BME 261: Prototyping, Simulation and Design
Estimated study time: 8 minutes
Table of contents
Sources and References
Primary texts — Ulrich, Eppinger, and Yang, Product Design and Development, 7th ed. (McGraw-Hill). Budynas and Nisbett, Shigley’s Mechanical Engineering Design, 11th ed. (McGraw-Hill).
Supplementary texts — Horowitz and Hill, The Art of Electronics, 3rd ed. (Cambridge). Chacon and Straub, Pro Git, 2nd ed. (Apress, open). Lemieux and Schoukens, MATLAB and Simulink for Engineers (MathWorks open courseware).
Online resources — MIT OCW 2.008 Design and Manufacturing II and 6.131 Power Electronics Lab. Ultimaker and Prusa open FDM design guidelines. Autodesk Fusion 360 documentation. Git SCM official documentation. Stanford Design Group open lectures on prototyping.
Chapter 1: The Integrated Prototyping Mindset
1.1 Why Prototype
A prototype is a physical or computational embodiment of a design hypothesis. Prototyping shifts uncertainty from belief to evidence: will this mechanism clear the obstacle, will this amplifier reject line noise, will this algorithm converge. Biomedical systems are multi-domain — mechanical, electrical, software — and prototyping forces the domains to meet.
1.2 Fidelity and Purpose
The fidelity of a prototype should match the question it answers. Cardboard and zip ties answer layout questions. A breadboarded op-amp circuit with clip leads answers small-signal questions. A finite-element model answers stress-concentration questions. A Simulink model with a hardware-in-the-loop fixture answers closed-loop stability questions. Over-investing in fidelity before the right question is asked wastes time; under-investing guarantees irrelevant answers.
Chapter 2: Mechanical Design and CAD
2.1 Parametric Modelling
Modern CAD is feature-based and parametric: sketches are constrained geometrically and dimensionally, features extrude or revolve sketches, and a model tree records the design intent. Equations link dimensions; configurations capture variants; assemblies mate parts through geometric relations. Edits at upstream parameters propagate downward through the tree.
2.2 Design for Manufacturing
Each process imposes design rules. Injection molding requires draft angles typically 1–3°, uniform wall thickness, and avoidance of sharp internal corners. Fused-deposition modelling is layer-anisotropic, with tensile strength roughly 60–70% of bulk along the layer and weaker across. Laser cutting produces kerfs requiring nominal allowance; CNC milling defines internal corner radii by the cutter. Early consultation with manufacturing constraints avoids redesign after tooling.
2.3 Tolerances and GD&T
Tolerance stack-up follows either worst-case or statistical summation. For \( n \) independent Gaussian contributors with standard deviations \( \sigma_i \),
\[ \sigma_{\text{stack}} = \sqrt{\sum_{i=1}^{n} \sigma_i^2} , \]substantially smaller than the worst-case sum. Geometric dimensioning and tolerancing (ASME Y14.5) separates size from form, orientation, and location, letting designers specify only the tolerance that matters functionally.
Chapter 3: Electrical Prototyping
3.1 From Breadboard to PCB
A breadboard is fine for DC and audio-frequency exploration; at RF or with fast edges, parasitic capacitance and inductance between adjacent rails dominate behaviour. Printed circuit boards replace ad hoc wiring with controlled impedance, proper ground returns, and manufacturable reproducibility. Schematic capture and layout in tools such as KiCad separate electrical intent from physical placement; design-rule checks enforce clearance, track width, and annular ring.
3.2 Signal Conditioning for Biosignals
Bioelectric potentials are small (µV–mV) and sit atop large common-mode disturbance from the mains. Instrumentation amplifiers with CMRR > 100 dB, active driven-ground shields, and isolation barriers are standard. The effective input-referred noise
\[ v_{n,\text{tot}} = \sqrt{\int_{f_1}^{f_2} S_v(f)\,df} \]sets the lower bound on resolvable signal; component selection and bandwidth design choose the tradeoff.
3.3 Sensors, Actuators, and Interfacing
Resistive, capacitive, piezoelectric, optical, and MEMS inertial sensors span the biomedical measurement space. Bridge circuits linearize small resistance changes; charge amplifiers match piezoelectric sources; transimpedance amplifiers match photodiodes. On the actuator side, motor drivers, PWM, and H-bridges control DC motors; solenoids and shape-memory alloys deliver discrete motion.
Chapter 4: Software, Simulation, and Version Control
4.1 Embedded Software Basics
Microcontrollers (ARM Cortex-M, AVR, ESP32) run bare-metal or RTOS code. Peripheral access uses memory-mapped registers; tight timing requires interrupts and DMA. For biomedical devices, deterministic timing and bounded latency matter as much as computational throughput.
4.2 Simulation
Multibody dynamics (Simulink, Simscape, OpenSim) models mechanisms before fabrication. Circuit simulation (SPICE) checks bias points, gain, stability, and noise. Finite-element analysis (FEA) checks stress, deformation, and thermal response. Simulation is not truth — it is a model with assumptions — but a calibrated model exposes sensitivities and failure modes far more cheaply than experiment.
4.3 Revision Control with Git
Git tracks snapshots of a project and the history of changes through a content-addressable object store. The essential operations:
\[ \text{init} \rightarrow \text{add} \rightarrow \text{commit} \rightarrow \text{push} \rightarrow \text{pull} \rightarrow \text{merge} . \]Branches enable parallel development; merges reconcile histories; tags mark releases. For teams, pull-request workflows enforce review before code enters the main branch. For biomedical software, traceability between a commit and a design-history entry supports regulatory evidence.
v1.4.2 denotes major.minor.patch. Breaking API changes bump the major number; backward-compatible feature additions bump minor; bug fixes bump patch. A user pinning ^1.4 accepts any 1.x.y with x ≥ 4.Chapter 5: System Integration
5.1 Interfaces and Interface Control
When domains meet, the bugs live at the interface. Interface control documents specify electrical, mechanical, and communication contracts: connector pinouts, voltage levels, timing, message formats, units, and coordinate frames. Every integration error traceable to a missing interface specification is preventable by writing the specification first.
5.2 Development Cycles
Waterfall — sequential specify, design, build, test — suits well-understood products with regulated verification. Agile iterations — two-week sprints with working increments — suit exploratory problems with evolving requirements. V-model aligns each development step with a verification step. Medical-device practice commonly blends these: agile within a phase, structured across phases, with frozen interfaces at each level.
5.3 Debugging Integrated Systems
Observability is the design property that makes bugs visible. Logging, test points, serial consoles, oscilloscope probes, and dedicated debug modes cost little at design time and pay enormously during integration. The rule of thumb: every subsystem should expose its inputs, its outputs, and at least one internal state for inspection.
Chapter 6: Digital Manufacturing and Building a Working Prototype
6.1 Additive and Subtractive
Material extrusion (FDM), vat photopolymerization (SLA, DLP), powder bed fusion (SLS, SLM), and material jetting cover the common additive spectrum, each with resolution, strength, biocompatibility, and cost tradeoffs. Subtractive CNC milling and laser cutting remain unmatched for accuracy and for engineering plastics and metals. Hybrid workflows — 3D-printed housings over machined structural members — are routine.
6.2 Assembly and Fasteners
Heat-set threaded inserts convert plastic holes into reusable threaded fasteners. Snap fits and living hinges exploit polymer elasticity but require cyclic fatigue analysis. Adhesives (cyanoacrylate, epoxy, UV-cure) require surface preparation and cure control. Assembly sequences should be planned alongside geometry to avoid designs that cannot be put together by a human hand.
6.3 Verification
A finished prototype is not a finished product. Verification shows that the build meets its specifications: dimensional inspection, electrical characterization, software unit and integration tests. Validation — separate and later — shows that the product meets user needs. The habit of distinguishing these two, even in a student prototype, is a career-long asset.
