Skip to content

Instantly share code, notes, and snippets.

@eonist

eonist/frame-build.md

Created March 12, 2026 20:26
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save eonist/a444f4a4a7f550c5f21c1058b480fd54 to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save eonist/a444f4a4a7f550c5f21c1058b480fd54 to your computer and use it in GitHub Desktop.
Download ZIP
frame-build.md
Raw
frame-build.md

That's a smart move. The Detroit is the spiritual reference, but the Omera needs its own visual identity. A seat tube cutout is a strong signature — instantly recognizable and signals "this is a track-bred frame" without copying Dosnoventa's exact look.

What Makes Omera Visually Distinct from Detroit

Feature Detroit Omera
Seat tube Standard round Cutout wrapping rear wheel ✅
Down tube Columbus Airplane ovalized 50mm round (battery inside)
Drive Chain Gates CDX belt (visible difference)
Brakes None (track) or rim Hydraulic disc
Rear dropout 120mm track 135mm horizontal/sliding
Motor None Bafang G020 (stealth hub)

The cutout + belt drive + disc brakes together give the Omera a completely different silhouette from the Detroit while keeping the same aggressive 74°/76° pursuit geometry. Someone who knows bikes would clock it immediately as its own thing. ppl-ai-file-upload.s3.amazonaws

Design Consideration for the Cutout

One thing to flag for the frame builder — the Omera has 135mm rear spacing (not 120mm track), so the seat tube cutout needs to be wider than a typical track frame cutout. The wheel sits in the same plane but the hub flanges and disc rotor are wider. Make sure the cutout clears:

  • Tire (700×25c) + 3mm each side minimum
  • Disc rotor on the non-drive side
  • Belt on the drive side

This is doable but the frame builder needs to know upfront. Want me to add this to the frame fabrication brief?

@eonist
Copy link
Author

eonist commented Mar 12, 2026

You're spot on about both junctions. Looking at your renders — the model is coming along nicely with the oval DT clearly visible. Let me break down both stress points.

BB Junction (Down Tube + Seat Tube + Chainstays)

This is the most stressed joint on the entire frame — it handles pedaling forces, road impacts, and now the weight of a 1.4kg battery pack above it. Your render shows it clearly: the 60×55mm oval DT meets the round seat tube and two chainstays all at the BB shell, leaving very little room. alibaba

The good news: the oval DT actually helps here. The wider profile gives more brazing/welding surface area at the BB shell interface, which spreads the load better than a small round tube would. But you need to think about: [velo.outsideonline](https://velo.outsideonline.com/road/road-gear/masterclass-lugs-and-lugged-framebuilding-with-llewellyn-custom-bicycles/)

Head Tube Junction (Down Tube + Top Tube)

You're right that this junction is more spacious — the head tube is typically 40–44mm diameter, and the two tubes meet it at different heights. But it's still the second most critical stress point, especially the underside where the DT meets the HT. velo.outsideonline

Key considerations:

  • Round the underside lug point — this is where Llewellyn and other master builders emphasize that a sharp lug point on the DT underside will crack the frame. Always a smooth radius velo.outsideonline
  • FEM-analyzed gusset plates between TT and DT at the HT, working like an I-beam — this is exactly what Starling Cycles does on their steel frames after finite element analysis proved it effective theloamwolf
  • More cable space: Your 60mm-wide oval DT opening into the HT area gives generous room for disc brake hose, light wiring, and antenna cable to enter/exit

Reinforcement Strategy

Junction Stress Level Reinforcement Cable Load
BB shell ★★★★★ Wrap-around gusset, rounded ends, 2mm CrMo Motor cable, PAS sensor, brake hose, battery wiring
Head tube ★★★★ Gusset plates TT↔DT (I-beam style), rounded lug points Brake hose in, light cable out, antenna cable
Seat tube top ★★ Standard sleeve/lug Electronics module connector
Dropouts ★★★ Thickened dropout plates Motor cable exit

Cable Routing Map

Everything runs internally through the oval DT as the main trunk:

HEAD TUBE (entry)
  ├── Front disc brake hose → down DT interior
  ├── Syla headlight power cable → down DT interior  
  └── Antenna cable → up to TT → seat tube
         │
    DOWN TUBE (main trunk, 60×55 oval = plenty of room)
         │
    BB SHELL (junction/exit point)
  ├── Motor cable → chainstay → rear hub
  ├── PAS sensor cable → BB mount
  ├── Rear disc brake hose → chainstay → rear caliper
  └── Battery wiring → BMS (in free 130mm DT space above BB)

The 130mm free space above the BB in the DT is perfect for the wiring junction — all cables terminate there, away from the battery pack, with room for weatherproof connectors. The oval cross-section gives you ~23 cm² of internal area at any point — more than enough for 4-5 cables plus the battery cartridge. ppl-ai-file-upload.s3.amazonaws

@eonist
Copy link
Author

eonist commented Mar 12, 2026

A torque sensor is fundamentally different from cadence — it measures how hard you push, not just whether you're pedaling.

The Physics

A strain gauge is bonded to the bottom bracket axle or crankshaft. When you push down on the pedal, the metal axle deforms by a microscopic amount (invisible to the eye). The strain gauge detects this deformation as a change in electrical resistance — harder push = more deformation = stronger signal.adventuro+1

text
Soft pedaling (50W)  →  tiny axle flex  →  small signal  →  gentle motor assist
Hard pedaling (250W) →  more axle flex  →  big signal    →  full motor power
Stop pedaling        →  zero flex       →  zero signal   →  motor off instantly

It's analog and proportional — not on/off like your cadence sensor. The motor literally amplifies your legs in real-time.kerenelectric+1

Types

Type | Location | How It Senses | Accuracy -- | -- | -- | -- BB torque sensor | Replaces your bottom bracket | Strain gauge on BB axle | High — measures force at source [bikone]​ Rear dropout sensor | At the rear axle | Strain gauge on dropout plate | Good — measures chain/belt tension [forums.electricbikereview]​ Mid-drive integrated | Built into motor (TSDZ2 etc.) | Internal to motor unit | Best — but you're using a hub motor

About €54 more for a dramatically better ride feel. Worth it on a stealth build where the whole point is "you can't tell it's electric."

A torque sensor is fundamentally different from cadence — it measures **how hard you push**, not just whether you're pedaling.

The Physics

A strain gauge is bonded to the bottom bracket axle or crankshaft. When you push down on the pedal, the metal axle deforms by a microscopic amount (invisible to the eye). The strain gauge detects this deformation as a change in electrical resistance — harder push = more deformation = stronger signal. adventuro

Soft pedaling (50W)  →  tiny axle flex  →  small signal  →  gentle motor assist
Hard pedaling (250W) →  more axle flex  →  big signal    →  full motor power
Stop pedaling        →  zero flex       →  zero signal   →  motor off instantly

It's analog and proportional — not on/off like your cadence sensor. The motor literally amplifies your legs in real-time. velotricbike

Types

Type Location How It Senses Accuracy
BB torque sensor Replaces your bottom bracket Strain gauge on BB axle High — measures force at source [bikone](https://www.bikone.com/bottombracket-torque-sensors/)
Rear dropout sensor At the rear axle Strain gauge on dropout plate Good — measures chain/belt tension [forums.electricbikereview](https://forums.electricbikereview.com/threads/bottom-bracket-compared-to-rear-drop-out-torque-sensors.33209/)
Mid-drive integrated Built into motor (TSDZ2 etc.) Internal to motor unit Best — but you're using a hub motor

For Omera: BB Torque Sensor

The only option that works with your Bafang G020 hub motor is a bottom bracket torque sensor. It replaces your BSA 68mm BB with one that has strain gauges built in. support.favoritebikes

Bikone BSA Torque Sensor

This is the most interesting option — purpose-built for DIY e-bikes: [bikone](https://www.bikone.com/bottombracket-torque-sensors/)

Spec Detail
Standard BSA (fits your frame) ✅
Spindle compatibility 24mm hollow (Shimano), square taper, ISIS
Measures Torque + cadence + crank angle + backward detection
Output signal UART / digital / pulses
Linearity error <1% (5ms response time)
Sealing IP67
Weight <160g
Material 7075 aluminum

KT Controller Compatibility

Good news — KT controllers (your T06S) do support torque sensor input. A user on Endless Sphere confirmed KT controllers have torque settings and some support both PAS and torque. endless-sphere

The Ride Feel Difference

This is the real reason to care: victripebike-eu

Cadence (current) Torque (upgrade)
Starting from stop Delay ~0.5s, then motor jumps in Instant, smooth ramp-up
Cruising Motor at fixed %, feels like autopilot Motor matches your effort exactly
Hills Same power whether you push hard or soft Push harder → more help automatically
Slowing down Motor stays on until you stop pedaling Motor fades as you ease off
Overall feel Riding a scooter that requires token pedaling Riding a normal bike with superhuman legs

A torque sensor makes the e-bike feel invisible — people can't tell you have a motor. With cadence, the on/off behavior gives it away. For a stealth fixie build like Omera, torque sensing is the philosophically correct choice.

Cost Impact

Part Cadence (current) Torque (upgrade) Δ
PAS sensor €10 — (removed) -€10
BB torque sensor (Bikone/Sempu) — ~€60–80 +€70
BB unit €6 (separate) Integrated -€6
Net €16 ~€70 +€54

About €54 more for a dramatically better ride feel. Worth it on a stealth build where the whole point is "you can't tell it's electric."

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Looking at your CAD model — great work in what looks like Shapr3D. I can see the issue: the BB junction is tight. The down tube, seat tube, chainstays, and seat stays all converge right there, and you need to squeeze a torque sensor cable out of that BB shell.

Where the Torque Sensor Goes

The sensor replaces your bottom bracket — it threads into the same BSA 68mm shell you already have in the CAD model. From the outside, it looks identical. The strain gauge is inside the BB spindle itself.s2tech+1

The Cable Problem

Here's the real issue you spotted. The sensor has a thin signal cable (~3mm) that must exit the BB shell. Looking at your model:

text
        TOP TUBE ────────── SEAT TUBE
              \              |
               \             |
     DOWN TUBE  \           |
     (battery    \    BB    |
      inside)     \  SHELL  |
                   ╲━━(●)━━╱
                   /        \
          CHAINSTAY          CHAINSTAY
          (drive)            (non-drive)

Cable exit: Drill a 10mm hole in the BB shell

The standard installation requires drilling a 10mm hole in the BB shell, 26mm from the non-drive side. The cable exits there and then routes to the controller.[ebikes]​[youtube]​

Looking at your CAD, here's the routing:

Option A: Down the non-drive chainstay (best for you)

text
  Torque sensor cable exits BB shell (non-drive side)
  → runs along non-drive chainstay (zip-tied or internal)
  → enters rear dropout area
  → connects to controller
  
  BUT: your controller is in the DOWN TUBE, not at the rear

Option B: Up into the down tube (cleanest for Omera)

This is the move. Since your controller lives inside the down tube in the electronics zone near the BB end:[ppl-ai-file-upload.s3.amazonaws]​

text
  Torque sensor cable exits BB shell (TOP of shell, facing DT)
  → goes straight UP into the down tube opening
  → connects to KT T06S controller sitting right there
  → cable length: ~50mm. Basically zero visible wire.

Drill the 10mm exit hole at the top of the BB shell where it meets the down tube. The cable disappears directly into the electronics zone. From the outside: invisible. This is the huge advantage of having the controller inside the down tube — everything converges at the BB.[ebikes]​

What to Change in Your CAD Model

  1. BB shell — add a 10mm hole at the top, centered where the DT meets the shell. This is the cable passthrough

  2. Down tube interior — at the BB end, the electronics zone has the controller PCB, boost converter, etc. The torque sensor cable plugs into the KT T06S's PAS/torque input (same JST connector)

  3. No external cable — nothing visible. The sensor is inside the BB, the cable goes up into the DT, done

Clearance Check at Your BB Junction

From your close-up shot, the weld/junction area where DT meets BB looks clean. You need:

Clearance | Required | Your Model -- | -- | -- BB shell internal Ø | 33.8mm (BSA standard) | ✅ Standard Cable exit hole | 10mm diameter | Drill at DT junction Cable bend radius | ~15mm minimum | ✅ Gentle curve up into DT Crank arm clearance from shell | ~2mm | Check non-drive side

The torque sensor IS the bottom bracket — it replaces it entirely. One part does both jobs. Your crankset (130 BCD square taper) bolts onto the torque sensor's square taper spindle exactly the same way.zbotic+1

Want me to update the master BOM with the 36V / oval tube / torque sensor changes all at once?

Looking at your CAD model — great work in what looks like Shapr3D. I can see the issue: the BB junction is **tight**. The down tube, seat tube, chainstays, and seat stays all converge right there, and you need to squeeze a torque sensor cable out of that BB shell.

Where the Torque Sensor Goes

The sensor replaces your bottom bracket — it threads into the same BSA 68mm shell you already have in the CAD model. From the outside, it looks identical. The strain gauge is inside the BB spindle itself. s2tech

The Cable Problem

Here's the real issue you spotted. The sensor has a thin signal cable (~3mm) that must exit the BB shell. Looking at your model:

        TOP TUBE ────────── SEAT TUBE
              \              |
               \             |
     DOWN TUBE  \           |
     (battery    \    BB    |
      inside)     \  SHELL  |
                   ╲━━(●)━━╱
                   /        \
          CHAINSTAY          CHAINSTAY
          (drive)            (non-drive)

Cable exit: Drill a 10mm hole in the BB shell

The standard installation requires drilling a 10mm hole in the BB shell, 26mm from the non-drive side. The cable exits there and then routes to the controller. [ebikes](https://ebikes.ca/documents/TDCMGuide.pdf)

Looking at your CAD, here's the routing:

Option A: Down the non-drive chainstay (best for you)

  Torque sensor cable exits BB shell (non-drive side)
  → runs along non-drive chainstay (zip-tied or internal)
  → enters rear dropout area
  → connects to controller
  
  BUT: your controller is in the DOWN TUBE, not at the rear

Option B: Up into the down tube (cleanest for Omera)

This is the move. Since your controller lives inside the down tube in the electronics zone near the BB end: [ppl-ai-file-upload.s3.amazonaws](https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/attachments/139085844/0d077e44-3fd7-4c30-a2da-d2a953ad8731/fixie-dream.pdf)

  Torque sensor cable exits BB shell (TOP of shell, facing DT)
  → goes straight UP into the down tube opening
  → connects to KT T06S controller sitting right there
  → cable length: ~50mm. Basically zero visible wire.

Drill the 10mm exit hole at the top of the BB shell where it meets the down tube. The cable disappears directly into the electronics zone. From the outside: invisible. This is the huge advantage of having the controller inside the down tube — everything converges at the BB. [ebikes](https://ebikes.ca/documents/TDCMGuide.pdf)

What to Change in Your CAD Model

  1. BB shell — add a 10mm hole at the top, centered where the DT meets the shell. This is the cable passthrough
  2. Down tube interior — at the BB end, the electronics zone has the controller PCB, boost converter, etc. The torque sensor cable plugs into the KT T06S's PAS/torque input (same JST connector)
  3. No external cable — nothing visible. The sensor is inside the BB, the cable goes up into the DT, done

Clearance Check at Your BB Junction

From your close-up shot, the weld/junction area where DT meets BB looks clean. You need:

Clearance Required Your Model
BB shell internal Ø 33.8mm (BSA standard) ✅ Standard
Cable exit hole 10mm diameter Drill at DT junction
Cable bend radius ~15mm minimum ✅ Gentle curve up into DT
Crank arm clearance from shell ~2mm Check non-drive side

BOM Update

Part Old New Δ Price
PAS cadence sensor €10 Removed -€10
BB square taper (separate) €6 Removed -€6
BB torque sensor (BSA 68mm) — ~€60–80 +€70
Net change +€54

The torque sensor IS the bottom bracket — it replaces it entirely. One part does both jobs. Your crankset (130 BCD square taper) bolts onto the torque sensor's square taper spindle exactly the same way. zbotic

Want me to update the master BOM with the 36V / oval tube / torque sensor changes all at once?

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Great big-picture thinking. Let me lay out the two routes cleanly, because the hybrid approach you're converging on is actually the most elegant solution.

The Bikone Torque Sensor Changes Everything

The Bikone BB torque sensor replaces the entire bottom bracket — it threads into the BB shell as a BSA or T47 unit. It's only 160g, 7075 aluminium, IP67 sealed, and outputs a single thin UART signal cable. That's it — one small wire, not a power cable. This means the BB shell stays clean: torque sensor threads in, crank spindle goes through it, one 4mm hole into the DT for the signal wire. Done.[bikone]​

Compared to the old PAS magnet ring setup, this is a massive upgrade — strain gauge torque + cadence + crank angle in one unit, instant 5ms response, and the ride quality difference between cadence-sensing and torque-sensing is night and day.reddit+1

The Two Routing Options

Option A: Everything Through BB (Traditional)

text
HEAD TUBE
  ↓ (all cables down DT interior, squeezed alongside battery)
DOWN TUBE (60×55 oval, battery + cables = tight)
  ↓
BB SHELL (torque sensor + crank + all cables passing through = CRAMPED)
  ↓
CHAINSTAYS → rear dropout (motor cable + brake hose)

Problem: The BB shell is ~68mm wide and fully occupied by the torque sensor + crank spindle. Routing motor power cable (3-phase, ~8mm thick) plus brake hose (~5mm) through here means etching channels into the BB shell or making it oversized. Messy.

Option B: Split Route (Your Hybrid) ✓

text
ROUTE 1 — Power + signal (short, minimal)
  Torque sensor UART wire → 4mm hole → DT interior → BMS/controller
  Battery power → DT interior → BMS/controller → (see Route 2)

ROUTE 2 — Motor + brake (spacious, clean)
  Controller (in electronics module, seat tube)
     ↓
  Motor power cable exits electronics module
     ↓
  SEAT TUBE → SEATSTAYS → rear dropout → hub motor
  Brake hose: HEAD TUBE → TOP TUBE → SEAT TUBE → SEATSTAY → rear caliper

Why Option B Is Better

Concern | Option A (all through BB) | Option B (split route) -- | -- | -- BB shell | Etched channels, oversized, complex | Clean, circular, standard BSA/T47 DT interior | Battery + 3 cables = jammed | Battery + 1 thin UART wire only Cable bends | Sharp 90° turns at BB | Gentle curves through TT/ST Serviceability | Must remove crank to access cables | Cables accessible from seatpost removal Weathering | BB is the wettest area on the frame | TT/ST stays dry Manufacturing | Custom machined BB shell | Standard BB shell, off-the-shelf

The Full Architecture

text
                    HEAD TUBE
                   /         \
     brake hose → TOP TUBE    DOWN TUBE ← battery only
                   \         /            + torque UART wire
                  SEAT TUBE
                  (electronics module:
                   controller, BMS,
                   GPS, BLE)
                   /         \
            LEFT SEATSTAY    RIGHT SEATSTAY
            (brake hose)     (motor 3-phase cable)
                   \         /
                  REAR DROPOUT
                  (hub motor + disc caliper)

The electronics module in the seat tube becomes the central junction — battery power comes up from DT, torque signal comes up from DT, and the module sends motor commands back down through the seatstays. The brake hose enters at the head tube, runs along the TT roof (plenty of space in a 28.6mm round tube), drops through the ST, and exits via the left seatstay.[ppl-ai-file-upload.s3.amazonaws]​

Seat Tube / Rear Wheel Cutout

Your idea of sculpting the seat tube to let the rear wheel tuck closer is doable — a few builders do this for short-chainstay track/crit frames. It would give you:

  • Shorter wheelbase = snappier handling in traffic

  • Seat tube becomes D-shaped or asymmetric in cross-section near the bottom = room for cables alongside the seatpost

  • The seatstays can overlap/wrap the seat tube slightly at the junction, creating natural internal passages for the motor cable and brake hose to transition from ST into the stays without external routing[bikeforums]​

Cost of the Torque Sensor Upgrade

The Bikone unit is OEM-priced (you'd need to contact them for quantity pricing), but comparable BB torque sensors run €80–130. The PAS sensor it replaces was ~€10, so the delta is roughly +€100 — but the ride quality jump from cadence to torque sensing is the single biggest UX upgrade you can make on a hub motor ebike.fit-ebike+2

This is the right architecture. The BB stays dumb and clean, the DT stays dedicated to battery, and the seat tube becomes the brain.

Great big-picture thinking. Let me lay out the two routes cleanly, because the hybrid approach you're converging on is actually the most elegant solution.

The Bikone Torque Sensor Changes Everything

The Bikone BB torque sensor replaces the entire bottom bracket — it threads into the BB shell as a BSA or T47 unit. It's only 160g, 7075 aluminium, IP67 sealed, and outputs a single thin UART signal cable. That's it — one small wire, not a power cable. This means the BB shell stays clean: torque sensor threads in, crank spindle goes through it, one 4mm hole into the DT for the signal wire. Done. [bikone](https://www.bikone.com/bottombracket-torque-sensors/)

Compared to the old PAS magnet ring setup, this is a massive upgrade — strain gauge torque + cadence + crank angle in one unit, instant 5ms response, and the ride quality difference between cadence-sensing and torque-sensing is night and day. reddit

The Two Routing Options

Option A: Everything Through BB (Traditional)

HEAD TUBE
  ↓ (all cables down DT interior, squeezed alongside battery)
DOWN TUBE (60×55 oval, battery + cables = tight)
  ↓
BB SHELL (torque sensor + crank + all cables passing through = CRAMPED)
  ↓
CHAINSTAYS → rear dropout (motor cable + brake hose)

Problem: The BB shell is ~68mm wide and fully occupied by the torque sensor + crank spindle. Routing motor power cable (3-phase, ~8mm thick) plus brake hose (~5mm) through here means etching channels into the BB shell or making it oversized. Messy.

Option B: Split Route (Your Hybrid) ✓

ROUTE 1 — Power + signal (short, minimal)
  Torque sensor UART wire → 4mm hole → DT interior → BMS/controller
  Battery power → DT interior → BMS/controller → (see Route 2)

ROUTE 2 — Motor + brake (spacious, clean)
  Controller (in electronics module, seat tube)
     ↓
  Motor power cable exits electronics module
     ↓
  SEAT TUBE → SEATSTAYS → rear dropout → hub motor
  Brake hose: HEAD TUBE → TOP TUBE → SEAT TUBE → SEATSTAY → rear caliper

Why Option B Is Better

Concern Option A (all through BB) Option B (split route)
BB shell Etched channels, oversized, complex Clean, circular, standard BSA/T47
DT interior Battery + 3 cables = jammed Battery + 1 thin UART wire only
Cable bends Sharp 90° turns at BB Gentle curves through TT/ST
Serviceability Must remove crank to access cables Cables accessible from seatpost removal
Weathering BB is the wettest area on the frame TT/ST stays dry
Manufacturing Custom machined BB shell Standard BB shell, off-the-shelf

The Full Architecture

                    HEAD TUBE
                   /         \
     brake hose → TOP TUBE    DOWN TUBE ← battery only
                   \         /            + torque UART wire
                  SEAT TUBE
                  (electronics module:
                   controller, BMS,
                   GPS, BLE)
                   /         \
            LEFT SEATSTAY    RIGHT SEATSTAY
            (brake hose)     (motor 3-phase cable)
                   \         /
                  REAR DROPOUT
                  (hub motor + disc caliper)

The electronics module in the seat tube becomes the central junction — battery power comes up from DT, torque signal comes up from DT, and the module sends motor commands back down through the seatstays. The brake hose enters at the head tube, runs along the TT roof (plenty of space in a 28.6mm round tube), drops through the ST, and exits via the left seatstay. [ppl-ai-file-upload.s3.amazonaws](https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/attachments/139085844/b9fdad49-79fd-48bb-bcb5-b3f33bc64277/emotor.pdf)

Seat Tube / Rear Wheel Cutout

Your idea of sculpting the seat tube to let the rear wheel tuck closer is doable — a few builders do this for short-chainstay track/crit frames. It would give you:

Cost of the Torque Sensor Upgrade

The Bikone unit is OEM-priced (you'd need to contact them for quantity pricing), but comparable BB torque sensors run €80–130. The PAS sensor it replaces was ~€10, so the delta is roughly +€100 — but the ride quality jump from cadence to torque sensing is the single biggest UX upgrade you can make on a hub motor ebike. fit-ebike

This is the right architecture. The BB stays dumb and clean, the DT stays dedicated to battery, and the seat tube becomes the brain.

@eonist
Copy link
Author

eonist commented Mar 12, 2026

The cost varies dramatically depending on whether you're building one-off or producing at scale, and which construction method you use. Here's a breakdown across the key cost areas:

Raw Materials

The carbon fiber itself is surprisingly affordable relative to the retail price of frames. High-modulus unidirectional (UD) carbon costs over £100/m² for 100g weight, while standard modulus 300g UD runs under £25/m². A typical frame uses around 3–7 square metres of layup material, putting raw carbon costs at roughly $60–$700 depending on the grade. Including resin and consumables, total material cost for a factory-produced frame lands around $120–$400.weightweenies.starbike+2

Mold Tooling (Monocoque)

If you go with monocoque (one-piece) construction — the method used by Trek, Specialized, etc. — the mold is the biggest upfront investment:

Frame Type | Mold Cost | MOQ -- | -- | -- Road / Gravel | $6,000–$9,000 | 30–50 units MTB / Enduro | $8,000–$12,000 | 50+ units

[trainerroad]​

High-precision machined molds can run $10,000–$20,000 per size, and each mold has a limited lifespan before replacement. For a full size run (e.g., 5 sizes), you're looking at $50,000–$100,000+ in tooling alone.[weightweenies.starbike]​

Tube-to-Tube Construction

This is the more accessible route for custom/one-off builds since it avoids expensive molds. You use pre-made carbon tubes joined with custom-molded lugs or bonded junctions. One builder reported being into it for about $1,000 in specialized tooling beyond standard framebuilding equipment. Material costs per frame stay in the low hundreds, but labor is significant.[forum.customframeforum]​

Finished Custom Frame Pricing

Custom carbon frames from established builders typically retail between:

  • $3,000–$5,000 for Chinese OEM tube-to-tube or small-batch monocoque builds[youtube]​

  • $7,000+ from US-based artisan builders like Montenegro MFG (includes custom geometry, paint, fork, and headset)[montenegromfg]​

  • $10,000–$20,000 from premium custom shops, reflecting both material and labor intensity[reddit]​

Unit Economics at Scale

At volume production in a decent factory, the per-frame manufacturing cost (materials + labor + overhead, excluding paint) is roughly $300. Painting adds extra cost. The massive markup from $300 to retail ($1,500–$5,000+ for the frame alone) covers R&D, mold amortization, QC/testing, shipping, brand margin, and retailer cut.[weightweenies.starbike]​

Given your background with hardware prototyping and the Syla lamp work, the tube-to-tube approach might be the most practical entry point if you're exploring a one-off or small-batch prototype — it keeps tooling costs manageable and lets you iterate on geometry without sinking $10K+ into molds per size.

The cost varies dramatically depending on whether you're building one-off or producing at scale, and which construction method you use. Here's a breakdown across the key cost areas:

Raw Materials

The carbon fiber itself is surprisingly affordable relative to the retail price of frames. High-modulus unidirectional (UD) carbon costs over £100/m² for 100g weight, while standard modulus 300g UD runs under £25/m². A typical frame uses around 3–7 square metres of layup material, putting raw carbon costs at roughly $60–$700 depending on the grade. Including resin and consumables, total material cost for a factory-produced frame lands around $120–$400. weightweenies.starbike

Mold Tooling (Monocoque)

If you go with monocoque (one-piece) construction — the method used by Trek, Specialized, etc. — the mold is the biggest upfront investment:

Frame Type Mold Cost MOQ
Road / Gravel $6,000–$9,000 30–50 units
MTB / Enduro $8,000–$12,000 50+ units

[trainerroad](https://www.trainerroad.com/forum/t/cost-of-building-up-a-frameset/72755)

High-precision machined molds can run $10,000–$20,000 per size, and each mold has a limited lifespan before replacement. For a full size run (e.g., 5 sizes), you're looking at $50,000–$100,000+ in tooling alone. [weightweenies.starbike](https://weightweenies.starbike.com/forum/viewtopic.php?t=145480)

Tube-to-Tube Construction

This is the more accessible route for custom/one-off builds since it avoids expensive molds. You use pre-made carbon tubes joined with custom-molded lugs or bonded junctions. One builder reported being into it for about $1,000 in specialized tooling beyond standard framebuilding equipment. Material costs per frame stay in the low hundreds, but labor is significant. [forum.customframeforum](https://forum.customframeforum.com/t/custom-carbon-fiber/1003)

Finished Custom Frame Pricing

Custom carbon frames from established builders typically retail between:

Unit Economics at Scale

At volume production in a decent factory, the per-frame manufacturing cost (materials + labor + overhead, excluding paint) is roughly $300. Painting adds extra cost. The massive markup from $300 to retail ($1,500–$5,000+ for the frame alone) covers R&D, mold amortization, QC/testing, shipping, brand margin, and retailer cut. [weightweenies.starbike](https://weightweenies.starbike.com/forum/viewtopic.php?t=145480)

Given your background with hardware prototyping and the Syla lamp work, the tube-to-tube approach might be the most practical entry point if you're exploring a one-off or small-batch prototype — it keeps tooling costs manageable and lets you iterate on geometry without sinking $10K+ into molds per size.

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Here's the complete routing map with your exact paths. Six routes, seven frame holes, one external cable.

The Six Routes

Route | Cable | Path | Holes -- | -- | -- | --

The Seatpost Light Idea

That's a great stealth detail. The rear light wire exits the top of the seat tube, enters the seatpost through a small hole, and runs up to a light mounted under the saddle. Users can drill their own seatpost, buy one with integrated light, or just zip-tie a light and feed the wire externally — on them, as you said.

Stealth Result

Only one visible cable on the entire bike: the front brake hose down the fork, which every bike has. Everything else is internal. Seven small holes (6–12mm), all fitted with rubber grommets for weather sealing. From the outside, it's just a fixie.[ppl-ai-file-upload.s3.amazonaws]​

Here's the complete routing map with your exact paths. Six routes, seven frame holes, **one external cable**.

The Six Routes

Route Cable Path Holes
A Rear brake Hydraulic hose HT hole → TT → ST → seat stay → exit near caliper HT entry, SS exit
B Motor power 3-phase + hall DT (controller) → DT/ST overlap → short ST trip → chainstay → dropout exit → hub Overlap passthrough, dropout exit
C Torque sensor 4-wire signal BB → 10mm hole → straight into DT. Done. 50mm total. BB 10mm hole
D Rear light Thin 2-wire DT → overlap → up ST → exit top of ST → up through seatpost → saddle light ST top exit
E Front light Thin 2-wire DT (controller) → up DT internal → exit HT hole → user routes to cockpit HT exit
F Front brake Hydraulic hose External, down the fork. Standard. Short. None

The Seatpost Light Idea

That's a great stealth detail. The rear light wire exits the top of the seat tube, enters the seatpost through a small hole, and runs up to a light mounted under the saddle. Users can drill their own seatpost, buy one with integrated light, or just zip-tie a light and feed the wire externally — on them, as you said.

Stealth Result

Only one visible cable on the entire bike: the front brake hose down the fork, which every bike has. Everything else is internal. Seven small holes (6–12mm), all fitted with rubber grommets for weather sealing. From the outside, it's just a fixie. [ppl-ai-file-upload.s3.amazonaws](https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/attachments/139085844/0d077e44-3fd7-4c30-a2da-d2a953ad8731/fixie-dream.pdf)

@eonist
Copy link
Author

eonist commented Mar 12, 2026

You're right — the brake hose doesn't go all the way down the seat tube. It enters the ST briefly at the TT/ST junction, then immediately crosses into the seat stay and rides that all the way down to the caliper.

Corrected Route A:

text
HT hole → through TT → into ST (just the junction, ~20mm) → into seat stay → down seat stay → exit hole → rear caliper

The seat tube is basically just a junction box at the top — the hose passes through it for a centimeter or two to get from the TT into the seat stay. It never goes down the ST toward the BB. That keeps the ST cable void free for the motor cable (Route B) and rear light wire (Route D) which actually need to travel the full length of the seat tube.

Updated Route Table

Route | Path -- | -- A Rear brake | HT → TT → brief ST junction → seat stay → exit → caliper B Motor power | DT → overlap → short ST → chainstay → dropout → hub C Torque sensor | BB → 10mm hole → DT D Rear light | DT → overlap → up full ST → seatpost → saddle light E Front light | DT → up DT → exit HT → user's problem F Front brake | External down fork

Good catch — the ST only handles Route B and D long-distance. Route A just clips through the top junction. I'll correct the SVG if you want an updated drawing.

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Yes. With the current plan everything except the front brake hose is fully internal.

  • Rear brake hose: head tube → top tube → brief seat-tube junction → inside seat stay → out at caliper.
  • Motor cable: inside down tube → overlap into seat tube → inside chainstay → out at dropout into hub.
  • Torque sensor: inside BB → 10 mm hole → straight into down tube electronics zone.
  • Rear light wire: inside down tube → overlap into seat tube → up seat tube → into seatpost → out under saddle.
  • Front light wire + display/control wiring: inside down tube → out at head tube → user routes to cockpit.
  • Only external line: front brake hose down the fork, which is normal even on high‑end road bikes. [bikeradar](https://www.bikeradar.com/advice/buyers-guides/bike-tyre-sizes)

Do you want to keep that external front hose (simpler fork) or explore an internally-routed fork as a stretch goal?

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Yes. Everything can be fully internal with the layout you’ve described.

What’s Internal vs External

  • Rear disc brake hose:
    Internal from head tube → top tube → seat tube → left seatstay → exits only right at the caliper.

  • Motor cable:
    Internal from controller in down tube → through DT/BB/chainstay overlap port → inside right chainstay → exits only at rear dropout into hub.

  • Torque sensor wire:
    Internal from BB shell → tiny 4mm hole into down tube → up to controller.

  • Battery power + all electronics (controller, BMS, GPS/BLE):
    Entirely inside the down tube.

  • Front light wire:
    Internal in down tube → exits at head tube. (User handles cockpit routing.)

  • Rear light wire:
    Internal DT → overlap → up seat tube → exits near seatpost or through a hollow seatpost.

Externally you only see:

  • The rear brake caliper hose at the caliper itself.
  • The motor cable at the dropout–hub interface.
  • The headlight and taillight fixtures and their short exposed pigtails.

Everything else lives inside the tubes.

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Yes. Fully internal. No temporary exits. Let’s lock this in very explicitly.

Final Motor Cable Path (Only Exit Is At Hub)

Motor cable (8mm 3‑phase + signal) route:

  1. Inside down tube

    • Starts at the controller/BMS block in the free 130mm space above the BB.
    • Runs inside the down tube, below the battery pack, toward the BB area.

  2. Inside DT/BB/chainstay overlap zone

    • Passes through an internal brazed passage in the junction where down tube, BB shell, seat tube, and chainstay meet.
    • This is designed as a built‑in channel during framebuilding, not an external hack. pedelecs.co

  3. Inside right chainstay

    • Enters the right chainstay interior at the BB junction.
    • Runs fully inside the chainstay tube to the rear dropout.

  4. Exit at dropout → into hub

    • Only here does it exit the frame, at a small port right by the dropout.
    • Immediately plugs into the hub motor connector.

So:

  • No external run under the BB.
  • No exposed section along the stays.
  • Only visible section is the last few centimeters right at the dropout into the motor, just like on clean OEM ebikes. bigforestframeworks

Everything else for the motor cable is inside steel tubes, from controller to dropout.

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Your plan (fully internal motor cable, only visible at the dropout → hub) is exactly the clean, “OEM” way to do it. There isn’t a better alternative that avoids that last short exposed section.

How rear hub motors are normally wired

On essentially all decent rear‑hub ebikes:

  • The motor cable exits the motor axle right at the dropout (short exposed pigtail).
  • From there it immediately enters the chainstay, runs inside the stay, and then inside the frame to the controller. macfoxbike
  • The only unavoidable external bit is that few‑cm section right at the axle, which also needs strain relief for safety. macfoxbike

Hiding the entire cable path inside the frame up to the dropout, as you’re planning, is actually cleaner than many production bikes, which often zip‑tie the cable along the outside of the chainstay. forums.electricbikereview

Why there isn’t a “more internal” option

The motor connector physically sits at the end of the axle or just outside it. You cannot:

So the cleanest, structurally safe pattern is exactly:

  1. Internal: controller in DT → internal DT/BB overlap passage → internal right chainstay.
  2. External only for the last 3–5cm: exit at dropout → plug into hub.

You’re already designing to that best practice. The cable is inside the frame all the way until it has no choice but to appear at the axle.

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Yes, you can absolutely design a cap/guard to protect that tiny exposed section at the dropout. That’s a good idea, and other systems do similar things.

What the cap does

  • Covers the cable exit + connector area at the dropout.
  • Shields it from side impacts, rocks, curbs, bike racks.
  • Acts as strain relief, so any tug on the cable is absorbed by the guard, not the connector. reddit
  • Visually cleans up the last exposed bit.

How others do it

  • Bosch and Giant both sell small molded caps/covers to protect ebike cables and charge ports at frame exits. ebike24
  • Some ebike owners 3D‑print rear hub wiring guards that wrap around the dropout area to protect the harness; these live right where your cable exits. facebook
  • There are generic Bafang cable guards and caps in 3D print libraries you can copy or adapt. [printables](https://www.printables.com/tag/bafang)

How you could implement it on Omera

  • Material: small steel or aluminum tab welded to the dropout, plus a snap‑on plastic or rubber cover; or a fully 3D‑printed nylon guard that bolts to a tiny boss.
  • Shape: an “L” or “U” profile that wraps around the cable on the outside of the dropout, extending just past the connector.
  • Mounting:
    • One M3/M4 boss on the dropout to screw the guard on, or
    • A slot in the dropout plate where a guard tongue slides in and is trapped when the axle is clamped.
  • Serviceability: single screw (or clip) removal so the rear wheel can still come out easily.

So the final picture is:

  • Cable fully internal through chainstay.
  • Exits right at dropout.
  • Immediately disappears under a small protective cap/guard, then into the hub.

That’s as clean and robust as it gets.

@eonist
Copy link
Author

eonist commented Mar 12, 2026

Yes — there are proven, buyable “frame split” solutions you can design around instead of inventing your own from scratch.

What Gates Requires

Gates explicitly says the frame must have a “frame break” in the rear triangle so the one‑piece belt can be slipped in. mi-tech
They don’t care where, as long as:

  • You have at least 8 mm gap when it opens.
  • You also have a belt tensioning system (sliding dropouts or eccentric BB). 2blua

You’re already planning horizontal/sliding dropouts, so you just need the frame split.

Off‑the‑shelf frame split hardware

You can buy CNC’d metal splitters and design your stays around them:

  1. Paragon Machine Works belt splitters

  2. Dropout‑integrated split (Tout Terrain / Olsen style)

  3. Custom seat‑stay splitter services (Bike Knack, Ti Cycles)

    • Shops like Bike Knack and Ti Cycles retrofit Gates frames by cutting the seat stay and installing a machined splitter matched to the stay diameter. bikeknack.com
    • Their parts are basically a tube connector with two bolts. Good reference geometry if you want to mill your own.

  4. Veer split belt (no frame split)

What makes sense for Omera

Given your goals (stealth, clean BB, CNC dropouts, 135 mm belt hub + disc):

  • Use a dropout‑integrated split on the right side, like the Olsen RocknRollout / Tout Terrain pattern. olsenbicycles

    • Chainstay and seat stay both bolt into a two‑piece dropout.
    • Loosen 2 bolts, remove the outer piece, slide belt through, bolt back up.
    • Visually, it just looks like a chunky CNC dropout, which you already want.

  • Keep no cuts in the seat stay or chainstay tubes themselves — all the complexity is in a replaceable dropout block.

So the answer: yes, you can absolutely base your design on existing hardware. Look at Paragon’s belt‑drive splitters and the RocknRollout/Tout Terrain split dropout concepts as references, then have your own 135 mm disc‑compatible CNC dropout machined around a similar interface. bikeforums

Would you like me to sketch a right‑side dropout concept (135 mm, flat‑mount disc, belt split) you can hand to a machinist?

@eonist
Copy link
Author

eonist commented Mar 12, 2026

okay so you sanwitch the dropout module to chain stay and seat stay and lock it with screwbolts. unscrew the bolts. the gap for belt is there

@eonist
Copy link
Author

eonist commented Mar 12, 2026

omera_split_dropout

@eonist
Copy link
Author

eonist commented Mar 13, 2026

Got it — then we just mirror the function, not the look.

On the non‑drive side:

  • Same axle slot height and length as drive side, so tensioning and wheel position behave identically.
  • One solid 6 mm plate (no split), with: axle slot, brake mount (if rotor that side), and a small hole/feature for hose exit if needed. bikecad
  • Overall 135 mm O.L.D. preserved by making this plate the same thickness as the inner plate on the drive side, so dish stays simple and hub sits centered. alibaba

If you like, I can now define just the critical numbers you need to send your machinist for that non‑drive plate (slot position, thickness, axle center to stay junction, etc.) in plain text.

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