I started out exploring patterns and grids to create flexible structures.

I created a parametrised base geometry (basic shapes (rect, hexagons, kites) with input parameters (length, angle, pattern spacing) from the user ) in Fusion 360. Marvelled at how a simple geometry could transform a stiff disobedient piece of cardboard into a gracefully bending entity. (Pictured) But patterning on Fusion was involved and time-consuming, so I started looking for a more powerful tool for generative design. Took a weekend retreat with Grasshopper.

We used a 1/32" end mill for the traces and a 1/64" end mill for the cutting.

We printed a test piece to gain confidence in the minimum trace width and gap width one can mill with the machine. These cute tuning forks in this case are used to tune thickness. I then milled the Amtel SAM D11C board that we will be using as a USB serial programmer for future projects. NOTE: I turned my back on the machine while it was milling and at the end of the procees I got: nothing. The end mill had lifted up as I tightened the screw and lost contact with the board. Be sure to hold the end-mill down while you tighten it into place.

LUCYSD

LucySD will be some sort of dark room umberella/canopy with a rotating reel of beautiful and calming paintings and backgrounds, and ambient music. An individual can prop their phone up in the holster and overlay a reel of memories (using iPhone memories mirrored/projected onto a block of translucent glass) on the rotating backgrounds and 'watch the show' and use the time to imagine, reflect, reminisce.
They can then view a log of their heartrate during the experience and map back to the instances where they may have experienced increased/lowered heartrate and use the insights howsoever they may please.

This might have made the Rajasthani artisans cringe, but I drew up in freehand one side of the profile for the board and mirrored it to create some semblance of symmetry. I played about with the height/width proportionality of the board to ensure it would not be too hard to balance on the stands.
I created some vents and a large through-square-panel on the bottom of the board where I could add my designed panels.

Just a note - when you're exporting your sketch as a dxf from Fusion, make sure that the sketch is on the top plane. V-Carve (and most other vector graphics tools) project your CAD from the top plane. If it's on the front plane, you'll only get a line in your VCarve workspace. Alternatively, you can right click on the sketch and select 'Export to DXF'. This operation is indifferent to the plane you're sketching in.

We were given 8'x4' boards to machine on, so I imported dxf profiles of my CAD to V-Carve to play around with placement.

V-Carve is a nice tool that links the ShopBot proccesses to your design. The UI was friendly and I used the following operations:
- Open dxf profile in V-Carve
- Select all using rectangular selection and go to 'Tools' and select 'Join Vectors'
- Scale, move transform, replicate to create and nest as many pieces as you need for your design
- Select 'Toolpaths' on the extreme right corner and pause to think about how you can optimise your cutting paths across the boards.
- First machine out any rasters, notches, lips that will only cut partly into the board. You can do this using the 'Pocket' toolpath and set the max depth to your choice. I used a max depth of 6mm for a board that was 13 mm thick.
- Finally, use the 2D Panel Cut toolpath to cut out your parts. Set the max depth to about 2mm more than the actual thickness of the board.
- For tool depth on each pass, the general rule is to set a pass depth equal to or less than the diamater of your tool (0.25" or 6 mm in my case)

Once all my parts were ready, I used the Orbital sanding tool to smoothen out the itchy wonky back of my OBS. I had to almmost sit on it to hold it down straight on the wooden surface but it was worth the effort and it created a smooth, un-chafing surface within a few minutes. I used a file to get into the smaller cuts and notches. I then joined my 3 panels together using gate hinges. Zach made the very useful suggestion of getting specifically gate-hinges for the OBS material for it's wide and short geometry, as it allows us to drill in attach into the meat of the OBS material, farther in from the edge, and avoid chances of splintering or detachment. I attached the hinges alternatively on the front and back side of the centre board, to allow a nice zig-zag placement of my 3-panel divider. I then lasercut some arabesque panels to add as ornamental fill-ins for the sqaure cuts on my main panels.

Voila! With a little more sanding, polishing and love, my divider will be ready for my room!

Here's what you need to feed your bits of code to the microcontroller

The process-chain is like so:
-Your newly born microcontroller (I used ATSAMD21E18A ARM Cortex)
-A programmer to connect the uC to the computer on start-up (I used Atmel ICE)
-A command-line interface on your computer that allows you to talk to the Atmel ICE (I used EDBG)
-A bootloader - a mini program (think of it as the uC's receptionist - checks for new programs/instructuions every morning when the uC wakes up) (I used Trinket M0)
-An IDE to build and write actual programs for the uC to execute (I used Arduino)

The higher forces prepared several tutorials on installing and running EDBG, so I'll just add a link here. Once I loaded the trinketM0 bootloader, I could simply interface with my board through Arduino.

Link: https://mtm.cba.mit.edu/2021/2021-10_microcontroller-primer/edbg/
A few notes:
-Be sure to add the '-I' before specifying your file path to the hidapi libraries. '-I' tells the navigator to include all files within that directory.
- EDBG is fussy about which functions you can club together in the command line. For example, -p and -v (-progam, -verify) can be happily clubbes as -pv, but this doesn't work for all or any of the functions.
- If you want to check if your EDBG install has been successful, and that your uC is indeed alive and awake, simply plug it in and out over USB and check the 'Tools>Ports' option on Arduino to see if your board is being identified over /usbmodem101 and the likes.
- After you upload your arduino program, you might get a desktop notification asking you to 'Eject USB Drive before unplugging', as if you've accidentally disconnected your board while uploading. Ignore this and plough ahead, your install has most likely been successful.

Even though everything seemed to have worked correctly so far, my LED refused to light up. I tried:
- Changing the state of my button pin to INPUT_PULLUP (suggestion courtsey of Ido Calmam), which basically plays the part of having a pull-up resisitor hooked up to 3.3V on the input signal side of the button.
- Switching the direction of your LED. The head of the 'T'-symbol on your LED should face the + side. An easy way to rmember is to just convert the 'T' to a baby 't', with now the '+' (cross) going toward the + (anode).

I made a very simple model of a snitch.

I didn't have much time, plus, a super exquisite affair just isn't fitting for a non-champion, say what?
I then imported my model to SB Part-Works 3D to set-up the 3D toolpath. I split the model in half using a cutting plane and made a symetrical 2-part mold.
We were skeptical about how the small wavy features (~2mm thickness) at the lower edges of the snitch would turn out, but the toolpath render looked really good.
I started my milling job and it took about ~20ish minutes for the roughing path, and 12 minutes for the finish path.
Since I had a symetric 2-part mold, I had the choice to either mill twice (~35 minutes/run) or make my silicon mold twice (~ 3 hours/run (which includes time for pouring, mixing, panicking, casting, curing)).
The choice was obvious.

I used 1A:1B OOMOO and mixed both solutions individually before combining and mixing again for 5 minutes.
I had a large quantity of OOMOO, and though the mixture turned a promising shade of blue on the top and sides, after pouring I saw that it wasn't well mixed at the bottom.
After 75 minutes my molds were super sticky but I was still able to peel them out of the wax mould.
Defintely want to put more elbow grease and time into the next batch of OOMOO that I mix, and perhaps even consider making it in 2 bathes with smaller quantities.
I then did a post-cure for 4 hours at 65 degrees C, which made my molds a lot better, smoother and non-sticky. Post-curing is a great fix if you have the time, there is still hope.

I wanted to do a test first with a smaller part, so I used one of the exisisting turtle molds.
I cut a piece of a bath-bomb for a sacrificial insert. I then sanded it carefully on a very dull file to get it to the right size without fracturing/crumbling.
I put it into the mold and poured the metal, but the difference in densities of the two materials made it incredibly hard to keep in place udner the surface.

I then forecfully clamped the piece within the mold using a U-clamp.
For the snitch, I stuck the bath-bomb like a lollipop on a snitch and cut two openings in the silicon mold to position the sphere in the centre of the mold.

After pouring the metal, I let it set for about 1 hour before demolding.
I was scared the U-clamp might not come off easily, but it snapped right off the bottom surface of the mold cleanly.
One of my bath bombs actually rose to the surface of the turtle to give him a cool party-turtle vibe. I'm going to use this one as a bath-bomb holder and let them sink to the bottom of the tub for a slow release of scent. :)
The other one I cut and dissolved in water to get a nice empty cavity. I can attach hinges on these to get a secret chamber to put something fun like a note or a ring in the turtles belly!

But in my case I need this.

I don't fully have a grip on what components I need to operate the sensor, but in the spirit of spiral development I got started on the identifying the the barebones of of circuit. We have:
- ATSAMD21E18A microcontroller
- JTAG 2x5 header to load a bootlader via edbg and Atmel-ICE
- 5V to 3.3 V regulator
- Reset Button
- LED (always helpful to have one onboard for debugging)
- Resistors for pull-ups (typically 10k) - Capacitors to stabilize ground and power (typically 1uF works) - MAX30102 Pulse Oximeter
- 1.8V supply (5V to 1.8V or 3.3V to 1.8V divider)
The schematic on the left shows crucial connections. LucySD

I decided to make breakout boards for all the components on which I couldn't easily attach/solder jumpers.
Quickly broke out all my pins to 1xX headers (X = number of useful pins) and routed traces to the edges.
I milled and soldered them as shown alongside.
For the pulseoximeter, I used the reflow technique bu first tinning the pads, carefully placing the sensor on board, and then heating gently with heatgun at 230 degrees C. Caution: Check the breadboard you have in-house to make sure they have as many consecutive pins as on your breakout boards so you can actually stick the part on to the breaboard without having to use jumpers for each connection.

Breadboards have a notorious reputation with good EE engineers and I received several raised eyebrows from trusted sources for my pure bread-board prototyping approach.
So I decided to put the MC staples on single board and breakout the pins to attach peripherals in Development Kit fashion.
The potential for symmetry with the 4x8 MC pins was inviting, so I made my board in the shape of a locket that I intend to wear around.

After a few hours of patient milling, soldering and riveting, I had myself nice Trinket AS to attach peripherals to.
Ofcourse it wasn't that easy. Here is a short documentation on my suffering:
- Make sure your rivet size is big enough to get the rivet through and small enough catch the flange of the rivet.
- For Eagle users: mods is temperamental and won't resolve the legs of your microcontroller on a fairly complex circuit even if it has in the past. Fight fire with fire. Here are some absurd fixes:
* Changing the diameter of your trace tool AS WELL AS outline tool (from say 0.0156" to 0.0145" for mill and 0.0312" to 0.031" for outline)
* Changing the DPI from 2000 dpi to 1999 dpi or 1998 dpi.
* Making sure the traces leading to the pin come in exactly head-on to the pin, no slinking around the sides or staggering slightly from centre.
* Using a different CAD footprint with THINNER leg traces or LARGER gaps between the legs.
- MAke 3 separate files for traces, vias and outline and mill in 3 steps with outline being last.

After my board was ready, I spent a good amount of time poking and prodding the pins and pads to ensure I had continuity between the required pins and the right logic levels at the requisite locations.
I had lots of floating GND problems, so I added a bunch of flux and solder on the backside of my board at the rivets to ensure good contact with the copper. Because the back is just a vast sea of copper, is acts a heat sink and you have to hold your soldering iron in place for a while (5-10s) to ensure the solder mmelts and flows correctly.
I kept getting the following error on EDBG, and we were pretty certain it was a hardware problem.
"Debugger: ATMEL Atmel-ICE CMSIS-DAP J41800007177 1.0 (SJ)
Clock frequency: 16.0 MHz
Error: invalid response during transfer (count = 0/3, status = 7)
" I stuck my board under a microscope and looked at all the pins 1 by 1, finally realised that my MC grounds were connected to eachother but not to the USB ground.
Added a quick jumper and solved the issue. My eyes were finally graced with the much sought after message from the EDBG terminal.

So I have my board and a wonderful Arduino library that supportes the MAX30102 BUT nobody said that the breakout board for the MAX30102 already has a ton of functionality built in.
And here I am with my naked little barely reflowed MAX30102 sensor on breakout pins.
This sensor needs a very fussy lean diet of 1.8V in on the data power lines and 3.3V in for the LED lines.
I will now proceed to build my own breakout board, wherein the first step is to add an opamp that can shift the logic levels on the outcoming SDA/SCL pins from 1.8V to 3.3V or 5V.
I also used a generic voltage regulator (VR) with 2 resistors (1k and 2k ) to tune the voltage on the adjust pin of the VR and get a V-out of 1.8V.
David from CBA gave me a logic shifter as well, as I'm ready to start on hooking up the pulse oximeter.

Two hours of pouring over Grasshopper pallette and the smirk was wiped clean off my face.

Creating solid structures (or surface with thickness) on Grasshopper requires a completely different mind-set from Fusion/Solidworks. 'Extrudes' work only on lines and curves, creating hollow bodies. You then have to 'Cap' these which then fools the computer into thinking they are solid bodies. There are several ways to create patterns on surfaces with thickness. A few include:
- Creating your pattern on a flat 'sheet' and 'morphing/wrapping' it around the curved body
- Creating 'drill-bits' using capped extrudes that then pierce your target surface to a certain distance
- Subtracting one set of solid bodies from another

- Generating a random cloud of points within a 3D body that can be used to seed a pre-defined pattern (e.g. Vonoroi) and then performing a region intersection I used the second and fourth approach.

Finding the names of components required was an absolute treasure-hunt in flash of videos, written documentation, and people's past blogs

Finally, I had my part list down and squared from the fab library: - 1x ATMEL SAM D21 E18A microcontroller
- 1x 3.3V 1A Voltage Regulator
- 1x JTAG 2x5 Headerpin
- 2x Button
- 1x LED
- 1x Connector 05X2 ARM Debugger
- 2x 10k resistors
- 5x 0k resistors
- 2x 1uF capacitors
- 2x 10uF capacitors
Naming the airwires was SUPER helpful, prevented the schematic from looking, well, worse than a ratsnest.

Instructables to the rescue!

This terse little instructable got me everything I needed in terms of connections, Arduino libraries, speed control and microstepping.
I used a DRV2885 driver and set the potentiometer to about 1/4th the way up to control the current in the circuit. I additionally set a current limit of 1A on my 12V power supply. This was sufficient for rotating my ~2kg canopy with the load in the axis of the motor shaft.
https://www.instructables.com/Nema17-Stepper-Motor-Microstepping/
My motor ran 200 steps/rev and this resolution was sufficient to get a fairly smooth and slow motion for my application. I set a speed of 50 in the units of the arduino library. I then fabriacted my own coupler to fit onto the shaft. I gave it some holes to accept 16G wire from the wire frame on my revolving reel and 2 thru-holes for 6x32" set screws to grip the shaft
I then tapped threads manurally using a fractional tap - PLA is a nice soft and unproblematic candidate for thread tapping.

I used a python compatible UI library to generate my user interface

This created a simple interactive window using Python (HR_UI.py - attached) where the person inputs their name and presses a button to start their LUSCYD log.
The data is then logged over PySerial through an OS call to another script called HRoverSerial.py (script attached). HRoverSerial logs data for N samples (30 seconds in this case) and calculates the fft of the datasets every 5 seconds.