YOU CAN BUILD A RIG THAT PULLS GLASS FIBRE THINNER THAN A HUMAN HAIR, FOR 556 EUROS
A bench rig that pulls telecom fibre down to 8 micrometres in seven seconds, using a plasma arc and two linear stages you can buy on AliExpress.
by L.F. Granados-Zambrano, J.P. Korterik, J.M. Estudillo-Ayala, R. Rojas Laguna, D. Jauregui-Vazquez, H.L. Offerhaus, J.A. Alvarez-Chavez
ScienceOpen-hardware
Built withArduino
- difficulty
- ●●●●●
- time
- a weekend-plus
- license
- GNU GPL
- repo
- repo FINISHED0 stars
●●●●● · a weekend-plus · GNU GPL · 0 stars · repo FINISHED
WHAT YOU’LL NEED
Jump to section
COMPAREE VERDICT
This is a published research instrument, not a weekend Arduino project. It stretches standard single-mode telecom fibre down to an 8-micrometre waist by heating it with a plasma arc while two NEMA 23 ball-screw stages pull it. The reason to build it is speed: in the paper's comparison table a ceramic microheater takes about 4 minutes per taper, flame brushing about 3 and an electric arc about 1; this rig takes 0.12 minutes, roughly 7 seconds. A 5 mm taper took 3.6 seconds and a 10 mm taper 7.3 seconds, and the coefficient of variation across 10 tapers was 7.3 percent. The parts come from AliExpress, Amazon and standard suppliers, plus a few small machined aluminium parts, for 555.94 euros in total. The paper is the build guide: it gives the wiring, the stage layout (100 mm apart), the electrode gap (6 mm) and a table of pulling settings. You are working with high voltage, though: a ZVS Tesla-coil flyback driver produces more than 10 kV across the tungsten electrodes, and the paper insists on grounding everything to mains earth and isolating the HV electrode post. If you are setting up a photonics lab and need tapered fibre for sensors, this is a documented, working design from a real optics group. If you have never handled bare fibre or high voltage, start with something simpler.
IN THE REPO
GOOD TO KNOW
- —Design files are on OSF (DOI 10.17605/OSF.IO/VUW4J), the article is open access in HardwareX (DOI 10.1016/j.ohx.2024.e00578, CC BY 4.0).
- —The Arduino code is GNU GPL v3, the process video CC BY 4.0 and the circuit diagram under a CERN licence; commercial use is allowed under their terms.
- —Bill of materials is complete with supplier part numbers and a total of 555.94 euros.
- —Arduino code is provided; there is a wiring diagram and a fabrication parameter table.
- —The paper is the documentation — there is no separate build guide or forum.
- —The paper says the tungsten electrodes have prolonged durability but gives no figure for how many tapers they last.
Parts to buy
11 itemsFrom our check of the build. Exact quantities and part numbers are in the creator’s BOM.
- Two NEMA 23 ball-screw linear modules with motorsFind
- Arduino UNOFind
- Two TB6600 stepper driversFind
- ZVS Tesla-coil flyback driver boardFind
- Two tungsten TIG electrodesFind
- Two 12 V XP Power suppliesFind
- 2-channel relay boardFind
- Neodymium magnetsFind
- Machined aluminium V-groovesFind
- Bases and electrode holdersFind
- Aluminium and a plastic enclosureFind
Can I build this?
Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.
More builds like this
All projectsGallery
Start here
Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.
- 1.Read the HardwareX article in full (The article is the build guide, with step-by-step assembly and operating instructions; videos are in the OSF repository.)
- 2.Download the design files from OSF (Arduino code, circuit diagram, machining drawings for the small aluminium parts, and process videos.)
- 3.Source the two NEMA 23 linear ball-screw modules first(They are the longest lead-time item and more than half the cost.)
- 4.Learn fibre preparation: stripping the coating(Strip the coating from the section you will taper (see the fibre preparation video on OSF), then hold the bare fibre with the neodymium magnets in the V-grooves.)
KNOWN ISSUES
- Set the electrodes 6 mm apart at the same height as the top of the V-grooves (paper step 7, Fig. 22), and use a piece of fibre to align the gap with the grooves; small misalignments and air currents change the waist.
- The ZVS Tesla coil driver (12-36 V input) generates >10 kV at the electrodes — follow the paper's wiring diagram exactly, ground it to mains earth, and isolate the HV+ electrode post as described.
- The paper says the tungsten electrodes resist the discharges well, but it gives no number of tapers per electrode pair; they cost 1.70 euros each, so keep spares.
- The 7.3 percent repeatability is measured across 10 tapers after calibration — your first few tapers will not hit that number.
- There is no community forum or build log; if something does not match the paper, you are on your own.
- If you have never cleaved or handled bare fibre before, this is not the project to learn on — a botched cleave or a snapped taper is immediate and you will not know why it failed.
Can I use a different plasma or arc source?
The paper tested this specific ZVS Tesla coil flyback driver and these tungsten TIG electrodes; a different arc source would need re-characterisation of the heating profile and pulling speed.
What kind of fibre does it work with?
The paper uses standard single-mode fibre (SMF) and does not name a specific type. Other fibre types are not tested, so expect to re-tune the settings.
How do I know the taper waist diameter without an SEM?
The authors measured tapers optically: a machine-vision camera on a 10× microscope objective gives about 0.7 µm per pixel, enough to see an 8 µm waist. A cheap USB microscope will not resolve it reliably; transmission loss is a useful secondary check.
Is there a way to avoid the high-voltage driver?
The plasma arc is what makes it fast. In the paper's Table 1, a ceramic microheater takes about 4 minutes per taper, flame brushing about 3 and an electric arc about 1. All of the fast heat sources involve high voltage or open flame.
Community builds
No community builds yet — be the first, we feature the best ones.
Discussion1
FROM THE COMPAREE TEAM
About seven seconds per taper, versus roughly three minutes for flame brushing or four for a ceramic microheater. What would you use tapered fibre for if you could make it that fast?
L.F. Granados-Zambrano, J.P. Korterik, J.M. Estudillo-Ayala, R. Rojas Laguna, D. Jauregui-Vazquez, H.L. Offerhaus, J.A. Alvarez-Chavez
A collaboration between the Optical Sciences Group at the University of Twente (Netherlands), Universidad de Guanajuato and CICESE (Mexico). They published the rig in HardwareX as a low-cost, open alternative for making tapered fibre for sensors.
DISCLAIMER
- Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
- Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
- Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
CompareeTEAM26d agoedited
Practical notes from our verification: this is a working rig published in HardwareX by optics researchers from the University of Twente and the University of Guanajuato, not a hobby project — the bill of materials is complete with supplier links and totals exactly 555.94 euros as stated, with the two ball-screw linear stages making up 314 euros of that. The Arduino code, circuit diagram and videos of the tapering process and fibre preparation are in the OSF repository, and the paper has the full build instructions. There are no 3D-printed parts: the V-grooves, base and electrode holders are machined aluminium, with drawings in the repository. The tungsten TIG welding electrodes cost 3.40 euros for the pair, but the paper does not say how many tapers you get before they need replacing. The speed advantage is documented in Table 1 of the paper, and the authors report 7.3 percent repeatability across ten tapers — expect test runs and calibration before you match that. Correction (4 October 2026): we re-checked this page line by line against the project's own repository, documentation and videos, and fixed errors in earlier versions.