Ataary’s pink and cyan illustration of retro game consoles, controllers, and handhelds
SeaGL 2026
Share the dead ends too
A hands-on argument for sharing your failures
Trevor Johnson · johnsonfarms.us · ataary.com
Why share the failures
Share the failures, not just the wins.
Open source and open hardware run on shared knowledge, but most of what we share is the finished thing that works. Document the attempts that didn’t, share them too, and:
You learn more
Writing it up makes you explain what actually went wrong.
You have a record
No relying on memory. The fix is often sitting in your own notes.
New ideas spawn
A failure raises questions you wouldn’t have asked otherwise.
Other uses appear
What failed at one job may be good at another.
Open knowledge grows
The next person starts where you stopped, not from zero.
This talk isn’t really about a flashlight.
It’s my example of all five.
Version one
This one barely lit. I posted it anyway.
Version one: a pink printed flashlight with black conductive bands on its end cap, next to a CR123A battery
On Printables
Flashlight with Conductive Filament - Limited Success - Learning Experience
“An attempt at a simple flashlight design using conductive filament achieving limited success and mostly failure.”
Write-up and files: printables.com/model/425558
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Version one
Print the wires in
PrusaSlicer preview of a version one part, showing the color change partway up where the conductive filament startsPrinted version one caps with the dark conductive band visible in each
A color change in PrusaSlicer swaps in conductive filament partway up each part. That dark band is the wire, and twisting a cap is the switch.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Version one
Round after round of not quite
Five version one iterations standing side by side, pink with black conductive stripes
Round one
Two AAs, 3 volts, and a 3 volt bulb. It barely even turned on.
Then more of everything
  • Three CR123A cells, 9 volts
  • A 5 to 24 volt LED bulb
  • Thicker walls and caps
  • A 60/40 split, regular to conductive
  • A flat battery contact, not a sphere
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Version one
Technically, it made light
Version one lit in a dark room: a faint point of light against a wooden surface
“If I was stuck in a mine and had nothing else, sure, but a match or lighter would be brighter.”
350
ohms along the body, even with thick walls
500 to 3,000
ohms across the whole light, depending on how the threads and caps bite and line up
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
For reference
Why resistance matters
V = I × R
Voltage = current × resistance
Flip it: I = V ÷ R. More resistance, less current, dimmer light.
Copper wire the length of this flashlight is a tiny fraction of an ohm. This plastic measured hundreds to thousands.
The LED takes its cut first
A white LED needs about 3 volts just to turn on. Two AAs left almost nothing to push through the plastic.
Then the threads decide
After the LED’s 3 volts: 6 V ÷ 3,000 Ω is 2 mA. 6 V ÷ 500 Ω is 12 mA. Six times the current, from how the caps lined up.
Resistance becomes heat
Current through resistance warms the part instead of lighting the bulb. Hold that thought for the fuses.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Version one
What the failure said
The lesson
“Any design that requires the current be carried any distance will suffer from resistivity.”
My note for next time
“I would use an alternate conductor along the length of the body and only have conductive ends, or use batteries with conductors only on one end.”
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Version two
A 9-volt battery has both terminals on one end.
The fix was already written down, in the post about the failure.
A 9-volt battery with both its negative and positive terminals on the top end
Version two
Shorten the path
A small conductive cap sits right on the 9-volt, its lug on the negative terminal.
The bulb rides in a conductor that threads into that cap, over the positive terminal.
Screw it down until the LED touches positive: on. Back it off a little: off.
Version two parts: a yellow body and screw cap, the black conductive cap and bulb conductor, and a 9-volt batteryVersion two lit up on top of a copy of What If?
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Version two
Same 9 volts. A much shorter path.
Version one
Version one lit in a dark room, a faint point of light
Current crosses the whole body and every threaded joint. 500 to 3,000 ohms, so 2 to 12 mA at best.
Version two
Version two throwing a bright beam across the cover of What If?
Current crosses just the cap and the bulb conductor. About 300 ohms or less, and 15 to 20 mA measured.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Version two
The small lessons, written down too
100% infill
On the conductive parts. 5% and 15% were fine on the rest.
Layer 15
An optional color change there saves about 2.5 grams of conductive filament.
It is DARK
Purging takes a while. I print something small afterward to clear it.
It really sticks
Hard to remove from smooth plates, and it fills in textured ones.
Tight threads?
Print the body at 98%. Print setups vary enough to bind.
Weak conductor
The first bulb conductor was fragile. A stronger one is in the files now.
Optional printed diffusers lit up on the printer bed: a bulb shape, a globe, and a tall tube
Optional diffusers, posted too.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Where it led
The failure pointed somewhere new
What the flashlight taught me
This plastic resists current. A lot.
The question it raised
Resistance makes heat. Could that heat be useful?
The next experiment
A fuse that melts itself open when the current gets too high.
Writing up the dead end is what turned it into a starting point.
The fuses
Five fuses, measured
Standard 5 × 20 mm barrel fuses, printed in conductive PLA.
The five printed fuses on a blue background, labeled 1x1, 2x2, 3x3, Hollow, and Side Hole
Resistance by center shape:
1 × 1 square
1,250 ohms
2 × 2 square
500 ohms
3 × 3 square
350 ohms
Hollow
300 ohms
Side hole
300 ohms
All 8 mm long and printed flat. The hollow one passed the most current.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
The fuses
Terrible fuses. Decent resistors.
The bench test: a power supply reading 31.03 volts and 0.035 amps, a multimeter, a drill, and a printed fuse in a fuse holderA multimeter reading 349.5 ohms across a printed fuse in its holder
Over 30 volts barely lit an LED on a drill.
In line with my truck battery, it didn’t blow, and the truck didn’t start.
Left as a short on the battery, it wasn’t even warm, and it crept up to just over 600 ohms.
It resists so much that it chokes off the current that would melt it. But that steady resistance makes a decent resistor for LEDs and buzzers.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
The straight answer
Conductive PLA, honestly
Good at
  • Short paths
  • Parts that are the contact and the switch at once
  • Resistors for LEDs, buzzers, and small projects
  • Sticking to the print bed, maybe too well
Bad at
  • Carrying current any distance
  • Anything that needs real current
  • Conducting between layers
  • Steady contact through threads
  • Being a fuse
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Your turn
Why we don’t document or share our failures
“It’s not important. Nobody will care.”
You can’t know who needs it. My barely lit flashlight led to a working one, a fuse experiment, and this talk.
“Nobody will see it.”
Views aren’t the point. If the only reader is future you, it still did its job.
“It’s embarrassing.”
An honest title is enough.
“It isn’t finished.”
Nothing open ever is. Write it down now, link forward later.
“It’s not that kind of project.”
Code, circuits, recipes, research. It all counts.
If it taught you something, document it.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Your turn
How to document a dead end
1
Title it honestly.
If it mostly failed, say so right in the title.
2
Say what you tried, and why it seemed like a good idea.
Printing the wires in made sense on paper.
3
Give the numbers, even the bad ones.
350 ohms told version two what to fix.
4
Say what you would try next.
“Use batteries with conductors only on one end.”
5
When it works, go back and link forward.
Version one now opens with “Update: Success!” and a link to version two.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Your turn
Where I share mine
For reference. Every name is a link.
Site
What I use it for
Platform
The whole story, on my own site
Open source (GPL)
Models, print settings, photos
Proprietary
Build logs, failures included
Proprietary
Electronics notes and guides
Proprietary
Code, READMEs, and issues
Proprietary (Git is open source)
Research data, with a permanent DOI
Open source, run by CERN
Most of mine are proprietary. Whatever the platform, your license is still yours to choose.
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Your turn
Worth a look
For reference. Every name is a link.
I’d recommend, and they’re open
Code, issues, and Pages. A GitHub alternative.
Open source (Forgejo), nonprofit
A wiki for open hardware projects.
Open wiki, CC BY-SA content
Also made for projects like these
Site
Good for
Platform
Electronics and PCBs
Open source
Code
Open core, self-hostable
3D models
Proprietary
3D models
Proprietary
Step-by-step builds
Proprietary
Electronics projects
Proprietary
Share the dead ends too · Trevor Johnson · CC BY-SA 4.0
Try something. When it fails, document it anyway.
Share it, and it becomes open knowledge.
Slides licensed CC BY-SA 4.0 · Trevor Johnson
Trevor’s 8-bit portrait from ataary: a bearded man in a backwards cap and glasses, holding a game controller