Well its progress. 

I burned some early hours this week putting this together and although it bends the way it should and the parts come together, I have found some areas for improvement for sure.

A lot of the issues come from the fact that when I asked for adhesive on the back of the flex circuit, PCBWay decided that I could no longer have through holes. So all the through holes that I had on the circuit which was every solder joint now became pads. 

Now some of those changes actually worked well for me as I found a nice SMT double header that fit perfectly on my double row of pins to connect to the board. However, I now have to make my own drill holes for some of the pads. When I pierce them the pads like to come off. So for the motor, to flex connections, I had bodge wires on half of them. Not the end of the world but an insult to my sense of order.

This further complicated the assembly of the foot which is crowded with the battery and and charger.  Admittedly, I did not think I was going to hit it on the first try and I sent the flex out before I designed the boots complete.  I expected to run some wires in the foot but without the through holes the wire joints are not as robust as I would like. It is pretty easy to pull the solder pad off the flex and there goes an hour of soldering and $20 in parts.

So I gave up on the sensor on the bottom of the foot, and then the battery as well. I can always connect a battery at the brain board for testing so it was just a trial.  I am going to have to go back through the foot and try to make the flex, the battery and the charger board line up more naturally so it is not a 3 handed job to put it together..

The bottom of the foot is design to take a flexible printed sole but printing it is another item on my punch list. I had tried to print it in Ninja flex but I could not get a full 2mm to print without the filament refusing to feed, even though the first 5 layers seem to work. I bought some Amazon Basic TPU which is a little less flexible and I hope to print it this weekend.

Going up the leg, the flex routing seems pretty good. I have not taken the liner off the adhesive except on the bottom but even without taking it down it seams OK. The liner provides a little reinforcement which is necessary and I flip it  50 times to get the screws in. I had one radius start to tear so this is a concern. Maybe when I redesign the layout of the boots it will de-risk this assembly.

The 3D printed  leg joints came together pretty well and the leg flexes well. The sensors fit under the sensor ramps well and  I look forward to running some motors and testing the sensor readings today.   I can solder up a flex in about a half an hour with about a 90% first time connection rate so this has improved greatly from my first flex design. I found that if I put some screws through the board and flex, it makes sure it is soldered in the right place. It makes me work harder on the soldering but the eventual assembly is at least possible. The sensor board to holes on the hip joint seem to need to be shifted. The board is not quite symmetric so when I flipped it, it did not line up as well.

In my adventures in creating through holes, I found that I I pierce below the pads it seems to be more robust than piercing through the pads. The pads are less likely to come off.

The double row of headers is pretty tight on the brain board, both tot the leg and do the PCA9685 board on the other side. To make it fit, I had to cut away the unused Pin on on the  right side when looking from the front and cut away at the PCA9685 board on the left. Lastly the motor and the PCBA board are touching so none of this is ideal.  It is squished like this because I was trying to stay with in the 10cmx10cm cheap limit for the JLCPCB board. I might try to move the board up under the ESP by about 7mm. This will make routing a little more difficult but it should be doable.

Other board changes that need to be made had to do with drive voltages and signals to the ESP. The ESP will be fried with 5V communication signals but the MP3 board and the HR-SR04 ultrasonic transducer only work with 5V. I forgot to include the level shifter for these devices. I will have to re-spin for this but I may as well figure out what else I messed up before I go through the effort.  

Mechanical punch list.

Build other leg while document process and irritations.

Redesign foot to try to line up flex with the switch and the 

Electrically, my punch list is as follows:

Move the motors

Read the sensors

Move the motors to a sensor location

Add the accelerometer

Move the motors to a target inclination.

Programming wish list

Output on a display or phone the joint sensor readings

Implement a method to record positions of interest

Ideally have the robot be able to move through a sequence by just sending it a file.

Or if it is on the device, record a set of points and let it cycle through them.