Sebastian’s Party Geometric Jacket

Sebastian has commissioned us to re-make his favorite hoodie/jacket with controllable/programmable lines of light on it that can light up to form geometric shapes and patterns.

Party Geometric

documentation on github >>

Progress Blog

The Pattern – to copy or re-make?

Since taking a closer look at Sebastian’s jacket I’ve also seen other people wearing a similar design but constructed differently. Some from non-stretch fabric. And some with a more defined hood.

Very similar:


We’re trying to decide if we should re-make the pattern to be more fitted around the shoulders, or to keep it closer to the original design with no defined shoulders and more of a “wrapped up in a blanket feel” as the shoulders slouch over the shoulder and cling to the body because of friction.

Original shoulder slouch:

Toile made up in non-stretch calico fabric. Not sure if it will feel/look good in non-stretch? Might need to solve it with a more fitted sleeve, but then we loose the blanket feel.

One concern is that we’re not sure how this slouch looks will feel and look work in a non-stretch fabric. The reason to use a non-stretch fabric would be because we intend to cut the pattern up into geometric pieces and insert the fiber optics and circuitry along these seams, which makes sense to do in non-stretch.

This is an example of a similar pattern but with a very fitted should area. Could this be an option?

This is just an example to show the use of very light/translucent fabric, which might be an interesting solution considering a design that causes you to sweat less and needs to be washed less too.

3D Geometry?

Next steps, ideas…. not this extreme, but to add some geometric shaping to the jacket…


Translucent fabric test

Mesh light test


We’re starting by making samples to try out how to embed fiber optics into the seams. As well as selecting the material we will re-make the jacket from and how to reconstruct his jacket’s pattern either through tracing the original or constructing it based on a shirt pattern.

Fiber optic seam line tests
Fiber optic seam line tests
Fiber optic seam line tests

“Bumblebee Ring”

For Sebastian’s partyGeometricJacket we are segmenting his jacket into triangular polygons. the 3mm diameter fiber optics will run along the seams of these polygons and be individually addressable.

We started out refining the jacket pattern so that it would be well fitted. We then then marked the polygon pattern directly onto the jacket and transferred these lines (which will all become seams) back to the flat paper pattern pieces. In doing so we started to think in more detail about how to run the circuit through the garment. The neatest solution we can think of is to run the traces (power, ground and data) through the polygon seams. At first we imagined we would use something similar to individual Bumblebee connectors for connecting the conductive threads running through the seams to the LEDs at the ends of the optic fibers. We are currently counting about 40 LEDs total.

Pattern from KOBAcam

But in making first prototypes of the polygon-seam solution and drawing up the polygon pattern up on paper, we realised two things:

A) There is no unique path through all the polygon seams that only passes every intersection once (which would be necessary for electrical connections to not touch).

B) That the Bumblebee (optic fiber to LED connector) covers the tube, meaning that if the connector wants to sit directly on the seam, then the lit optic fiber can not extend all the way into the corner(s) of the triangles.

Solutions to the above could be:

Instead of spreading the LEDs around as much as possible, to instead cluster them around as few junctions as possible. And then design “Bumblebee Rings” that solve all the LED connections on-board. See bellow illustration.

Bumblebee Ring sketch

Bumblebee Ring sketch

The optic fibers could bend/curve over to the nearest seam…. but this does not solve problem A.

Bumblebee Ring

Instead, the Bumblebee could sit on an extended end of the optic fiber, meaning electrical connections from the seam would have to extend past the seam. An opportunity for a unique connector design – the “Bumblebee Ring”

Bumblebee Ring dev

Bumblebee Ring layers

Sebastian’s 3D printed resin Neopixel to E-Textile connectors

Version 2:
Sebastian's 3D printed e-textile Neopixel connector
Sebastian's 3D printed e-textile Neopixel connector

Version 1:
Sebastian's 3D printed e-textile Neopixel connector
Sebastian's 3D printed e-textile Neopixel connector
Sebastian's 3D printed e-textile Neopixel connector
Sebastian's 3D printed e-textile Neopixel connector

Reflective End Caps

Reflective caps at the ends to reflect the light back into the tube!
Fiber optic WITH reflective cap
Fiber optic WITHOUT reflective cap

Flex PCB Ring Mount (26.4.2018)

The neopixel SMD mini LEDs are mounted vertically on flex PCB. The top side connection is brought to the side with copper tape. This is only possible with hand soldering, and not suitable for manufacturing.. This test version has 3 LED+tube placed in 120 degrees angles. (they are equally angled) We used 3mm brass tube to fix the optic fiber. The tube is soldered onto copper foil underneath. +,-,dataIn, dataOut connection of the ring is broken out as copper foil pad with 2x 1.2mm hole so one can sew the connection like buttons. The next step will be to see if this design can be used together with seam conductive connection channel.
ring mount test

“Party Button Circuit” (27.4.2018)

Today we prototyped a double sided PCB version of the “party button circuit”, and it looks like it could work really well.

Simulated Double Sided Hard PCB Version:


Cut file:

This version has little tabs on it for helping to fix the fiber-optic ends tube recepticals with shrink-tube to the PCB, but of course these will not be necessary if the 3D print is designed with the fiber-optic holding included.
If we have these boards manufactured, they can probably get a few mm smaller. The minimum size will probably be given by the size and spacing of the holes for sewing.

Next we will go back to the pattern to re-look at all the LED clusters and see what kinds of angles and how many there are total. One last check to see if we can’t find that unique path through the pattern that would allow us to only have one LED at each junction…

PCB + Modeling (28.4.2018)

Sebastian came to the shop today and we went over the last “Party Button Circuit – Simulated Double Sided Hard PCB” version and discussed how the PCB and 3D print will come together.

While Sebastian worked on the 3D model, I put together the PCB in Eagle to start working towards the final dimensions and a file/format that we can order manufactured PCBs.

I made a side-mount footprint for the mini Neopixel:

For which I needed to look up the LED dimensions in it’s datasheet…

WS2812B MINI – 3535 RGB LED mit integriertem WS2811 LED-Treiber IC:


In EagleCAD I made up the schematic:

And laid out the circuit:

And exported the DXF file for Sebastian to mill on his CNC mill…

WS2812BMINI vs. SK6812MINI ???


WS2812B MINI datasheet

WS2812B MINI datasheet

SK6812MINI (similar?)


The joys and pleasures of switching from EAGLE to KICAD

But I made a mistake in the dimension of LED. 5mm instead of 3.5mm…. and this was the point i really started to reconsider switching to KiCAD (open source and free PCB CAD software). No size or layer restrictions, plus open community and non-commercial values. So, I made the decision and for the last two weeks have been VERY slowly learning the joys and pleasures of switching from EAGLE to KICAD.

First, because I knew I would need to construct a custom footprint for the LED to be mounted on the edge of the PCB, I looked into how to create custom footprints in KiCAD’s footprint editor. This has it’s quirks of active libraries and netlists and updating netlists…. but it all works and makes sense.

How to create, load, activate and save libraries:

Sparkfun beginner’s guide to KiCAD:

How to create custom footprints with pads on either side of PCB:

KiCad Tutorial #6 Create PCB footprint component:

Kicad Tutorial 12/2014 – Part 7/12 – Custom PCB Footprints:

But then when it came to constructing the correctly dimensioned footprint from the dimension drawing in the datasheet, I got really frustrated by not finding a simple way to work precisely. In the end I started asking questions in the KiCAD IRC/Matrix room and found out that a footprint for the SK6812MINI already exists in this LED_SMD library:

This allowed me to skip the exact dimensioning and instead I just worked off the existing (trusting it is correct). Then also had to figure out how to move a solder pad from top copper layer to bottom copper layer in footprint editor. Could not find flip function but easy enough to do in edit window (hover over part + e) where you can select the layer.

O! The joys of KiCAD

Footprint complete, the next step is to find a way to import the board outlines, which I had already drawn in Illustrator. KiCAD has:
File —> Import —> DYF File
…but DXF files exported from Illustrator are not “the right kind of DYF” so nothing appears.
Instead tried following tutorials that showed how to export DXF from Inkscape… but same thing here… nothing appears.

In the Matrix room I am told to use LibreCAD to export DYF for KiCAD and this works! But shape is distorted:-(

O! The joys of KiCAD

Go back to Illustrator and add in more anchor points, export as DXF. Import DXF to LibreCAD and it looks good!

O! The joys of KiCAD

O! The joys of KiCAD

O! The joys of KiCAD

Import into KiCAD and it looks perfect!

O! The joys of KiCAD


But then!

DXF exported from Illustrator uses bezier curves to define shape. These take up memory resulting in larger file size. This large file size (14MB) slowed down KiCAD extremely so that it was basically unusable.

O! The joys of KiCAD

To bring down the file size by eliminating the use of bezier curves I added many anchor points to the curve:
Object —> Path —> Add Anchor Points

And then turned the bezier curves into lines:
Object —> Path —> Simplify

Then export this as DXF. Open in LibreCAD. Export as DXF R12. In KiCAD:
File —> Import —> DXF File


O! The joys of KiCAD

An alternative way of doing this could have been to use this board-generation plugin that a user in the KiCAD IRC/Riot chatroom showed me a screenshot of:

O! The joys of KiCAD

Then I need to go through all the LEDs and rotate them to the exact same angles as the outline.
Hover over the LED + “e” –> opens edit window. To rotate enter the angle *10! 63 = 630!

O! The joys of KiCAD

O! The joys of KiCAD

The First Board!

I got the first board layed out and it looks… well… interesting.
it is definitely going to be a lot of handwork for each board. and I have to let go of wanting to control the shape of the traces and just go with what KiCAD kind of does of it’s own accord.

O! The joys of KiCAD

Am a bit worried about:
– is the footprint of the LED really going to work well – we have no way of testing this before ordering.
– will the the cutout for the LED be wide enough given the diameter of the milling bit? (i should be able to check this in the design specs of manufacturer)
– on the boards that have very close LEDs, will there be space for the traces and vias?
– will the soldermask coating hold up as insulation against the conductive thread sewing between the holes as there are traces laid between these holes.

O! The joys of KiCAD

O! The joys of KiCAD

PCBs Ordered (01/09/2018)

After a nice long summer break, have FINALLY finished laying out all 14 PCBs for Sebastian’s jacket and placed the order with Elecrow

– total panel size: 42 x 152mm
– number of panels: 5 (or less – we only really need two)
– different PCB designs: 14
– copper layers: 2
– PCB thickness: 1mm
– solder mask: matte black (on both sides)
– silkscreen: none
– surface finish: HASL lead free
– Castellated Hole:No
– Copper Weight:1oz
– PCB Stencil:NO
– Production Time:4-7 working days
– Country:Germany


WS2812B MINI – 3535 RGB LED mit integriertem WS2811 LED-Treiber IC:

1uF 0603 Capacitor:

PCBs Ordered (08/10/2018)

i soldered up one board and noticed one embarrassing mistake. the footprint was designed for 3mm thick board, and in my head i mixed up lulu&partygeo just before placing order. so with 1mm thick board there is a bit of a gap between LED pads and solder pads, but it is still easy to solder. stencil for holding LED exactly half-way would be nice.

Assembled Party Geometric PCB:

Meet with Sebastian (12/10/2018)

Try on the assembled jacket minus sleeves:
Arne's gloveless datagloves
Arne's gloveless datagloves

Talk about 3D print solution for the PCBs.
here the photo of the angles of the LEDs on boards:
Party geo pcb angles

the plan is to solve all LED to optic-fiber tube connections in single 3D print per PCB.
you will do some 3D printing experiments to see if:
– this could be done in flexible filament so that it simply pulls over the PCB.
– otherwise 2-piece hard versions that click together.
– we try to avoid using any glue, for the case that we have to replace a broken print at any time.

we aim to meet again early November.

Sebastian Fiber-Optic Coupling Design V1 (8/11/2018)

This week Sebastian emailed us renderings and photos of his first tests for the optic-fiber coupling to his party geometric button PCBs.

Party Geometric optic-fiber connection tests

Party Geometric optic-fiber connection tests

Optic-fiber 3D print coupling...

Overview created for Lulu:

Sebastian Fiber-Optic Coupling Design V2 (4/12/2018)

Going over Sebastian’s 3D printed PCB buttons that connect the ends of the optic-fibers to the LEDs.

Sebastian Fiber-Optic Coupling Design V3 (11/12/2018)