Main power input protection¶
Everything the board runs on enters through one path:
battery / alternator bus → BAT+ terminal → protection stage → protected rail → LMR36510 buck → 5 V rail → TLV62569 buck → 3.3 V rail
The protection stage is a TPS48000-Q1 controller (U12) driving two back-to-back MOSFETs. Everything downstream of those MOSFETs (the buck regulator, both logic rails, and the field driver's supply) is isolated from the bus in about 5 µs when the controller trips on overvoltage.
Sensor inputs have their own protection and are not covered here.
Where the field connects¶
The field return diode's cathode lands on the BAT+ input terminal, on the battery side of the protection MOSFETs. Two things follow, and both matter more than any single trip level on this page:
- Field current never crosses the protection MOSFETs, in either wiring type. The freewheel loop runs outside them, so isolating the board does not strand the field's stored energy. On N-type it runs through the return diode, the BAT+ terminal, the Regulator BAT+ wire, and the alternator output cable. On P-type it runs through D5, the Regulator Ground Wire, and the engine ground.
- The protected rail is fed only through the protection MOSFETs. Nothing back-feeds it, so a trip shuts the board down.
Together those make a protection trip with the field energized a clean event: the board goes dark, the field switch (Q3) opens as its driver rail decays, and the winding's current decays in its freewheel loop. The field wiring itself is covered by the field fuse; see Fusing on the P-type or N-type wiring page.
Controller¶
TPS48000-Q1, VSSOP-19 (DGX). A 100 V automotive high-side protection controller with an integrated charge pump for N-channel pass MOSFETs.
Features used in this design:
- Reverse-polarity tolerance to −65 V
- Programmable overvoltage protection
- Programmable undervoltage lockout
- Fast MOSFET shutdown (~5 µs)
- Charge pump for N-MOSFET gate drive
The part is AEC-Q100 qualified. Automotive electrical environments, with load dump, cold crank, and large transients, are generally harsher than marine ones, which is what makes an automotive part the right starting point here.
Protection MOSFETs¶
Infineon BSC072N08NS5, TDSON-8 (SuperSO8 5×6 mm), two in back-to-back configuration. The same device is used for Q3 in the alternator field drive stage.
| Parameter | Value |
|---|---|
| Drain-source voltage (Vds max) | 80 V |
| Rds(on) max @ Vgs = 10 V | 7.2 mΩ |
| Continuous drain current (Id) | 47 A (Tc = 100 °C; 74 A at Tc = 25 °C) |
| Gate threshold voltage (Vgs(th)) | 3.8 V |
| Total gate charge (Qg) | 29 nC @ 10 V |
| Input capacitance (Ciss) | 2.1 nF |
| Reverse transfer capacitance (Crss) | 26 pF |
| Power dissipation (Pd) | 69 W |
Sources are tied together at the SRC node (U12 pin 13); gates are tied together at the G_MOSFET node. Back-to-back sources cancel the body diodes in both directions, so the pair blocks reverse-polarity current and gives full bidirectional isolation on a fault.
Conduction loss is negligible in normal operation: 14 mW at 1 A through both devices, and 1.4 W at the 10 A rating of the external BAT+ fuse.
Gate drive and inrush current control¶
PU (pin 15), the pull-up. Sources gate current from the bootstrap rail toward SRC, through R30, 1 MΩ. R30 sets the turn-on slew: the pull-up charges the gates at roughly 10 µA, so the pass MOSFETs enhance over about a millisecond rather than nanoseconds.
PD (pin 14), the pull-down. Sinks the gate to SRC directly, with no series resistance, so the gate discharges in about 5 µs on an overvoltage trip. R30 plays no part in how fast the board isolates.
Bootstrap capacitor C5 = 100 nF between BST (pin 12) and SRC (pin 13). The charge pump holds about 11 V above SRC across the whole input range, so the MOSFETs stay fully enhanced at any battery voltage.
Turn-on behavior. With ~11 V of bootstrap headroom across 1 MΩ, gate current is about 10 µA. The source follows the gate, so the protected rail ramps up over roughly 1 ms at any input voltage, about 0.05 V/µs at 48 V. Inrush into the 21.5 µF bulk bank is then about 1 A at 48 V, and proportionally less at 12 V, so the board can be plugged onto a live bus without a current step the external fuse would see.
Why the ramp rate matters. The trace between the MOSFET stage and the bulk capacitors (15 mm, 2 mm wide, 1 oz copper, ≈ 8 nH) forms an underdamped LC tank with that bank:
| Parameter | Value |
|---|---|
| Trace inductance (L) | ≈ 8 nH |
| Bulk capacitance (C) | ≈ 18.8 µF at the rail node |
| Resonant period | ≈ 2.4 µs |
| Resonant frequency | ≈ 410 kHz |
| Characteristic impedance (Z₀) | ≈ 21 mΩ |
| Damping (Rds(on) + shunt + ESR) | ≈ 17 mΩ |
| Damping ratio (ζ) | ≈ 0.4 |
A turn-on edge fast compared with the 2.4 µs tank period would ring the rail up above the bus, and the overshoot scales with system voltage: at 48 V it would approach the buck regulator's 70 V absolute maximum. The 1 ms ramp spans hundreds of tank periods, so the tank is never excited and the rail rises cleanly at every system voltage.
Charge pump ripple. The charge pump runs in burst mode to keep quiescent current low, which puts about 1 V peak-to-peak of ~21 Hz ripple on the gate drive. It is intrinsic to the part, independent of load and input voltage, and harmless: the minimum gate-source voltage stays well above full enhancement.
Undervoltage lockout (EN/UVLO)¶
Divider R45 = 750 kΩ / R70 = 220 kΩ, 1 %, on pin 1 (INP is strapped to the same node).
| Threshold | Pin voltage | Battery voltage |
|---|---|---|
| Rising, minimum | 1.176 V | 5.19 V |
| Rising, typical | 1.23 V | 5.42 V |
| Rising, maximum | 1.287 V | 5.67 V |
| Falling, typical | 1.136 V | 5.01 V |
Below about 5 V the controller stays off, so a flat or disconnected bank leaves the board off rather than browning out mid-operation. Divider current is 12 µA at 12 V.
Overvoltage protection¶
Divider R90 = 511 kΩ (0.1 %) / R91 = 10.2 kΩ on the OV pin, a ratio of 0.01957, so a 1.225 V typical comparator threshold puts the trip at 62.6 V.
| Condition | Battery voltage |
|---|---|
| Trip, minimum | 59.8 V |
| Trip, typical | 62.6 V |
| Trip, maximum | 65.3 V |
| Re-enable, earliest | 55.6 V |
| Re-enable, typical | 58.2 V |
| Re-enable, latest | 60.6 V |
When OV crosses the rising threshold the controller pulls PD to SRC, discharging both gates and opening the pass MOSFETs. Propagation delay is 4.5 µs typical, 5.4 µs maximum. Recovery is automatic once the bus falls back below the falling threshold.
On a 48 V system the window between an absorption setpoint near 57–58 V and a low-tolerance falling threshold is narrow, so a board that has tripped on overvoltage may stay off until the bank comes down off absorption. That is the intended behavior of a protection trip, but it is the one case where it is visible as more than a momentary dropout.
Divider current is 23 µA at 12 V and 115 µA at 60 V.
Reverse polarity¶
Three things cover a reversed battery connection:
- The back-to-back MOSFETs cancel each other's body diodes, so no current path exists through the pass stage.
- The crowbar diode across the input terminals (anode on GND, cathode on BAT+) conducts the fault at one junction drop and clears the external BAT+ fuse, so the fault is carried by a diode rated for it and ends in a blown fuse rather than anywhere else on the board. It is idle at all other times.
- The controller itself tolerates −65 V at its input.
D3, an SS2H10 Schottky from GND to the protected rail, clamps negative excursions on the rail itself.
Transient protection¶
D6, an SMBJ64CA bidirectional TVS, sits across the raw input:
| Parameter | Value |
|---|---|
| Stand-off voltage | 64 V |
| Breakdown voltage | 71.1–78.6 V |
| Clamp voltage | 103 V @ 5.8 A |
| Peak pulse power | 600 W (10/1000 µs) |
It is off in normal operation and starts conducting above roughly 71 V. It protects the raw side, between the battery terminals and the MOSFET stage, and does not clamp the protected rail, whose clamp voltage of 103 V would in any case be above the buck regulator's 70 V absolute maximum. The protected rail is kept safe by the overvoltage trip disconnecting it at 60–65 V, well before the TVS ever conducts, and by the controlled turn-on ramp.
Raw-side bulk capacitance (C38 10 µF, C40 1 µF) absorbs the high-frequency edge of a surge ahead of the TVS.
Lightning. A nearby strike couples surges of either polarity into vessel wiring. The layered response is the TVS clamping the fastest energy, the input capacitors absorbing high-frequency content, the controller detecting the overvoltage, and the MOSFETs isolating everything downstream. A direct strike is not survivable by any of this, and designing for one is not attempted.
Downstream energy buffer¶
Capacitance on the protected rail, all 100 V rated:
| Capacitor | Quantity | Location |
|---|---|---|
| 4.7 µF | ×4 | rail node, ~15 mm from the MOSFET stage |
| 100 nF | ×1 | rail node |
| 2.2 µF | ×1 | at the buck regulator's VIN pin |
| 220 nF | ×1 | at the buck regulator's VIN pin |
Total ≈ 21.5 µF. The four 4.7 µF parts sit at the rail node and connect back to the MOSFET stage through a 2 mm wide, 1 oz copper trace; the remaining two are at the buck regulator's own pins.
The bank buys time during the controller's shutdown delay. A 20 A surge into the node raises it at 0.93 V/µs, so the rail moves about 5 V in the ~5 µs before isolation completes.
What the board does in each fault¶
Both wiring types behave the same except in the rows that split them, because the field's freewheel loop runs outside the pass MOSFETs in both. The loop itself differs by type throughout. On N-type it runs through the return diode, the BAT+ terminal, the Regulator BAT+ wire, and the alternator output cable; on P-type it runs locally to ground, through D5, the Regulator Ground Wire, and the engine ground.
"Board shuts down" means exactly that: the protected rail is isolated, the ESP32 stops, and the field driver's rail decays, opening Q3.
| Event | Behavior |
|---|---|
| Overvoltage on the bus | A load dump, a BMS opening under charge, or a failed external charger. The controller trips at 60–65 V and isolates the board in ~5 µs. The field's stored energy decays in its freewheel loop, through the return diode (N-type) or through D5 (P-type). The board restarts by itself when the bus falls below the falling threshold. |
| Short on the BAT+ wiring | The external BAT+ fuse clears it. The protection stage trips on voltage, not on current, so the fuse is what ends a short on that wire. P-type: the field decays through D5 as above. N-type: the short holds the BAT+ terminal at ground, so the winding is excited through the return diode into the short, as with reversed leads below, for as long as the short lasts. The current comes from the battery through the alternator output cable, so it flows with the engine stopped too. Full field current is normal current for the 10 A field fuse, so the fuse stays whole while the battery drains and the rotor heats with no fan turning to cool it. Pull the field fuse to end it, and do not run the engine until the short is repaired. |
| Short on the protected rail | The external BAT+ fuse again. The rail has no other feed, so nothing else on the board conducts into the fault, and the board stays dark until the fuse is replaced. |
| Battery leads reversed | Crowbar diode carries the fault and clears the BAT+ fuse. Pass MOSFETs block, board undamaged. P-type: the field is fed from the battery through the bridge, so the blown fuse also removes field power. N-type: the field is fed from the alternator's own B+, and with the leads reversed the winding is excited through the return diode (alternator B+ → winding → return diode → BAT+ terminal) for as long as that terminal has a path to the battery. Do not run the engine until the wiring is corrected. |
| Bus above 71 V | The overvoltage trip has already isolated the board at 60–65 V. D6 absorbs the raw-side energy. A TVS that fails short takes out the BAT+ fuse, which on P-type also removes field power (the bridge is downstream of the fuse); an N-type field is fed from the alternator's own B+ and is not affected by that fuse. |
| Battery below ~5 V | Undervoltage lockout holds the stage off until the bus recovers. |
| Firmware shutdown or field-off | Q3 opens; the winding's current decays in its freewheel loop, through the return diode, the Regulator BAT+ wire, and the alternator output cable (N-type), or through D5, the Regulator Ground Wire, and the engine ground (P-type). Nothing reaches the protected rail. |
| Battery opened while charging | P-type: the field freewheels locally through D5 and the board shuts down on the lost supply. N-type: it depends on where the circuit opens. When the Regulator BAT+ wire and the alternator output cable stay joined, as they do when a drop-in battery's internal BMS opens, or a battery switch between the bank and a bus bar that carries both, the freewheel loop stays intact and the field decays in it. When the opening separates the two and leaves the Regulator BAT+ wire without the battery, for example a battery switch with the Regulator BAT+ wire on its load side and the alternator output cable on the battery side, the return path ends at an open terminal, so the winding's energy lands on the raw input side and is absorbed by D6. When it is the alternator output cable that loses the battery and the Regulator BAT+ wire stays on it, the board stays powered and keeps regulating, and a field cut decays into the battery through the alternator's own rectifier diodes, not into D6. In every case, opening a charging path under load is a load dump for the alternator as well: shut charging down before opening a battery switch or terminal. |
| Field wire shorted | Outside this stage entirely. The field switch is not fed through the pass MOSFETs, and a locked-on field is normal current to the field fuse. See the short hazard warning on the P-type or N-type wiring page. |
Standby current¶
Worst-case controller quiescent current is 55 µA.
| VIN | EN/UVLO divider | OV divider | Controller Iq | Total |
|---|---|---|---|---|
| 13 V | 13 µA | 25 µA | 55 µA | 93 µA |
| 26 V | 26 µA | 50 µA | 55 µA | 131 µA |
| 52 V | 52 µA | 100 µA | 55 µA | 207 µA |
At 52 V that is about 10.8 mW of standby dissipation.
With the battery absent, the controller has no supply and draws nothing from the battery rail; the TPS2116 power multiplexer routes USB power to the 5 V rail instead, at a few microamps of its own.