# P3.6 first physical experiment — dry-air routing and reaction Status: protocol and nominal geometry only. Physical measurements: **0**. The first experiment tests one useful part of CryoPulse: a finite powered air stream redirected through controlled upper outlets can produce a measurable downward reaction, with an asymmetric outlet producing a yaw moment. It does not establish a high-speed stopping distance, a self-charging cycle or a tyre coefficient. Keep the cold branch separate from this dry, room-temperature rig. ## A measurement boundary that can close Mount the complete air unit on one six-axis balance: both blowers, collector, short feed duct, manifold, valves and actuators. Air enters from the stationary room and leaves through the four measured outlets. Keep power/data leads in loose symmetric service loops and measure their effect. An external pressurised hose would apply a feed reaction that could be mistaken for useful duct force; it is excluded from this first complete-unit force measurement. Two **Sanyo Denki 9BMB24P2F01** blowers are candidates in parallel, not series. Each is rated at 24 V, 10.8 W and 4500 rpm, with catalogue maxima of 1.04 m³/min free airflow and 410 Pa static pressure. These two maxima are different operating points. A parallel pair does not double static pressure. The measured installed curve must establish the real flow and input power. Catalogue maximum pressure is not a certified relief-device setting. The four-outlet manifold uses 30 mm bare bores, 29.1 mm soft-seat free bores, split 5 mm stubs, 0.8 mm gate plates and actual ROBOTIS XC330 actuator CAD. The rig is a 0–40 °C, dry-air concept with a 500 Pa working target and a separate 750 Pa pressure-test target requiring a qualified fixture and procedure. The full car's 150 ms valve target is not qualified by a 250 ms bench-stroke target. Select a force-balance mounting and complete-unit mass below its combined six-axis load envelope. The candidate ATI Nano17 SI-25-0.25 has 35 N Fz range and 250 Nmm moment range; neither number establishes achieved accuracy. Centre the gravity load over the balance, including the collector and two blowers. Measure all six static loads before energising. Detailed collector, blower brackets, balance adapters and guarding remain fabrication-detail work. ## Instruments and declared targets * Full mounted-system expanded force uncertainty target: 0.02 N; yaw: 0.001 Nm. These must be demonstrated by calibration, repeatability and cable tests. * Dry differential pressure: Sensirion SDP810-500Pa candidate; manufacturer range ±500 Pa, accuracy 3% of reading plus 0.1 Pa zero-point accuracy. Its bulk silicon temperature is not an independent gas-temperature measurement. * Independent air temperature, ambient pressure, inlet and four outlet flow traverses, blower tachometers, four actual gate angles, supply voltage/current on every electrical branch, and common timestamps. Log at 200 Hz or faster. * Determine each outlet momentum-profile factor from a velocity traverse. A uniform-jet approximation is a planning screen, not a measured momentum map. * Flow uncertainty target 2% requires a suitable calibrated reference. Do not install a restrictive meter whose pressure drop consumes the blower margin. ## Commission and calibrate 1. Independently inspect the as-built bore, gate travel, seats, stubs, joints, balance mount and electrical limits. Complete the separately qualified leak and pressure checks. Do not connect liquid nitrogen to this rig. 2. Calibrate force and moment in both directions using traceable loads over the actual measurement range. Include cross-axis terms and the installed cables. 3. Record zero before and after each run. Repeat the zero with leads in the allowed positions. Retain this variation as uncertainty, not a fitted force. 4. Establish actual blower flow/pressure/power curves with open outlets first. Pressure-trip target 375 Pa; verify real overshoot and instrument response. The operating limit is the lowest qualified component/fixture limit. 5. Characterise each gate's angle, stroke time, current, temperature, opening hysteresis and leakage. Stall torque is not a continuous actuator rating. ## Preregistered sequence Use at least five independent repeats per steady state, randomise state order, settle for 5 s, retain 10 s of steady data and a separate transition trace. Repeat an all-open reference between asymmetric states to track drift. * Blowers off; valves all open; all channels and zeros recorded. * Both blowers at 25%, 50%, 75% and 100% PWM with all outlets open. * At the highest qualified steady point: each single gate closed in turn, each bank isolated, and mirror-image left/right restrictions. * All four gates step open/closed at the verified low-pressure duty, with actual angle and full electrical input logged. Do not hold an unqualified stall. * Repeat relevant states with the alternative valve/stub assembly only after a matched flow comparison is defined. Preserve both equal-flow and equal-power comparisons; either can answer a different engineering question. ## Reduction and acceptance Choose +z downwards for reported reaction and document the balance transform. Integrate measured velocity profiles and outlet pressure forces to predict reaction for the **complete-unit boundary**. For a simple planning estimate with equal outlets and ambient exit pressure, Fdown = ρ Q² sin(130°) / A_total. This formula predicts about 0.20 N at 25 L/s for four 30 mm exits and ρ=1.2 kg/m³. It does not promise that the chosen blowers reach that flow. Estimate mean uncertainty from repeat runs and one-second blocks, so correlated 200 Hz samples are not treated as independent. Add calibration, zero drift, cross-axis, lead/fixture, flow-profile and dimensional uncertainties. Report coverage assumptions. A retained synthetic software check is labelled SYNTHETIC and must never be counted as a physical run. Accept a downward-reaction demonstration only when the lower uncertainty bound of the zero-corrected force is positive, the measured control-volume balance closes within its combined uncertainty, and electrical input is fully recorded. Accept controlled yaw only when mirror restrictions reverse the moment with consistent sign and uncertainty. Publish leakage, response and losses even if a target is missed; change the gate, collector or control and repeat matched runs. ## Primary supplier references * https://products.sanyodenki.com/en/sanace/dc/blower/9BMB24P2F01/ * https://docs.robotis.com/docs/dxl/model_reference/x_series/xc_series/xc330-m288/ * https://sensirion.com/products/catalog/SDP810-500Pa * https://www.ati-ia.com/products/ft/ft_models.aspx?id=Nano17 No parts have been ordered and no supplier has been contacted.