within the community (this size, and brand, of motor are used by HUGE and Lucky) and are relatively economical. There is an additional option to use 63xx-sized motors; these are exceptionally well-understood in the community, and their power capability is commensurate with other 2-wheel-drive, horizontal-spinner robots (for example, Tombstone uses “short” Magmotors on P3 gearboxes). MadCatter has used one 6374 per wheel as well.

 

Of particular importance will be “soft start” circuitry. VESC does not handle the inrush current spike well, and simply closing a switch will have a very high likelihood of killing the speed controllers. Hal has already seen this in Duck. The Endgame team (at least used to?) sell “soft start” retrofit kits for Whyachi switches, and sell their own switch with integrated soft start mechanisms, which we are already using on the weapon.



Option 1: Shortest Lead Time, Most Expensive

 

Drive motor: Maytech 8085

Drive motor controller: Trampa Boards VESC6-High Power *or* VESC 75/300

Drive motor voltage: 8S (29.6v nominal)

Drive motor speed: 4736RPM @ 29.6v (160 RPM per volt [Kv])

Drive power: Approximately 5 horsepower (3750 watts) of drive motor power per side

Wheel size: Approximately 12 inches per wheel

Gear reduction: Whyachi P3 12:1 planetary gearboxes

Weight per motor: 2.5 pounds 

 

Approximate ground speed: 20 feet per second (13.6 miles per hour)



Discussion:

 

This option makes use of parts that are available with the shortest lead time. Maytech manufactures high-quality brushless motors, however the only motors kept in stock are 160Kv, which is notably slower than the current motor system. As a result, this option requires new batteries. This option saves approximately 8 pounds over the Magmotors. 



Parts required for this option:

 

– Custom gearbox mount plates and rear-support hardware

– Custom shafts for 8085 motors to interface with existing gearbox pinions/keys and rear motor support

– New 8085 brushless motors

– New motor controllers (1 controller per motor)

– New 8S drive batteries

– “Soft start” hats for Whyachi switches OR Endgame switches with “soft start” capability




Option 2: Longest Lead Time, Most Direct Comparison

 

Drive motor: Maytech 8085

Drive motor controller: Trampa Boards VESC6-High Power *or* VESC 75/300

Drive motor voltage: 6S (22.2v nominal)

Drive motor speed: 4884RPM @ 22.2v (220 RPM per volt [Kv])

Drive power: Approximately 5 horsepower (3750 watts) of drive motor power per side

Wheel size: Approximately 12 inches per wheel

Gear reduction: Whyachi P3 12:1 planetary gearboxes

Weight per motor: 2.5 pounds 

 

Approximate ground speed: 21 feet per second (14.3 miles per hour)



Discussion:

 

This makes use of made-to-order 220kv 8085 motors from Maytech. These have a lead time, according to Maytech’s website, of 25-50 days. This option uses the existing set of Malice drive batteries, and is slightly faster. This option does not require new batteries. This option saves approximately 8 pounds over the Magmotors.



Parts required for this option:

 

– Custom gearbox mount plates and rear-support hardware

– Custom shafts for 8085 motors to interface with existing gearbox pinions/keys and rear motor support

– New 8085 brushless motors

– New motor controllers (1 per motor)

– “Soft start” hats for Whyachi switches OR Endgame switches with “soft start” capability









Option 3: Shortest Lead Time, Most Uncertainty, Lowest Cost, Lowest Speed

 

Drive motor: Maytech 6396

Drive motor controller: Trampa Boards VESC6-High Power *or* VESC 75/300

Drive motor voltage: 6S (22.2v nominal)

Drive motor speed: 4218RPM @ 22.2v (190 RPM per volt [Kv])

Drive power: Approximately 5 horsepower (3750 watts) of drive motor power per side

Wheel size: Approximately 12 inches per wheel

Gear reduction: Whyachi P3 12:1 planetary gearboxes

Weight per motor: 3 pounds



Approximate ground speed: 18 feet per second (12.2 miles per hour)



Discussion:

 

This makes use of rapidly-available 6396 190kv motors from Maytech. These motors are new to me, but the 63xx motors are extremely well-understood as a family, with the primary difference here being that the 6395 has a longer stator (thus, more torque and larger ability to dissipate heat) than the 6374. These are the highest-Kv motors that are available in the shortest period of time, but still cost about 10% of our drive speed. This option does not require new batteries.  This option saves approximately 7 pounds over the Magmotors. 



Parts required for this option:

 

– Custom gearbox mount plates and rear-support hardware

– Custom shafts for 6396 motors to interface with existing gearbox pinion/keys and rear motor support 

– New brushless 6396 motors

– New motor controllers (1 per motor)

– “Soft start” hats for Whyachi switches OR Endgame switches with “soft start” capability




Option 4: Shortest Lead Time, Additional Speed

 

Drive motor: Maytech 8085

Drive motor controller: Trampa Boards VESC6-High Power *or* VESC 75/300

Drive motor voltage: 12S (44.4 nominal, 2x of our current 6S batteries in series)

Drive motor speed: 7104RPM @ 44.4v (160 RPM per volt [Kv])

Drive power: Approximately 5 horsepower (3750 watts) of drive motor power per side

Wheel size: Approximately 12 inches per wheel

Gear reduction: Whyachi P3 12:1 planetary gearboxes

Weight per motor: 2.5 pounds 

 

Approximate ground speed: 31 feet per second (21 miles per hour)



Discussion:

 

This option makes use of rapidly-available 160kv 8085 motors from Maytech, but makes a major change with the battery input voltage. By tying two of our existing 6S drive batteries together in series and doubling the input voltage to the controllers, Malice gains a substantial amount of maximum speed. This option additionally lowers the total amperage requirements, which generally results in cooler-running, more efficient motors and controllers. This option will require redesigning the drive battery holder, but does not require new batteries. This option saves about 8 pounds over the Magmotors. 



Parts required for this option:

 

– Custom gearbox mount plates and rear-support hardware

– Custom shafts for 8085 motors to interface with existing gearbox pinions/keys and rear motor support

– New 8085 brushless motors

– New motor controllers (1 per motor)

– “Soft start” hats for Whyachi switches OR Endgame switches with “soft start” capability

– Series wiring harness for batteries

– New drive battery box




Conclusions



The above options will save a considerable amount of weight, and should offer largely-comparable performance to the existing motor drive system with zero required modification to the existing Malice frames and spares. It may increase the robustness (and as such, mechanical reliability) of the robot drive system, through better-supporting motors that are lighter and smaller, making them more difficult to shear off of the gearbox in the first place. However, these changes come with costs: one option involves replacing drive batteries (a very large monetary investment) and the three other options will involve, at minimum, custom part manufacture and time on behalf of the robot’s owners to retrofit, test, tune, and configure the robot’s drive controllers. 

After reading that thorough and amazing proposal, I knew I had to talk to our Top David again. While the decision falls to Liaw, I can overrule him if I don’t think he’s being reasonable. 

When I brought this well researched proposal to Liaw, he read it thoroughly. when he came back, he gave me the reasons he still believed we should not go brushless this year for Malice.
1. We have about $4k invested in our current 6s batteries, and to go brushless and get the same power we currently have, we would have to invest $400 per battery, and get 4 sets for 8 total batteries. $3200 more for just the batteries, not including the additional parts we need, when we have reliable, working batteries right now (which we know can take large hits) is not an ideal solution.

2. In addition, we have heard from both Matt Spurk (Kraken) and noticed that our favorite team to emulate, Witch Doctor, have decided that brushless drive is not reliable enough after big hits, and either are, or have returned to brushed drive. 

In the end, after much discussion with our team, we unanimously decided that for now, the slight weight savings we would gain is not enough to justify the large extra investment on a problem that is right now, not losing us fights. 

We are so excited to have added Brian though, and we have tasked him to looking into data to see if we can upgrade our weapon to brushless instead!