On e-stop the Racer does not push energy back onto the bus. It stops the motor with its own brake resistor and, later, a mechanical brake (an active-release disk brake with a solenoid).
The hardware e-fuse class is fixed per output by part population. A load's declared contract is accepted only if it fits inside that class. A load that asks for more is refused and flagged.
High-power output target: 1 kW at 48 V, 10 mA resolution. Proposed spec: 25 A continuous (1 kW at a 40 V pack), ±50 A sense range for pre-trip capture, 10 mA resolution, ±0.5 % of reading accuracy. Slow precision monitor for telemetry and coulomb counting, plus a fast path for the hardware trip and the capture.
Note on G4: on the NineBot hub (λ = 0.026 Wb, 15 pole pairs) the peak line back-EMF at 483 rpm is about 34 V, below the 40 to 54 V bus. A coasting wheel with the bridge off pushes no current until roughly 650 to 700 rpm. The brake resistor handles active braking after the bus is gone and the overspeed case.
Open, with the recommended answer
#
Question
Recommendation
G1
Does S stand for Sensing or Sensors? The bus bar brief says Sensors, ADR-0001 says Sensing.
Sensing
G2
Authority split: the bus bar owns power and e-stop, the MIB owns motion intent, every node owns its local safety.
Yes. "No central controller" becomes "no single node that every function depends on".
G3
The bus bar owns the e-stop: a two-channel hardware input opens every motion output without passing through the MCU, then firmware publishes ESTOP/STO on the network. No separate wire to each Racer.
Yes. Replaces the per-zone loop in ADR-0006.
G5
Adopt the firmware's stop behaviour: SAFE_STOP and command loss ramp then hold; ESTOP and faults go Hi-Z. Rewrite PRD F-04.
Yes
G6
Every Racer, seating included, gets a bus bar output. Low-power RJ45 nodes use PoE (802.3bt, up to about 71 W at the device). Single-pair power over data is limited to small sense nodes (about 20 W and under).
Yes. Rewrites ADR-0004's logic-power plan.
G8
Drive outputs (1 kW and 300 W) are fully reverse-blocking, so an off output is off in both directions. Compute and accessory outputs use a plain high-side FET.
Yes
G9
The brake solenoid runs from the drive's motion output, so any cut engages the brake without firmware. A hardware hold-up of about 300 to 500 ms lets the electrical stop happen first.