April 22nd: Research and Designing the First CAD models of Mark III

For the Mark III, I aimed for a total overhaul of the Mark II (shown below), primarily to incorporate more powerful servo motors, have better walking gaits, and achieve a more conventional hexapod appearance.

I took a lot of inspiration from hexapods like the MX-Phoenix and Aecert Robotics' hexapods, and held them as my goal for the final appearance.

MX-Phoenix: Aecert Robotics:

So, clearly there was a lot of work to be done. The first objective is to create a way to mount the servos into the 3D printed chassis (I'm using MG996R's which don't have easy screw holes to mount to). My first idea was to follow Aecert Robotics and make a sort of sleeve for the servo (shown below). However, with this, there would only be a single point of connection to each motor (at the horn); if left alone, this would be really unstable. I added an extension to the bottom of the servo (in line with the gear) to act as another rotation point.

Issues: Bottom mount part is very annoying to print; very likely to mess up. The tolerance of the sleeve is very tight because slightly too small would prevent the motor from fitting, and too big would prevent the mounting pegs from fitting. Additionally, the bottom mount doesn't secure well to the sleeve and is thus unable to bear weight.

Total time spent: 2hrs

April 24rd: Redesign and Femur Has Been Created

The bottom mount is proving to be inefficient, so I decided to redesign it. Since mounting was the issue, I decided to make the peg part of the bottom piece of the motor case (it can be removed by taking out the screws).

Next, I started to make the Tibia because it looked like the simplest part. I started with simple connecting bars to each mounting point on either side of the motors. The actual mounting of the servo motors would be through the screw holes on the motor horns. Finally, I made a cool-looking X middle connector with a bunch of chamfers. I later decided to move the center connector slightly up, as to avoid issues with it hitting the motor.

Issues: All of the femur parts had to be sized up 102% in slicing in order to fit with the servo horn I was testing it with. Also, the X connector had a really, really tight fit with the side bars.

Total time spent: 2hrs

April 25th: Making the Tibia

Today I designed the Tibia. Because I wanted to make it into a curve (so it looks cool), I used ChatGPT to find that the ideal absolute length of the tibia is 130mm. Next, I used a straight line from the servo case outwards with a curve on top to begin making the leg. I had to be careful to place the start of the line where the rotating gear would be. I added screw holes and protruding holes in the tibia design to ensure a strong connection with the servo mount. After, I changed the tip of the tibia into a ball because it would be brought down on different angles. This would make sure that there wouldn't be much imbalance where based on the angle where the tibia is brought down.

Issues: the mounting protrusions don't really fit too well (tolerancing). While printing, the bottom side of the tibia often has slight deformations.

Total time spent: 1hr

April 26th: Making the Coxa

To finish off the leg, I finally made the Coxa (base joint). I just made a simple connector between two servo holders with different orientations, while using the same mounting protrusions

Upon printing, I realized there were a few errors. With tolerancing, the servo horn part had to be scaled up 102% in slicing, while the servo mounting had to be exact. To fix this, I manually sized up the horn coupler in CAD. With printing, there isn't an easy way to print the coxa, since either the horn coupler would have to be printed sideways, or there would be a massive overhang. Finally, the original design I made was facing the wrong direction...

Total time spent: 1hr

May 3rd & 4th: Making the Centerpiece

For the centerpiece, I began by making a simple connector to hold the coxa servo motors. I made a sort of pie slice that would connect to the motors and copied it around to make a hexagon. I also added a simple dome to cover up the center. Later, I changed the mounting to just have the motor sleeve as a part of the centerpiece. However, with this design, I wouldn't be able to fit it on my A1 Mini printer.

To make the centerpiece fit on the build plate, I decided to have the motor-mounts print separately and screw on later. I also added grooves to ensure a stronger connection.

Issues: The grooves don't print sharply, so they don't make a complete seal. Upon printing, I realized that the center hole isn't large enough to fit my MCU.

With this, the hardware is complete and I can begin with the electronics / code.

Total time spent: 3hrs

May 11th: Playing Around With Inverse Kinematics

I'm starting to play around with how I want Mark III to move. With the earlier models, I just used hardcoded angles for movement, but for Mark III I want to use inverse kinematics for smooth, new movement.

Turns out, inverse kinematics is quite simple! You use known segment lengths as well as a relative destination to use trigonometry to solve for the angles. However, with my first attempts at IK, the leg wouldn't move directly to the destination and would move wildly. To fix this, I made the leg move in smaller ticks in a straight line so it would just follow points along a line instead of just going to the destination.

2 Dimensions: https://github.com/user-attachments/assets/a3eb9b45-d414-4b76-a091-abc09decaa17

3 Dimensions: https://github.com/user-attachments/assets/2f3c27b4-28a0-40f4-89b9-6f2e65d93d94

Total time spent: 3hrs

June 18th: Finished the Box Plot

I finally finalized my box plot for the PCB! It's also somewhat heavily inspired by Aecert Robotics' design, but I removed the pressure sensors because I don't see the point in them. Also, I'm going to use the Bno055 as the gyro because the one Aecert uses is kind of bad :/. However, I want this design to be original - uniquely mine, so I'm going to base the power from better buck converters, have the main MCU be a Nucleo F446re, and use more affordable budget servos.

Total time spent: 4hrs

June 20th: Made first Schematic :D

The box plot makes wiring so easy! Passives like capacitors are annoying, though (but that's what ChatGPT is for). Good thing that datasheets give circuit diagrams, though! I'm having a hard time comprehending some parts of the circuits, like resistor dividers, but I'm slowly learning. Additionally, there are a bunch of ERC errors, and I don't think the nucleo footprint has nearly enough pins? Finding symbols/footprints was a massive pain, too. For example, for stuff like the passives, I kinda don't know what footprint to use because there are so many and I don't want to use the wrong one! Also, for things like the Bno055, all I could find on the internet were symbols of the chip itself, not the breakout board. I ended up having to make my own symbol.

Total time spent: 3hrs

June 25th: Finally fixed my schematic ugh -_-

I finally found the rest of the Nucleo footprint(s)! Turns out, there are 6 different parts. Although, I'm only using the two that have morpho pins XD. Also thought to clean up the space a little. The buck converter passive circuit was becoming confusing so I just wired it. The servo motor array is the bane of my existence - so annoying to rearrange. An annoying thing is, each pin has like 5 different purposes and I can't find any concrete this pin does specifically PWM/UART/I2C. Despite that, I CAN FINALLY START MAKING THE LAYOUT.

Total time spent: 5hrs

June 26th: Layout Designing go brrr (SO MANY ERRORS UGH)

The buck converters' passives are sooooo annoying. I have to place so many things so many times. I'm also starting to realize I might not have enough space (maybe it was a bad idea to use ChatGPT to find footprints). I'm going to use something like Digikey to find new footprints for everything. Also, the servo --> PWM connections are all over the place (rearranging them is going to be a painnnnnnnn). I genuinely don't think I have enough space for the PCA servo driver, so I might get rid of it. Also, I don't know how to route power because 15 amps is a lot and my trace calculator said it has to be 300 mils...

Total time spent: 3hrs

June 27th: Layout Designing go brrr - Take 2

I finally got a more optimized power system!!! Also, I found a way to fit everything on the board. Turns out that the footprints that ChatGPT gave me were wayyyy too big to fit. However, I do think that I'm placing things a little close together (in order to fit everything on the board) that there might be heat interference or smth. Routing is going to be fun!

Total time spent: 5hrs

June 30th: Layout Designing go brrr (fml)

I HATE DRC. IT TEARS APART MY HOPES AND DREAMS. Routing was fun while it lasted, but there. were. so. many. errors.............. I feel like I don't know how big I should make the buck converter traces. My calculator said that they need to be crazy wide (which cannot fit on the VERY small space that I have), but ChatGPT (that I kind of don't trust anymore) reassured me that the short length of the traces would negate damage or something.(Also, I realized I can use pours for the power :O)

Total time spent: 6hrs

July 1st: DRC PASSED!

After a lot of adjustments, all the DRC errors are now gonnnneeee!!!! I also learned about and implemented JLCPCB-specific design rules. Most of my errors consisted of: courtyard overlaps from components being close together, clearance issues of traces because the Kicad auto-route feature (I discovered) is based on a preset clearance that is just shy of the JLCPCB clearance, and some miscellaneous footprint/net errors. However, I'm still concerned on whether the servo voltage pours / vias will be able to carry the current. Prob a few tweaks left, but almost done!

Total time spent: 2hrs

July 7th: I'M DONE

I moved around the layout and increased the trace width of the buck converter circuits. I uploaded my schematic and designs onto several Kicad communities for final insight. Additionally, I also exported a 3d model of the PCB from kicad into OnShape CAD to check if it would physically fit. I also fully mapped out all of the necessary components in Digikey to the correct dimensions and requirements I need. I think I'm ready. I'm confident with the board and it's ready to be made!

Total time spent: ~2 hrs