Design phase: no unit has been built or measured yet. Help build the first one

Open Bed Climate
Repo
Menu

Thermal & Fluidics

All numbers come from the lumped model in software/opod/src/opod/thermal.py. They are first-order estimates until a prototype is measured. Re-run the sweep after changing any parameter.

Model

  • TEC: TEC1-12706, with parameters fitted to its datasheet maxima (Qmax 50 W, ΔTmax 66 K, Vmax 14.4 V at Imax 6.4 A, 25 °C hot side): S = 0.0453 V/K, R = 1.78 Ω, K = 0.467 W/K (TecModule.from_datasheet). For a different TEC, enter its datasheet values; the test suite checks that the fit reproduces them.
  • Water loop per side: 1.2 kg water + 800 J/K hardware → C ≈ 5.8 kJ/K.
  • Sleeper load: UA body↔water 6 W/K at skin 34 °C (≈ 50–80 W into the bed when cooling). Room losses 2.5 W/K.
  • Cold plate: 12 W/K per TEC. Heatsink: R per side (fans on).

Sizing sweep (25 °C room, sleeper in bed, best-case current)

TEC/side Heatsink R (K/W/side) Coldest water TEC power/side Hot face
2 0.11 23.5 °C 143 W 48 °C
3 0.11 21.5 °C 189 W 55 °C
3 0.07 20.1 °C 217 W 47 °C
3 0.05 19.3 °C 213 W 41 °C
4 0.07 18.2 °C 250 W 51 °C
4 0.05 16.9 °C 291 W 46 °C

Design point (Duo): 4 TECs per zone, heatsink ≤ 0.07 K/W per zone, current capped at 4.5 A per TEC (see below). Pod 6 publicly uses 3 TECs per side. With TEC1-12706-class modules our model needs 4 to reach a similar range, so Pod 6 probably uses larger modules; that is not public. The Solo Starter kit uses 3 TECs at 0.11 K/W. Contributions with stronger modules (e.g. 12710/12715 class) are welcome: add their datasheet values and a kit.

Implications:

  • Under full sleeper load in a warm room, cooling bottoms out around 18–19 °C. The 13 °C end of the range is reachable only in a cooler room or with the bed unoccupied (pre-cooling). Pre-cooling before bedtime is the practical answer. The UI must show this honestly.
  • Heating is easy: TECs in reverse run at COP > 1. The constraint is the TEC water-side face, which runs about q/(12 W/K) above the water. Without a limit it reaches ~48 °C while heating to 43 °C. The controller therefore tapers heating current (0.5 A/K below 48 °C water). Simulated peak face is 45.3 °C, and 43 °C is reached in 7–12 min (all kits are tested in 16 °C and 22 °C rooms).
  • The heatsink is the main design lever. 0.07 K/W per side means a pair of tower-cooler-class fin stacks with 120 mm fans. Two-sided placement with fans between them (as in Pod 6’s public description) is a good layout.

Current cap: the cheap win

Cooling power has strongly diminishing returns above ~4.5 A per TEC (4 TECs, 0.07 K/W):

A per TEC Coldest water TEC power/zone
3.5 19.8 °C 101 W
4.0 19.1 °C 132 W
4.5 18.6 °C 167 W
5.0 18.3 °C 207 W
6.0 18.2 °C 298 W

The Duo Standard kit therefore caps at 4.5 A (max_tec_current_a). It loses 0.4 K versus 6 A but uses 44 % less power, so each zone fits one certified 280 W supply. The cap is a firmware limit (Kit.controller()) and the current limit written to the TEC driver. Tests run a full pull-down in a 30 °C room and check the supply stays under 80 %.

Lessons from earlier DIY builds (public)

  • opencooler V1 (Hackaday 205182, June 2026): cooling was too weak; one TEC side got warm. Heat rejection was undersized.
  • TechTeamGB (2024): heating worked better than cooling; quick disconnects didn’t self-seal.
  • Several builds drove TECs with raw PWM (capacitors ran hot). Open Bed Climate requires smooth DC.

All of this is consistent with our model: cooling is the hard part, and the heatsink is the main lever.

Water loop

  • Flow: 0.4–0.6 L/min per side. Q = ṁ·cp·ΔT gives ΔT ≈ 2 K at 70 W, so cover temperature is even.
  • Pump: 12 V brushless DC centrifugal pump with a tach output, ~1 m head, ≤ 5 W. A tach is mandatory: TECs stay off until flow is confirmed, and the MCU raises NO_FLOW after a 20 s grace period.
  • Reservoir: about 0.5 L internal reservoir with a fill port at the top, plus a capacitive level sensor (non-contact on the wall). Low level → LOW_WATER: TECs off, pump off.
  • Cold plates: aluminium (anodized) or copper water blocks, 40×40 mm per TEC. Use a corrosion inhibitor (distilled water + glycol/biocide) and avoid mixing copper and aluminium in the same loop.
  • Cover: TPU-laminated channel mat (welded TPU film, similar to medical cooling pads), 4 lines in the umbilical (supply/return × 2). Self-sealing quick couplings (CPC dry-break type) at the hub so unplugging doesn’t spill.
  • Priming: pump runs with TECs off until flow and level are stable. Air purge through the reservoir.

Open measurement items

  • Real UA body↔water through mattress topper fabric (sets everything else). Measure on the first prototype with a heated dummy (resistive pad at 34 °C).
  • TEC cold-plate thermal resistance with real paste and clamping.
  • Heatsink R vs fan noise at night (bedroom target ≤ 30 dB(A) at 1 m).

Rendered from docs/design/01-thermal-fluidics.md in the repository. View or edit the source.