Ezar
Sadeque

Why This Project?

Education is an extremely important component of personal and community development, but unfortunately, rural areas are often limited by unreliable power. To help address this challenge, I developed a solar-powered reading light and USB charger to provide students with a dependable source of illumination for studying after dark.

Project Overview

A solar panel (8 W or 12 W) charges a Li-ion battery pack inside a rectangular base. Power is managed by a USB controller, then boosted/regulated to drive high-output LEDs and provide 5 V USB charging on two ports. Two push buttons control the lamp heads. An articulated, spring-loaded arm lets you aim the beam precisely over a desk.

Design Process

Impact

With two hours in the sun, the solar lamp offers 3 hours with a single LED on and 1.5 hours when both LED lights are activated. Currently distributed to two siblings, and in the works of finding investors to manufacture more lamps for rural schools in Valuka, Bangladesh.

This is a short video overview of the new iterations.

This is a short video overview of the 2026 FTC competition robot.

FTC Robot (Team 14156) — Mecanum drivetrain, intake + outtake

Why this project

For the 2024–25 season, FTC's "Into The Deep" game challenged robots to collect and score small "pixels" into high baskets and ascend a central structure. As lead designer for Team 14156, I developed a mecanum-drive robot with a vertical roller intake, bucket outtake, and dual linear slides to reach the high basket and climb the structure.

Design Process

Chassis

We chose a square goBILDA frame for packaging and serviceability, then modified members for slide clearance and an auxiliary motor mount for the collection linkage.

Linear Slides

Two independent slide motors raise a cross-hopper. Early on, uneven rise skewed the hopper and changed the outtake angle. I planned encoder-sync and strengthened mounts where space allowed.

Spools

Initial issue: Narrow spools let the string wander and slip onto the axle.
Update: Widened the spool and added a center rim to split up/down runs and prevent crossovers.

Hooks (ascent)

I slanted the hook geometry to increase tolerance during lining up to the rung. The profile reduces "perfect hit" requirements and speeds driver handoff into the climb.

Final 3D Designs

What I learned

• Syncing dual actuators (even just with sensors + guard rails) prevents cascading skew issues.
• Routing + labeling saves matches: two-color lines, idlers, and clear tie-offs cut pit time.
• Designing for the game meta (high basket + high ascent) focuses effort where points matter.

Why this project

As Team 14156 scaled up from a finalist-alliance season to Super-Qualifiers, we rebuilt our process: plan earlier, CAD end-to-end, and design for repeatable scoring. The portfolio captures that shift: from whiteboard sketches to Fusion 360 assemblies, 3D-printed parts, and a drivetrain + manipulator package tuned for consistent cycles and endgame execution.

Intake

After evaluating a claw concept, the team adopted a vertical roller intake with custom flexible flaps—stiffer than surgical tubing for better force transfer. Iterated from two-end to three-end flap geometry.

Outtake

Simple, reliable bucket outtake with a small roller to transfer pixels from the ramp into the bucket—then a quick flip to score on the backdrop.

Linear Slides

Switched from prior REV slides to Misumi linear rails for speed and weight savings; upgraded to durable Kevlar string to prevent fray.

Side Plates

CNC-cut polycarbonate side plates provide rigid mounting for slides and electronics (REV Hubs).

Drone Launcher

Onboard endgame launcher: rubber-band energy, servo release, whiteboard → CAD → 3D-print pipeline.

Build Process & Materials

• CAD-first: Full Fusion 360 robot model; more up-front reviews to reduce rushed late-build time.
• 3D-printing choices: Polycarbonate for load-bearing parts; PETG/ASA for non-critical; Filaflex (82A) for flexible intake flaps.

Programming Stack

• Android Studio + GitHub for collaborative dev and version control.
• Road Runner for autonomous pathing + localization; Dashboard for real-time telemetry.
• Computer Vision: Built on the FTC SDK opmode and adapted for team needs.

Season Takeaways

• Advanced to Super-Qualifiers
• End-to-end CAD pipeline
• Faster, lighter slides
• More reliable cycles

• Front-load planning & CAD to avoid build crunch.
• Design transfer paths to keep game pieces controlled end-to-end.
• Choose materials by role: polycarb for structure, flexible filaments for intake compliance.
• Use shared tools (GitHub, Dashboard) to speed debug and driver iteration.

Why this project

At Reverie Power & Automation Engineering Ltd., I produced a fabrication-ready radiator drawing for a power transformer. The goal: convert specs into a clean, buildable layout with correct fin count, flange geometry, and service features—so a shop can manufacture, assemble, and maintain the radiator bank with zero ambiguity.

Quick heat check

Heat per fin           = 275 W
Fins per radiator      = 20
Number of radiators    = 8
Total fins             = 20 × 8 = 160
Total dissipation      = 160 × 275 = 44,000 W

What I learned

• Spec → shop floor: turn client notes into unambiguous dims, weld notes, and a BOM that fabricators trust.
• Cooling logic: how fin area, oil path, and air side ΔT set the radiator's effective heat transfer.
• Tolerancing & assembly: flange PCD/slots for alignment; lifting points placed for safe rigging.
• Serviceability: valve/oil-cock & clearance placement so a single radiator can be isolated and swapped.

Transformer cooling — ONAN in one minute

Losses heat the oil. In ONAN (Oil-Natural/Air-Natural), hot oil rises through headers into finned radiators, sheds heat to ambient air by natural convection, then returns cooled. No pumps or fans — gravity and temperature differentials do all the work. The radiator fin count and geometry directly set how much heat the transformer can safely dissipate.