Thermal-Optical Validation Plan
Scope
Engineering characterization of the single-tube thermal and optical subsystem. These targets are preliminary engineering targets, not assay or diagnostic claims. Use safe color standards and noninfectious controls; assay and clinical validation remain separate.
Instrumentation
- Test rig USB CSV logging at >=5 samples/s thermal and configurable optical cadence.
- Calibrated reference thermometer and traceable dry-block or stirred-bath points for NTC calibration.
- Fine-wire K-type thermocouple connected at J6/MAX31856 and positioned in a sacrificial sealed tube or validated tube surrogate.
- Current-limited 12 V bench supply for bring-up, then the certified 12 V/3 A adapter.
- Oscilloscope on TP5/TP6 and DMM/current probe on input/heater path.
- Ambient-temperature sensor, stopwatch/reference timestamps, camera for condensation observations.
- Safe colored standards spanning expected hue/intensity range; opaque cap/blank fixture for dark tests.
Calibration
- Calibrate the block NTC chain at a minimum of 25, 50, 60, 65, 70, and 80 °C after soak. Record ADS1115 code, actual 3V3, reference temperature, and fit coefficients (Beta or Steinhart-Hart).
- Check thermocouple converter at ice point, ambient, and near 65 °C against a traceable reference. Record connector polarity and cold-junction offset.
- For optics, record dark offsets, white/reference tile response, channel linearity versus LED current or pulse width, and saturation thresholds.
Safety and fault-injection tests
- Verify F1 is the installed Littelfuse 02153.15MXEP 3.15 A time-lag part. At maximum normal load and worst-case ambient, measure fuse, holder, connector, and nearby-board temperature rise; repeat cold starts and representative PWM patterns to establish nuisance-opening margin.
- Create a controlled downstream overcurrent using a current-limited source and protected fixture. Confirm F1 clears before wiring, connectors, D1, or board copper exceed qualified limits. Do not short the certified adapter directly.
- Hold/reset/remove U1 while monitoring TP5 and TP6. Confirm R25 keeps
HEATER_PWM_CMD low and no heater pulse occurs through reset, bootloader entry, or an unpowered MCU.
- Open J4 during every active state and while serial traffic is incomplete. Confirm hardware gate drive and heater current fall independently of firmware parsing.
- Disconnect/open each block-sensor lead and substitute near-rail ADC voltages. Confirm a latched block-sensor fault and heater-off response.
- Disconnect, reverse, short, and intermittently disturb the thermocouple. Confirm MAX31856 FAULT handling, DRDY stale-data timeout, implausible value/rate rejection, and heater-off response. Before testing, continuity-check the corrected topology: U3
BIAS and J6 TC- are on connector-side TC_MINUS; R27 is the only series path from that node to U3 T- on TC_MINUS_FILT; C9/C10/C11 remain on the IC-side filtered nodes. Verify the filter does not create unacceptable settling delay or offset.
- Stop MAX31856 conversions or hold DRDY inactive in a test fixture. Confirm the firmware latches
thermocouple_stale within the declared 1 s limit.
- Feed a partial serial command without a newline for at least 10 s. Confirm the 100 ms control cadence, logging, interlock response, and overtemperature checks continue uninterrupted.
- Challenge equilibrium logic with one out-of-band sample, a >1.5 s sample gap, and a curved/drifting 60 s trace whose endpoint slope appears small. Confirm none can start
RUN; only all samples in band with an acceptable least-squares slope can pass.
- At cold startup with an intact and then an open thermocouple, confirm
ARM remains unavailable until the 2.0 s automatic open-circuit qualification interval completes and the MAX31856 status register has been read. Confirm the CSV retains the detailed current and latched fault bytes.
- Exercise the configurable active-cycle timeout using a shortened test build, then repeat a boundary test at the documented 90-minute engineering default or with a validated timebase simulation. Confirm expiry latches
active_cycle_timeout, heater duty goes to zero, and only the normal fault-clear procedure can return to SAFE.
Sample matrix
- At least 3 heater/block assemblies if available.
- At least 3 thermistors or repeated remounts of one thermistor.
- Tube brands/lots: minimum 3 representative lots.
- Fill volumes: 10, 15, 20, 25, and 30 uL, adjusted to intended workflow if different.
- Repeated hot insertions: >=10 per representative tube/volume condition.
- Ambient: low/nominal/high bench conditions, nominally 18, 23, and 28 °C.
- Optical repeats: >=10 reinsertions per standard, >=10 tubes per lot, with/without controlled bubbles and condensation.
Thermal procedure
- Inspect guard, cutoff loop, harness, and sensor attachment. Keep tube out.
- Preheat the empty block toward 65 °C. Log block temperature, liquid-reference channel, duty, interlock, and faults.
- Characterize time to first enter 65.0 ±0.5 °C, peak overshoot, settling time, steady-state mean/range/standard deviation, duty cycle, input current, and ambient.
- Open the guard: confirm TP6 and heater current fall to off without firmware cooperation.
- Reclose the guard, command guarded hot insertion, insert a sealed tube, and timestamp insertion.
- Track block recovery and liquid temperature. Define equilibration as 61 consecutive nominally one-second liquid samples, every sample within ±0.5 °C, no gap >1.5 s, a window spanning at least 60 s, and least-squares slope magnitude below 0.1 °C/min, unless data justify a revised criterion. Start the assay-duration timer only after this event.
- Continue for the intended hold duration. Observe condensation, top/bottom gradient, bubbles, and tube deformation.
- Repeat across matrix. Run a controlled separate one-shot cutoff qualification on a dedicated assembly; do not reuse a tripped cutoff.
Optical procedure
- With all four
LED_*_CMD low, acquire >=100 dark samples with optical chamber closed; repeat under bright room light to measure ambient rejection.
- Sequence white, red, green, and blue channels individually. Use fixed off-settle, on-warm-up, and acquisition windows; never overlap channels initially.
- Measure LED short-term warm-up, 30-60 min drift, supply sensitivity, and temperature dependence at room temperature and during a 65 °C thermal cycle.
- Measure tube removal/reinsertion repeatability, tube-to-tube variation, rotational sensitivity, fill-volume sensitivity, bubbles, condensation, and intentional ambient leaks.
- Preserve raw R/G/B/clear counts, integration time, gain, LED channel, command duration, dark counts, temperature, tube ID, orientation, and standard ID.
- Compute normalized values only in addition to raw values:
r=R/(R+G+B), g=G/(R+G+B), b=B/(R+G+B), and channel/clear ratios where clear is safely above dark.
Recorded CSV fields
Device CSV fields are timestamp_ms,state,fault,block_c,liquid_c,ads_code,heater_duty,interlock,tc_fault_now,tc_fault_latched,opt_led,opt_r,opt_g,opt_b,opt_clear,dark_r,dark_g,dark_b,dark_clear,liquid_slope_c_per_min,liquid_equilibrated,opt_command_ms,run_elapsed_ms,cycle_elapsed_ms. Host/operator metadata remains ambient_c,tube_id,tube_lot,fill_ul,orientation,standard_id,notes and must be joined by timestamp/run record rather than fabricated by firmware.
Preliminary engineering acceptance targets
Thermal
- Block overshoot =5 pooled standard deviations in the selected normalized feature space, with no saturation.
- Ambient-light shift with chamber closed: provisional <1% of illuminated signal after dark subtraction.
All optical thresholds remain provisional until real tube/dye measurements exist.
Analysis and design feedback
- Fit heater plant time constants, dead time, gain, required PWM frequency, and PID limits from step/insertion responses.
- Convert block-to-liquid offset versus fill, ambient, and elapsed time into firmware equilibrium logic and integrated-reader sensor placement requirements.
- Use maximum measured current and cable transients to revise fuse, connector, MOSFET, TVS, and wiring ratings.
- Use optical SNR, saturation, normalized separation, drift, and geometry sensitivity to select final sensor, emitter wavelengths/currents, chamber geometry, and calibration method.
- Reject any optical conclusion that depends on a single tube, standard, or unqualified breakout.