Skip to content

Instantly share code, notes, and snippets.

@miloh
Last active November 18, 2015 01:13
Show Gist options
  • Save miloh/9630058497ddc704c771 to your computer and use it in GitHub Desktop.
Save miloh/9630058497ddc704c771 to your computer and use it in GitHub Desktop.
Koruza: notes from a talk given by Luka Mustafa. Noisebridge, 17 Nov 2015

Koruza, a OSH/FLH FSO (free space optical) link solution by IRNAS

Creator luka mustafa gives a personal introduction

2015 shuttleworth fellow

other 2015 shuttleworth fellows

Luka works on a variety of projects

  • koruza.s
  • open hackerspace/fab lab tools
  • documenting open hardware (Q: what definition used for ohw?)
  • phd on the side :-)

Background of Koruza

2012- 2015 development history of koruza.

FSO never scaled for mass use. used by carrier, military, enterprise for 10 years or so.

Optical tranceiver development is still not widespread or accepted tech Current solutions too expensive for consumers, haven't acheive economies of scale or prove use cases in a specific sector.

There simply isn't a good last mile solution

Koruza Design

All slides are available online. Hunt for them on koruza dot net Koruza in the slides looks like 3d printed toy with some linear rails and a power unit. there is some attempt to make a hermetic seal with the O ring visible on the front o f the interior unit The outside

SFP FO modules...

the system inlcudes manual mounting system, sliding system, and inner motorized.

currently no part for continuious motion.

mike k asks about schelrian distortions.

slow tracking/positional only. currently no fast tracking

interesting to note there isn't a seperate channel for control systems of each module to communicate with each other, though this is in the works via modulation of the visible green alignment laser

following slides display the first prototype, second proto, proto two.

notes on design changes

  • water resistant by prototype two.o
  • square enclosure is from stock parts and cut to shape.
  • continually developing and documenting the prototypes: protootype gen 4, protype gen 5.
  • gen 5 designed in openSCAD, still finds ways to reduce material cost
    • installed 10 links around the world.
    • still issues to solve with proto gen 5

interesting results from testing with fog, using connectivity corralated with time/date to get a suggestion of issues with thermal expansion . project source is on github under IRNAS

other projects:

  • Also works on GoodEnoughCNC
    • fiberoptic control for the stepper motors in goodenoughcnc
    • some kind of plasma cutter example at 700GBP...
  • "10k fab" includes plasma cutter, cnc mill (2x3m), laser cutter 100W (2x3m), 3D printer, Waterjet cutter.

parts check

  1. motion and enclosure components:
  • smooth rods, estimate $40
  • simple bushings estimate $10
  • 3D printed parts no more than 500g, $20 in raw material max
    • estimate ~8 hours print time.
  • rubber (used as gimble bed), o-ring
  • fasteners, springs, small parts $10
  • stock Al enclosure $10
  1. cots electronics
  • tp-link SFP switch $40 (amazon)
  • TP-LINK MC220L Gigabit Media Converter, $25usd range. (amazon)
  • rpi 2 $40
  • green (532nm?) laser pointer
  1. bespoke electronics
  • media converter breakout boards (2 pcbs, smt)$15
  • 3 axis motion control board (connects to rpi) $15
  1. optics
  • lens (cost unknown, estimate <$50usd)
  1. other components and tools (used for calibration and operations)
  • infrared detecter paper, cost unknown, (used for alignment at distance, usable only in dark)
  • infrared optic sensor $200

bom total: ~$250, less any tools.

my own ideas for future design improvement:

DFHM (design for hacker manufacture)

  • improved 3DP materials. (ultem, nylons, or possibly pet)
  • more COTS enclosure parts to reduce cost and 3DP requirements

QA (quality assurances)

  • active cooling to reduce thermal expansion
  • active temp sensing for better data colleciton from field reporting

Range Boosting

  • active Vdd on the media converter breakout board, check on increased gain
  • QA and bin SFPs for alignment issues (small issues that might pass QA for standard use might be unacceptable for use with FSO at long range)

Reduce BOM cost

  • find OEM for media converter, SFP unit, work out oem purchasing and integration
  • use mchck or allwinner SOC based board for motion/system control
  • farm out 3DP for simpler it.
  • release electronics/optics only kits

Q&A

Juan talks from experience about the lack of a viable last mile solution - points out solutions are literally no more than 1/10th of the 'last mile'. range was always an issue -- beyond 100M range the signal falls off steeply

Questions about cost. Unit cost starts at 700EUR from Fabrikor

Wrapup

Attendees got to ask Luka some questions as he broke down and packed up the the sample units, which were running a short link across noisebridge during the talk, a distance of about 100 feet, or 30 meters.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment