# Walter R08: thermal design and physical qualification protocol

Status: development protocol, not laboratory certification, safe-temperature approval or final-print authorization. The complete build temperature is currently unknown. The existing air-only diagnostic cannot select an intake/exhaust fan or predict the finished head temperature.

## Current evidence and its meaning

- Actual Pi recording:62.0–64.8°C over117seconds, fan5147–5329RPM, no recorded throttle flags. Maxwell reports the current setup is a large cardboard V0 enclosure. Exact room temperature, cooler identity and matched workload are not documented. The recording is too short to establish steady state and does not measure the printed head.
- Calculated field:0.5second startup,12W assigned directly to the Pi PCB boundary, fan off, no solid conduction or thermal mass, other loads omitted.541254 air cells,29948 concave cells, narrow vents/seams unresolved. Sampled air/boundary temperature reaches267.37°C. This is a rejected-model diagnostic, not a forecast of CPU temperature or proof that the enclosure overheats.
- Exact current mesh checks found no overlaps above1mm³ between modeled solids and the four existing fan-intake, rear-harness, GPIO-adapter and future-power-bay reservations. Actual cable connectors and cooler revision remain unverified.
- The build has20 printed pieces and13 electronics references. Those33 objects are not33 required prints and are not33 validated thermal regions. Do not use the arbitrary volume of a solid electronics envelope as its actual thermal mass.

## Design candidates to compare

A. Existing R08 vents, complete rear cover, actual cooler and routed wiring.

B. Removable separator/baffle to keep the cooler's warm discharge from returning directly to its intake. Confirm blower direction before drawing the duct. Keep cable bends and the Pi service pull path clear. Preserve the smooth external helmet and symmetric ear speakers.

C. A selected case exhaust fan with a matching clear intake route. Intake-fan alternatives may also be compared if packaging, dust management or acoustic requirements favor them. Evaluate pressure-flow operating points, acoustic pickup at the microphone, and flow obstruction. Never equate a Pi blower's free-air rating or RPM with helmet exhaust flow. An extra fan is a test candidate, not a predetermined requirement.

Do not count speaker grilles as cooling vents without measuring the path through installed drivers, cloth and cavities. Start thermal qualification on a correctly rated external supply. Add a selected protected battery/converter configuration in a separate test; external power is a test sequence, not a permanent design decision.

## Freeze the actual test configuration

Maker records part numbers/revisions, measured dimensions and connector exit/bend envelopes, mounting hardware and tolerances, cable lengths/routes, power source and protections, exact fan/cooler, display brightness, audio setting/content, active camera/radar/IMU/touch modules, software/configuration versions and workload recipe. Treat optional or absent modules explicitly as absent. Unknown active loads are not0W.

Use the component register with geometry hashes. Measure main-display and both small-screen current, audio input under representative playback, Pi input, camera/mic/radar/hub consumption, regulator losses and selected battery behavior. Measure simultaneous total input as well as branches without double-counting downstream power. Nameplate speaker power is not its constant consumption. PMIC rail subtotal is not total Pi/system input. Do not apply total Pi board watts only to the CPU package unless the partition is experimentally justified.

## Instrumentation and records

Synchronize timestamps across the Pi log, calibrated external temperature logger and power meter. Record instrument model/serial, calibration date/report, range, accuracy, sampling rate and probe attachment. Record blank values for missing readings. A thermal camera may locate hotspots, but requires emissivity/reflection control and contact-sensor corroboration; shiny metal readings alone are insufficient.

Minimum monitored channels: room air upstream away from exhaust; CPU sensor; heatsink near its base without disturbing the thermal pad; cooler intake air; enclosure discharge air; hottest loaded shell/mount locations; total system input watts; fan RPM/control and throttle flags. For a battery test add pack case/cell-proxy temperature per its maker guidance and converter temperature. Use small probes and strain relief so sensors do not materially obstruct vents. Log additional hotspots separately when needed; the browser CSV is a core-channel review tool, not a replacement for the full laboratory data file.

Airflow: measure integrated inlet/outlet flow or velocity traverses with known area, instrument correction and uncertainty. A single spot velocity multiplied by gross vent area is not enough when flow is nonuniform. Measure pressure differential at the installed fan/vent system using appropriate taps. Smoke visualization, if appropriate for the test setup, is qualitative and must not contaminate electronics or replace flow measurement.

## Test sequence

1. First-fit coupons and mounts: dimensions, screw access, rear removal, screen retention, cable bend/strain relief, electrical continuity/protection and stable power. Install the actual cooler correctly. This remains necessary even if numerical checks pass.
2. Open-bench baseline with all intended peripherals on external power. Log idle, normal intended Walter operation and highest intended sustained simultaneous operation. Do not silently substitute CPU-only stress for the real workload.
3. Complete R08 engineering prototype with all covers, grilles, wiring and actual hardware. Repeat the same loads at matched ambient and power. Compare CPU rise above ambient and every external probe.
4. Repeat with separator/baffle candidate, then a selected extra fan if needed. Change one controlled variable at a time. Measure microphone noise and speaker interaction as well as temperatures.
5. Repeat trials after thermal reset to quantify variability. Include the intended highest room ambient, nominated orientation and plausible cable/vent obstruction conditions only within an approved supervised procedure.
6. Add the selected battery and converters. Test supported discharge and charging/use modes separately, following the chosen manufacturers' limits. No charging architecture is currently approved.
7. Review fault detection and protective response. Begin with software alarm/instrument checks; do not intentionally stall a real fan, block vents or overheat a battery during an unattended run. Set stop rules before any supervised physical fault test.

## Proposed project review criteria

These are engineering targets to review, not a cited product standard or certification.

- Run at least30minutes and extend until all critical temperatures and ambient have settled. Review the final10minutes: absolute fitted temperature slope≤0.05°C/min and span≤1°C, sampling gaps≤15seconds. Periodic fan/load cycling may require a longer cycle-based analysis instead; do not label it steady by averaging away peaks.
- Stable matched input power: final-window span≤5% of mean (0.1W numerical allowance at very low input). Capture fast startup/transient electrical events with suitable instrumentation separately; a5second log can miss them.
- Initial CPU design target≤75°C at the nominated highest operating ambient and intended sustained load. Review/stop the test at80°C; do not count operation while throttled as satisfactory thermal design. The Pi manufacturer's throttling behavior is a device protection, not protection for the battery, plastic or all peripherals.
- Every battery, converter, display, adhesive, printed mount and touchable surface needs its own reviewed limit with measurement uncertainty and margin. Those limits are currently unselected. Filament printing temperature or melting temperature is not an allowable service temperature. Account for load, creep, print orientation/infill and fastening stress.
- Passing the browser's measurement checks only makes data suitable for review. It does not approve calibration records, whole-build safety, cooling capacity or physical fit.

## Numerical model requirements

Build coupled fluid/solid heat transfer: board/packages, thermal pads/contact, heatsink fins/blower, shell, main display, auxiliary screens, audio and selected power components. Include actual heat capacity, conductivity and radiation/external convection where applicable; justify any simplified or homogenized region. Include physical cable/hub obstruction. Model closed joints according to their actual overlap/seal; do not silently seal unresolved gaps.

Use measured/traceable load cases, installed fan pressure-flow curves with control hysteresis, and proper pressure/hydrostatic initialization. Run zero-heat equilibrium and known heat-transfer reference checks. Repair rejected mesh quality and resolve pressure-loss-critical features. Perform at least three systematic mesh levels, time-step and exterior-domain sensitivity; quantify uncertainty in CPU/package temperature, component hotspots, vent mass flow and pressure. A fixed number of cells across a vent is a starting choice, not proof of adequacy.

Check mass closure and full transient energy balance including solid/fluid stored energy and all boundary transport. A12W wall integral only verifies an applied boundary, not conservation of the complete system. Record residuals, field extrema, mesh hashes, settings and run logs. Expose convergence failures rather than displaying them as accepted cooling predictions.

Calibrate only nominated uncertain parameters against designated baseline measurements. Validate on independent held-out closed-head/cooling cases; compare at the same probe positions and time histories, with repeatability and instrument/placement uncertainty. Do not tune separately to each validation run. Define tolerances before viewing validation results and retain adverse outcomes.

## Print release

Release the final cosmetic print only after the engineering prototype meets fit/retention/serviceability, electrical, thermal, acoustic and material limits with selected hardware. Obtain maker/printer-specific compensation and physical assembly sign-off. Preserve the smooth helmet exterior, but do not sacrifice cooling or cable service space to appearance. This protocol reduces mistakes; it cannot guarantee zero manufacturing errors.

## Primary references

- Raspberry Pi thermal management and current fan-control documentation: https://www.raspberrypi.com/documentation/computers/raspberry-pi.html
- Manufacturer cooler/case bench comparison: https://www.raspberrypi.com/news/heating-and-cooling-raspberry-pi-5/
- NASA verification versus validation: https://www.grc.nasa.gov/www/wind/valid/tutorial/overview.html
- NASA systematic grid-convergence guidance: https://www.grc.nasa.gov/www/wind/valid/tutorial/spatconv.html
- OpenCFD coupled fluid/solid solver: https://doc.openfoam.com/2306/tools/processing/solvers/rtm/heat-transfer/chtMultiRegionFoam/