[BUG] TLV76733DRVR thermal margin is insufficient at the annotated 800 mA

[RodrigoCAndrade]

1. Technical Objective

U20 is a TLV767 in the DRV package (6-pin WSON, 2 × 2 mm). At the load annotated on sheet 3/3 (3.3 V @ 800 mA):

• V_IN after D3 ≈ 4.6 V under load
• P_D = (4.6 − 3.3) × 0.8 ≈ 1.04 W

For a 2 × 2 mm WSON with moderate copper (RθJA in the 70–80 °C/W range) that is roughly 80 °C of rise: Tj ≈ 105 °C at 25 °C ambient, and close to thermal shutdown at 50–60 °C ambient.

2. Proposed Solution

1. Bench-measure worst-case current (LoRa TX at 22 dBm plus Wi-Fi active) before freezing the topology.
2. If sustained current stays below ~500 mA, keep the LDO but allocate generous copper pour and thermal vias under the pad during layout.
3. If not, replace the LDO with a buck converter, which removes the 1.3 V linear drop entirely.

3. Alternatives Considered

• Larger LDO package (HVSSOP DGN, 3 × 3 mm): better RθJA, but the dissipated power is unchanged.

References:

• [1] TLV767 datasheet: https://www.ti.com/lit/ds/symlink/tlv767.pdf

[RodrigoCAndrade]

Worst case (all peaks simultaneous + continuous): ~572mA normal / ~656mA with CAN fault. Electrically the TLV767 is fine, but treating that as continuous exceeds TJ=125°C above ~38°C ambient on a JEDEC-standard board (RθJA=77.7°C/W).

Datasheet has no transient thermal data, so this can't be closed on paper. Keeping open, not moving to a buck yet.

Plan: fab with thermal vias + copper pour under the pad (already planned), then measure real TJ under worst realistic firmware load (Wi-Fi TX + CAN together) via thermal camera/thermocouple on the package top + ΨJT (4.3°C/W). Fall back to a bigger LDO package before considering a buck if margin is bad.