Skip to content

Latest commit

 

History

History
485 lines (350 loc) · 22.1 KB

File metadata and controls

485 lines (350 loc) · 22.1 KB

Mamba MIL-STD Validation

Scope

This document defines the environmental validation approach for the Mamba autonomous drone platform. Testing is self-performed and documented per the tailored methods below. Compliance is claimed per individual test method, not to MIL-STD-810H as a whole.

Applicable standards:

Standard Title Application
MIL-STD-810H Environmental Engineering Considerations and Laboratory Tests Methods 501.7, 514.8, 516.8
MIL-STD-31000 Technical Data Packages Design documentation structure
MIL-HDBK-217F Reliability Prediction of Electronic Equipment MTBF calculation

Platform under test:

Subsystem Component Critical interfaces
Flight controller Pixhawk 1 (hw v2.4.5) GPS connector, peripheral VDD line
Propulsion RCTIMER 5010 360KV × 4 Arm mounting, ESC wiring
Frame 2mm CF plate + 13.2mm CF tubes Arm-to-base joints, vibration damping
Compute Raspberry Pi 5 4GB + Hailo8 USB/CSI connections, thermal
Power Skywalker Quattro 25A×4 + BEC Power module connector
Telemetry SiK radio pair TELEM1 serial port
Navigation u-blox LEA-6H GPS + compass GPS port (known power issue, see guidance)

Known issue: GPS VDD peripheral line drops to ~1V if GPS is connected during boot. This is a pre-existing hardware limitation of the Pixhawk clone and must be accounted for in all test procedures. GPS must be connected after boot or powered via TELEM1 VDD brick line.


1. MIL-STD-810H, Method 514.8 — Vibration

1.1 Purpose

Verify that the Mamba airframe, avionics mounting, and electrical connections survive and operate correctly under vibration levels representative of multirotor flight. Of particular concern:

  • Pixhawk 1 mounting (IMU sensitivity to frame vibration)
  • Carbon fiber arm-to-base joints (adhesive/mechanical retention)
  • GPS connector reliability (known fragile connector)
  • RPi 5 + Hailo8 module retention and thermal interface
  • Propeller balance and motor shaft integrity

1.2 Tailoring rationale

MIL-STD-810H Method 514.8 defines multiple Categories. For a small UAS platform, Category 24 (Jet Aircraft Stores — Materiel Installed in Jet Aircraft) is overly severe. The tailored approach uses:

  • Procedure I — General Vibration as baseline
  • Vibration profile derived from actual flight data (accelerometer logs from Pixhawk) rather than generic profiles, per 810H guidance on "use measured data when available"
  • Test duration: 1 hour per axis (3 axes) — reduced from the 810H default of 4 hours per axis for qualification, acceptable for engineering development testing

1.3 Test equipment required

Equipment Specification Alternative
Accelerometer ADXL345 or MPU-6050 breakout, ≥200 Hz sample rate Pixhawk onboard IMU log (VIBE message)
Data logger Raspberry Pi or laptop with serial/I2C logging QGroundControl .ulog export
Vibration source Electrodynamic shaker (preferred) Controlled motor run at various RPM on fixed test stand
Mounting fixture Rigid plate replicating operational mounting The actual airframe, fixed to bench

1.4 Test procedure

Pre-test:

  1. Photograph all connectors, joints, and mounting points. Record torque values on all fasteners.
  2. Record baseline IMU data: power on Pixhawk, log 60s of stationary VIBE data. Record X, Y, Z vibration levels and clipping counts.
  3. Perform functional check: GPS lock, telemetry link, RPi boot and Hailo8 inference test.
  4. Mark all connectors with alignment marks (paint pen) to detect any movement.

Test execution (motor-induced vibration):

Since a shaker table may not be available, the following motor-run profile substitutes:

Step Throttle Duration RPM (approx.) Rationale
1 25% 15 min ~3000 Ground idle / pre-flight
2 50% 20 min ~5500 Hover
3 75% 15 min ~7500 Cruise / maneuver
4 100% 5 min ~9000 Max thrust
5 Sweep 25→100→25% 5 min Variable Resonance sweep

Repeat with each axis oriented vertically (rotate airframe on fixture) if using a shaker. If using motor-run, a single orientation (normal flight attitude) is acceptable.

During test:

  • Log Pixhawk VIBE data continuously (.ulog)
  • Monitor for audible rattling, visible connector loosening
  • At each throttle step, verify telemetry link active and GPS lock maintained

Post-test:

  1. Repeat all pre-test inspections. Compare photographs.
  2. Check all alignment marks for movement.
  3. Record post-test VIBE baseline (same 60s stationary test).
  4. Perform full functional check.
  5. Inspect propeller balance and arm joints for cracking or delamination.

1.5 Pass/fail criteria

Parameter Pass Fail
Pixhawk VIBE levels X, Y, Z < 30 m/s² Any axis ≥ 60 m/s²
Clipping events 0 during stationary post-test Any clipping in post-test
Connector displacement No movement on alignment marks Visible displacement
Structural integrity No cracks, delamination, or loosened fasteners Any structural damage
Functional check All subsystems operational Any subsystem failure

1.6 Deliverables

  • Test plan (this document)
  • Pre-test photographs and baseline data
  • Raw .ulog files for all test steps
  • Post-test photographs and comparison
  • Test report with pass/fail determination

2. MIL-STD-810H, Method 501.7 — High Temperature

2.1 Purpose

Verify that Mamba electronics and structural components operate correctly at elevated temperatures representative of hot climate field deployment. Critical concerns:

  • Raspberry Pi 5 thermal throttling (known to throttle at 85°C die temperature)
  • Hailo8 accelerator thermal limits
  • LiPo battery performance degradation and safety
  • Carbon fiber epoxy matrix softening (typically Tg > 120°C, not expected to be limiting)
  • Pixhawk IMU accuracy drift with temperature

2.2 Tailoring rationale

Method 501.7 defines storage and operational temperature categories. For a tactical UAS operating in European/Mediterranean climates:

  • Operational hot temperature: 45°C (Category A2 — Basic Hot)
  • Storage hot temperature: 60°C (Category A2)
  • Test duration: 4 hours at operational temperature (reduced from 810H qualification default of 7×24h cycles, appropriate for development testing)
  • LiPo batteries excluded from storage test — tested per manufacturer limits only, not subjected to 60°C due to safety risk

2.3 Test equipment required

Equipment Specification Alternative
Climate chamber Programmable to 60°C ±2°C Insulated box + ceramic heater + PID controller
Temperature sensors K-type thermocouples or DS18B20 digital RPi internal temp sensor + IR thermometer
Data logger Multi-channel thermocouple logger RPi logging via 1-Wire bus
Timer Software timer

2.4 Test procedure

Pre-test:

  1. Record ambient temperature and humidity.
  2. Attach temperature sensors to: Pixhawk case, RPi 5 SoC (via vcgencmd measure_temp), Hailo8 module, ESC heat sink, motor windings (IR spot check), battery surface.
  3. Perform baseline functional check at ambient temperature.

Test execution — Operational (45°C):

Step Action Duration
1 Place fully assembled drone (without propellers) in chamber
2 Ramp to 45°C at ≤1°C/min ~20 min
3 Soak at 45°C 60 min (thermal stabilization)
4 Power on all electronics, begin functional test
5 Run Pixhawk self-test, verify IMU calibration 10 min
6 Run Hailo8 inference benchmark (fixed model, record FPS) 10 min
7 Activate telemetry link, verify stable connection 10 min
8 Maintain powered operation at 45°C 120 min
9 Ramp down to ambient at ≤1°C/min ~20 min

Test execution — Storage (60°C):

Step Action Duration
1 Remove LiPo batteries. Place drone in chamber
2 Ramp to 60°C at ≤1°C/min ~35 min
3 Soak at 60°C, unpowered 4 hours
4 Ramp down to ambient at ≤1°C/min ~35 min
5 Allow 1 hour ambient stabilization, then full functional check

During test:

  • Log all temperatures at 10-second intervals
  • Monitor RPi throttle flag: vcgencmd get_throttled (bit 19 = over-temperature)
  • Record Hailo8 inference performance every 15 minutes

Post-test:

  1. Visual inspection for deformation, discoloration, delamination.
  2. Full functional check at ambient temperature.
  3. Compare Hailo8 inference FPS to baseline.
  4. Check all connectors and solder joints.

2.5 Pass/fail criteria

Parameter Pass Fail
RPi 5 SoC temperature < 85°C at 45°C ambient ≥ 85°C (throttling)
Hailo8 inference rate ≥ 90% of ambient baseline FPS < 80% of baseline
Pixhawk IMU Passes self-test, no thermal cal errors Self-test failure
Telemetry link Stable connection throughout Link loss > 5 seconds
Structural No visible deformation or delamination Any structural change
Post-storage functional All systems nominal at ambient Any subsystem failure

2.6 Deliverables

  • Test plan (this document)
  • Temperature log (all sensors, full duration)
  • RPi throttle flag log
  • Hailo8 benchmark results (ambient vs. hot)
  • Pre/post photographs
  • Test report with pass/fail determination

3. MIL-STD-810H, Method 516.8 — Shock

3.1 Purpose

Verify that Mamba survives handling drops and hard landings without loss of function. This is particularly relevant for:

  • Pixhawk IMU (accelerometer and gyroscope damage from impact)
  • GPS connector (known fragile, see guidance notes)
  • Carbon fiber tube arm joints (brittle failure mode under impact)
  • RPi 5 and Hailo8 board-level connections (BGA solder joints)
  • Propeller retention on motor shafts

3.2 Tailoring rationale

Method 516.8 defines multiple procedures. The applicable ones for a portable UAS:

  • Procedure I — Functional Shock: simulate hard landing
  • Procedure IV — Transit Drop: simulate handling/transport drop

Tailored shock levels:

Procedure Shock specification Rationale
Functional shock (hard landing) 20g, 11ms half-sine pulse Representative of 2m/s vertical impact on hard surface
Transit drop 1.0m drop onto 50mm plywood over concrete Per 810H Table 516.8-II for items 0–9 kg

3.3 Test equipment required

Equipment Specification Alternative
Drop test rig Guided drop fixture with quick-release Measured height drop from string release
Impact surface 50mm plywood over concrete floor Per 810H spec
Accelerometer ADXL377 (±200g range) or similar high-g sensor Pixhawk IMU may clip (limited to ±16g on MPU-6000)
High-speed camera ≥240 fps Smartphone slow-motion
Measuring tape Steel tape, calibrated

3.4 Test procedure

Pre-test:

  1. Full functional check: Pixhawk self-test, GPS lock, telemetry, RPi + Hailo8 boot and inference.
  2. Photograph all structural joints, connectors, propeller retention.
  3. Record accelerometer baseline noise floor.
  4. Remove propellers for safety. Install with landing gear as in operational configuration.

Test execution — Transit drop (Procedure IV):

Drop Orientation Height Surface
1 Flat (bottom down, normal landing attitude) 1.0 m 50mm plywood/concrete
2 Nose down (forward arm leading) 1.0 m 50mm plywood/concrete
3 Side (left arms leading) 1.0 m 50mm plywood/concrete
4 Inverted (top down) 1.0 m 50mm plywood/concrete
5 Corner (single arm leading) 1.0 m 50mm plywood/concrete

For each drop:

  1. Suspend drone at specified height and orientation using quick-release mechanism.
  2. Start high-g accelerometer and high-speed camera recording.
  3. Release. Allow drone to come to rest.
  4. Record peak g-level and pulse duration from accelerometer.
  5. Visual inspection for damage before proceeding to next drop.
  6. Stop testing if structural failure is detected — document and report.

Test execution — Functional shock (Procedure I):

If a shock machine is available, apply 20g half-sine 11ms pulse in each of 3 orthogonal axes, both directions (6 shocks total). If not, the transit drop test above provides adequate engineering-level shock data.

Post-test (after all drops):

  1. Full functional check (identical to pre-test).
  2. Detailed photographic inspection of all joints, connectors, boards.
  3. Check propeller shaft runout (spin by hand, visual wobble check).
  4. Verify Pixhawk IMU calibration is still valid (compare pre/post accelerometer offsets).
  5. Verify GPS lock acquisition time is within normal range (< 60 seconds cold start).

3.5 Pass/fail criteria

Parameter Pass Fail
Structural integrity No cracks, delamination, broken joints Any structural failure
Pixhawk function Passes self-test, IMU cal within ±5% of pre-test Self-test failure or cal drift > 10%
GPS Cold start lock < 60s post-test No lock or lock > 120s
RPi 5 + Hailo8 Boots and runs inference normally Boot failure or inference errors
Telemetry Link established normally Link failure
Propeller shafts No visible runout Wobble or shaft bend
Connectors All seated, no displacement on alignment marks Any connector unseated

3.6 Deliverables

  • Test plan (this document)
  • Pre-test photographs and functional check record
  • High-speed video of each drop
  • Accelerometer data for each drop (peak g, pulse duration)
  • Post-test photographs and functional check record
  • Test report with pass/fail determination

4. MIL-STD-31000 — Technical Data Package

4.1 Purpose

Structure the Mamba design documentation as a Technical Data Package (TDP) conforming to MIL-STD-31000. This is a documentation standard, not a test. It demonstrates the ability to deliver engineering data in a format accepted by defense procurement.

4.2 Applicability

MIL-STD-31000 defines three TDP types:

Type Content Mamba applicability
Conceptual Functional requirements, trade studies Partial — README + design rationale
Developmental Drawings, specs, test plans Primary focus
Production Full manufacturing package Out of scope

4.3 Required TDP elements for Mamba (Developmental level)

The following mapping shows how existing and planned Mamba documentation maps to MIL-STD-31000 requirements:

MIL-STD-31000 Element Mamba document Status
System specification README.md + operational concept ⬜ Expand
Engineering drawings hardware/*.FCStd (FreeCAD) ⬜ Export to PDF/STEP
Parts list / BOM doc/design.md (partial) ⬜ Formalize as structured BOM
Interface documents doc/guidance.md (wiring, pinouts) ⬜ Expand with connector tables
Test plans doc/mil-std.md (this document)
Test reports ⬜ After test execution
Software documentation doc/ai.md, doc/vision.md ⬜ Expand with SW architecture
Associated data img/ directory
Quality provisions ⬜ Add inspection criteria

4.4 Implementation plan

Phase 1 — Drawing package:

  1. Export FreeCAD models (drprt.FCStd, frame.FCStd) to:
    • STEP AP214 format (3D exchange)
    • PDF drawings with title block, revision, dimensions, tolerances
  2. Create assembly drawing showing all components in context.
  3. Assign drawing numbers: MAMBA-DWG-001 through MAMBA-DWG-NNN.

Phase 2 — Bill of Materials:

Create structured BOM in tabular format:

Item Part number Description Qty Manufacturer Specification
1 MAMBA-001 Base plate, 2mm CF 1 Custom MAMBA-DWG-001
2 MAMBA-002 Arm tube, 13.2mm OD CF 4 Generic ASTM D3039
3 RCTIMER-5010 Motor 360KV pancake 4 RCTIMER Vendor datasheet
4 RCTIMER-16x55 Propeller 16×5.5" CF 4 RCTIMER Vendor datasheet
5 PX4-FMUv2 Pixhawk 1 (hw 2.4.5) 1 3DR/clone PX4FMUv2.4.5 schematic
6 UBEC-SW4x25 Skywalker Quattro ESC 1 Hobbyking Vendor datasheet
7 SiK-TEL SiK telemetry radio 2 Generic SiK firmware spec
8 UBLOX-LEA6H GPS + compass module 1 u-blox LEA-6H datasheet
9 RPI5-4G Raspberry Pi 5 4GB 1 Raspberry Pi Ltd Product brief
10 HAILO-8 Hailo8 AI accelerator 1 Hailo HAILO-8 datasheet

Phase 3 — Interface Control Document (ICD):

Document all electrical interfaces:

  • Pixhawk ↔ GPS (UART + I2C, known VDD issue)
  • Pixhawk ↔ ESC (PWM channels 1–4)
  • Pixhawk ↔ SiK radio (TELEM1 UART, 57600 baud)
  • Pixhawk ↔ RPi 5 (MAVLink over serial or UDP)
  • RPi 5 ↔ Hailo8 (PCIe M.2)
  • Power distribution diagram with voltage rails and current budgets

4.5 Deliverables

  • Drawing package (STEP + PDF)
  • Structured BOM (CSV + PDF)
  • Interface Control Document
  • TDP index document listing all data items

5. MIL-HDBK-217F — Reliability Prediction

5.1 Purpose

Calculate predicted Mean Time Between Failures (MTBF) for the Mamba electronics using MIL-HDBK-217F methodology. This is a calculation exercise, not a physical test. It demonstrates the ability to perform quantitative reliability engineering.

5.2 Method

MIL-HDBK-217F provides two prediction methods:

  • Parts Count — uses generic failure rates by component category. Simpler, suitable for early design.
  • Parts Stress — uses detailed operating conditions (voltage derating, temperature, etc.). More accurate.

For Mamba, the Parts Count method is appropriate. The Parts Stress method can be applied selectively to critical components (Pixhawk, RPi 5).

5.3 Environmental profile

MIL-HDBK-217F defines environments with different π_E (environmental factor) multipliers:

Environment code Description π_E multiplier (typical)
G_B Ground, Benign 1.0
A_UF Airborne, Uninhabited Fighter 12.0
A_RW Airborne, Rotary Wing 8.0

For Mamba, A_RW (Airborne, Rotary Wing) is the closest match. This is conservative for a small multirotor but demonstrates worst-case thinking.

5.4 Parts count reliability prediction

The following table uses generic base failure rates (λ_b) from MIL-HDBK-217F Tables for each component category, adjusted by the environmental factor.

Component Category (217F) Qty λ_b (per 10⁶ hrs) π_E (A_RW) λ_p = Qty × λ_b × π_E
Pixhawk (microprocessor-based) Hybrid/MOS complex IC 1 0.060 8.0 0.480
Raspberry Pi 5 SoC VLSI MOS microprocessor 1 0.14 8.0 1.120
Hailo8 accelerator VLSI MOS ASIC 1 0.14 8.0 1.120
BLDC motors (5010) Motor, AC 4 5.0 8.0 160.000
ESC (4-in-1, power MOSFET) Power semiconductor 4 ch 0.50 8.0 16.000
u-blox GPS module Hybrid IC 1 0.060 8.0 0.480
SiK radio module RF/transmitter hybrid 2 0.10 8.0 1.600
Connectors (est. 20 pins) Connector, general 20 0.010 8.0 1.600
Wiring harness (est.) Wiring, general 1 0.002 8.0 0.016
LiPo battery Battery, rechargeable 1 2.0 8.0 16.000
TOTAL λ_SYSTEM ≈ 198.4

5.5 MTBF calculation

MTBF = 1 / λ_SYSTEM = 1 / (198.4 × 10⁻⁶) ≈ 5,040 hours

This is the predicted MTBF for the full system under airborne rotary-wing environmental conditions.

Interpretation:

  • At 1 hour average mission time, this predicts approximately 1 failure per 5,040 missions.
  • The dominant failure contributors are the BLDC motors (81% of total failure rate), followed by ESC and battery (8% each).
  • Electronics-only MTBF (excluding motors, ESC, battery) would be approximately 156,000 hours — indicating that mechanical/electromechanical components drive overall reliability.

5.6 Recommendations from reliability analysis

  1. Motor reliability is the critical path. Consider: bearing quality, motor temperature monitoring, vibration-based predictive maintenance.
  2. ESC derating: Verify that motor current draw at max throttle does not exceed 80% of ESC rating (25A × 80% = 20A per channel).
  3. Connector reliability: The known GPS connector issue is consistent with connector failure being a significant contributor. Consider: conformal coating, strain relief, or connector upgrade.
  4. Battery: Implement cell voltage monitoring and enforce safe discharge limits (no LiPo cell below 3.3V under load).

5.7 Deliverables

  • Parts count prediction spreadsheet (this section, expanded to full 217F format)
  • Reliability block diagram
  • Critical items list (top 5 failure contributors)
  • Recommendations for reliability improvement

6. Test execution tracking

Test Standard Status Date Report
Vibration MIL-STD-810H Method 514.8 ⬜ Planned
High temperature MIL-STD-810H Method 501.7 ⬜ Planned
Shock / drop MIL-STD-810H Method 516.8 ⬜ Planned
Technical data package MIL-STD-31000 🔨 In progress
Reliability prediction MIL-HDBK-217F ✅ Initial calc complete Section 5 above

References

  1. MIL-STD-810H, Environmental Engineering Considerations and Laboratory Tests, Department of Defense, January 2019. Available: everyspec.com
  2. MIL-STD-31000B, Technical Data Packages, Department of Defense, November 2018. Available: everyspec.com
  3. MIL-HDBK-217F Notice 2, Reliability Prediction of Electronic Equipment, Department of Defense, February 1995. Available: everyspec.com
  4. PX4 Autopilot Documentation, Vibration Isolation. Available: docs.px4.io
  5. Raspberry Pi 5 Datasheet, Thermal Management. Available: raspberrypi.com