Skip to content

Week 18 : Project Development, Applications & Implications

This documentation focuses on higher-level planning of the project - please refer to the final project overview for further details on each building block (i.e. SBC, MCU, power supply, and payment system).

Scope

Kilat is a portable, pay-to-use solar pv system for street vendors. Customers trade mostly in the evening, earn on a daily basis, and have limited access to reliable or more affordable electrcity.

Generally, product features such as solar tracking and pay-as-you-go (PAYG) functionality are developed as separate modules targeting different market needs. This prototype, however, integrates selected features from these systems into a unified solution and catered for the street vendors inspired by the following past resources :

The scope is from there determined to be :

  • Prototype (current):
  • A single-axis (east–west) solar tracker that keeps the panel oriented toward the sun through the day so the battery is charged before the evening trading hours.
  • Energy is stored in a battery pack so power is available after sunset.
  • The user can draw power (the prototype focuses on USB phone charging as the representative load) only while their subscription / payment is active.
  • Activation is handled remotely through the payment/lock logic with a MOSFET load switch acting as the physical "lock".
  • The panel is raised on a manually-elevated mast held by a spring-detent button, and the components are packaged in a compact control unit that can be stored vertically or horizontally.
  • Future vision (documented, not built in prototype):
  • Additional loads beyond phone charging (e.g. lamps and other small appliances).
  • A real, transactional payment system rather than the dummy/token subscription used for the prototype.

Design, make-or-buy, and processes

Sub-system Made / Bought Process / Tools Design
Tracker head Made CAD → FDM 3D print (PLA/PETG) Panel bracket and east–west rotation joint carrying the servo and phototransistors
Control-unit enclosure Made FDM 3D print + laser-cut back panel Houses battery and PCB; chamfered edges, heat-set inserts; back panel laser-cut due to 3D-printer build-plate size limit
Mast + spring-detent lock Made / Bought 3D print + assembly Manually elevated mast with internal cable routing so no wiring is visible. spring detent lock selected as a purchased component to ensure sufficient mechanical strength and durability under load
Custom PCB Made PCB milling + soldering Phototransistor sensing channels, servo output(s), MOSFET load switch (payment/lock + USB enable), and the power path
Cast logo Made Molding & casting Fluorescent accent on the casing with crystal epoxy mixture, no added LEDs
Embedded software Made MQTT + Node-RED (Pi); Cloudflare + Vercel + MongoDB Local tracking loop (sensors → servo); MOSFET lock logic; remote payment path
Solar panel, regulator, battery, servo, sensors, Pi, LoRa, OLED Bought - Energy and compute components integrated into the system rather than fabricated (see BOM)

Bill of materials (BOM)

Almost all components were sourced externally through Taobao, while a small number of items were already available in-house or obtained off-the-shelf. The BOM below compares the planned quantities and costs with the actual quantities and costs incurred during the project. Costs are listed in RMB, with total costs also converted to USD for reference. An analysis of it is also conducted for future improvement.

Note: Actual quantities may be fractional where only part of a purchased multi-pack was consumed in the build (e.g. a portion of a 100-pack of screws, resistors, or connectors). Rows with an actual quantity and cost of 0 indicate components that were purchased or considered during development but were ultimately not used in the final prototype.

Component / Material Source Qty (Planned) Price (Planned) Qty (Actual) Price (Actual)
Filament PLA basic 10105 Taobao 2 116.2 1.0 58.1
2.42 inch OLED Taobao 1 56.0 1.0 56.0
Filament PETG Basic 30107 Taobao 2 103.6 1.0 51.8
Solar panel Taobao 1 45.0 1.0 45.0
Battery 5V5A 5600 mAh Taobao 1 42.0 1.0 42.0
MG996R 180 servo Taobao 3 44.2 2.0 29.5
ESP32-DEV CH340 Taobao 2 42.0 1.0 21.0
50x 22AWG dupont cable Taobao 1 36.1 0.5 18.0
Phototransistors Taobao 4 35.2 2.0 17.6
3P alternating pin socket Taobao 4 11.0 4.0 11.0
MG996R metal arms Taobao 1 9.0 1.0 9.0
Heat tubes Taobao 1 8.8 1.0 8.8
NCE01P30K PMOS TO-252 Taobao 5 7.7 5.0 7.7
Cable skeleton Taobao 1 6.8 1.0 6.8
White-blue fluorescent powder Taobao 1 11.6 0.5 5.8
10x30p pin socket Taobao 1 11.0 0.5 5.5
USB C breakout board Taobao 4 9.2 2.0 4.6
AMS1117 3.3V linear regulator Taobao 10 6.4 5.0 3.2
USB A connector with silicone casing Taobao 2 5.6 1.0 2.8
10x6p pin socket Taobao 1 4.8 0.5 2.4
Kafuter silicon Taobao 2 6.6 0.5 1.7
100x Ceramic capacitor 100 nF Taobao 1 3.0 0.5 1.5
100x resistor 100 Ohm Taobao 1 3.0 0.5 1.5
100x resistor 100 kOhm Taobao 1 3.0 0.5 1.5
10x Ceramic capacitor 10 µF Taobao 2 3.0 1.0 1.5
10x M3x6x4.2 heat insert Taobao 3 6.2 0.7 1.4
Buck converter Taobao 1 1.2 1.0 1.2
20x JST PH2.0 mm cable 2P red-black Taobao 1 4.7 0.3 1.2
20x JST PH2.0 mm cable 2P black-red Taobao 1 4.7 0.3 1.2
10x M3x4x4.2 heat insert Taobao 3 5.2 0.7 1.1
Spring detent lock F8 Taobao 1 1.1 1.0 1.1
KCD1 switch Taobao 2 4.4 0.5 1.1
5x Alternating pin header 40p Taobao 1 4.8 0.2 1.0
JST PH2.0 mm 4P Taobao 10 1.3 5.0 0.7
100x M3x4 countersink screw Taobao 1 2.6 0.2 0.5
100x M3x6 countersink screw Taobao 1 2.6 0.2 0.5
100x 2N7002 NMOS SOT-23 Taobao 1 5.0 0.1 0.5
JST PH2.0 mm 2P Taobao 10 0.9 5.0 0.5
Aluminum capacitor 100 µF Taobao 10 0.9 5.0 0.5
50x M3x4 counterbore screw Taobao 1 1.0 0.4 0.4
50x M3x6 counterbore screw Taobao 1 0.8 0.4 0.3
Yellow-orange fluorescent powder Taobao 1 11.5 0.0 0.0
White-yellow fluorescent powder Taobao 1 13.4 0.0 0.0
Spring detent lock F6 Taobao 1 0.3 0.0 0.0
Spring detent lock F3 Taobao 1 1.2 0.0 0.0
Spring detent lock AF Taobao 1 1.5 0.0 0.0
Spring detent lock T Taobao 1 2.1 0.0 0.0
Spring detent lock Taobao 1 0.2 0.0 0.0
Spring detent lock F1 Taobao 1 0.2 0.0 0.0
Spring detent lock AX Taobao 1 0.1 0.0 0.0
Spring detent lock AB Taobao 1 0.1 0.0 0.0
10x Grove-Universal 4 pin connector Taobao 1 11.0 0.0 0.0
MG996R 360 servo Taobao 2 32.8 0.0 0.0
Spring M6x10x3x3 Taobao 1 5.1 0.0 0.0
Spring M8x12x4.5x3 Taobao 1 6.0 0.0 0.0
Spring M3x8 Taobao 1 3.0 0.0 0.0
Spring M3x6 Taobao 1 3.0 0.0 0.0
Spring M3x5 Taobao 1 2.7 0.0 0.0
100x resistor 5.1 kOhm Taobao 1 3.0 0.0 0.0
20x 2N7002 NMOS SOT-23 Taobao 2 2.4 0.0 0.0
PH2.0 male and female set Taobao 1 5.3 0.0 0.0
30N06 NMOS TO-252 Taobao 1 2.6 0.0 0.0
Non alternating pin header 40p Taobao 2 0.3 0.0 0.0
USB Type-C connector Taobao 1 2.6 0.0 0.0
USB Type-C connector split Taobao 1 4.3 0.0 0.0
4P alternating pin socket Taobao 4 6.2 0.0 0.0
USB A connector without silicone casing Taobao 1 4.2 0.0 0.0
Solar regulator Taobao 1 13.2 0.0 0.0
Battery 5V1A 6000 mAh Taobao 1 36.3 0.0 0.0
Raspberry Pi 4 Off the shelf 1 - 0.0 0.0
Grove - LoRa 868 MHz Off the shelf 2 - 0.0 0.0
Grove - Compass V2.0 Off the shelf 1 - 0.0 0.0
Basswood Off the shelf 1 - 0.0 0.0
Total price (RMB) 852.7 427.3
Total price (USD) 127.90 64.09

A few observations stand out from the BOM :

  • The actual spend came in at roughly half of the plan — about 427 RMB (≈ $64) actual against 853 RMB (≈ $128) planned. In the future perhaps this could become a benchmark of budget that should be allocated for product creation.
  • The cost is dominated by a handful of items. Even after that optimisation, the seven most expensive parts — the two filament reels, the OLED, the solar panel, the battery, the servo, and the dupont cabling — already account for about three-quarters (~74%) of the actual cost, while the remaining line items together make up only the final quarter. Focus on supply-chain or alternative components would therefore be a priority for optimisation.
  • The economics are encouraging for the pay-as-you-go model. At an actual hardware cost of approximately $64, a vendor paying around $1 per day would recover the full device cost in roughly two months, after which the system could effectively be owned outright. That of course depends on the other features that will be added and whether similar price range could be obtained in Indonesia compared to China.

Evaluation, status and open questions

Aspect Success criterion Status so far Open question
Tracking Panel follows the sun east–west over several hours and parks/stows at night Servo + light-direction sensing drives E–W alignment Does single-axis tracking capture enough charge for an evening? Is one servo enough or is a second axis needed?
Energy Panel charges the battery in sun, and the battery delivers USB power afterward Stable 5 V rail and uninterrupted output once the UPS/pass-through battery was adopted Is the battery sized for a full evening of trading?
Payment lock USB output is off when "unpaid" and on when "paid", reproducibly, driven by a functional payment system MOSFET load switch gives a clean, electronically-controlled cut-off for the lock The payment system should be functional rather than a dummy. Would the user be willing to pay a deposit to guarantee the product is not stolen and is returned?
Portable form Stows compactly and deploys with the spring-assisted mast, without a fixed tower Compact control unit + manually-elevated mast with a selected spring-detent lock Is the mast quick to deploy/stow in a stall? Robust enough for daily outdoor use, and does the fit/placement hold after assembly?
Sensing / orientation Status and orientation shown reliably to the user OLED status display, RTC for timekeeping, and the compass heading are all working GPS is not added yet. If the product is stolen, can its location still be tracked and the unit retrieved? GPS works poorly in enclosed/indoor places, so is there a better-suited technology for theft tracking?
Affordability (product / users) Hardware cost low enough for a daily micro-payment model Actual hardware cost ≈ $64 → roughly two-month payback at $1/day What pricing model (daily / weekly / monthly) and price point would vendors accept?

Lessons learned & next-iteration improvements

  • Cable management :
    • The wiring is not tidy enough — signal-and-return pairs should be twisted and the routing planned more deliberately.
    • The cable skeleton did not perform as hoped. Perhaps the diameter should be bigger so that it creates less stress to the cables and minimizes connection failure. Alternative is perhaps to consolidate the connection of the phototransistors onto a separate PCB and from there less number of cables are needed to the control unit.
  • Design around the fasteners instead of the other way around. Too many options creates confusion.
  • A single MG996R motor is actually in the end sufficient for the tracking. Load analysis should be performed next time.
  • Mast locking mechanism : the spring-detent lock can be released / removed too easily, which is both a usability and a security concern for a deployed, potentially-unattended product. Perhaps a cam-lever clamp, twist/collet collar could be explored in the future.
  • The fluorescent logo did not work as well as hoped — as soon as the OLED display lights up, its brightness overpowers the faint glow and the fluorescent effect is lost. The glow is only really visible in full darkness, so it does not hold up next to an active light source. Thus a cast logo with back-light from an LED should be explored in the future.