SPACE status
Nodes tracked
9
2 in use or validated
Requirements met
13of 53
17 partial · 23 open
Decisions to make
11
1 accepted
High-severity issues
3
from PRD §14
Nodes by pillar
Every board on or near the SPACE network. Click one for its page.
Unmet requirements by area
Open plus partial, PRD v0.1 register with the updates below applied
Changed since PRD v0.1
Requirement statuses that moved in other repos
| ID | Was | Now | Why |
|---|---|---|---|
| E-08 | Open | Partial | Per-board watchdog exists. The MIB owns the deadman (joystick released means SAFE_STOP). No fabric-wide e-stop yet. |
| H-09 | Open | Partial | Console, firmware update and reset over Ethernet (tcp :5555, discovery udp :5556) since 2026-10-15. No secondary UART header yet. |
| F-01 | Open | Partial | ATOS motor API v2 (rammp-rtps MotorCommand/MotorState). v1 verified on one board 2026-09-26; both chair drive boards run v2 since 2026-10-15. Not yet run with two boards under load. |
| F-10 | Open | Partial | 200 ms command watchdog is in firmware. It ends in SAFE_STOP then an active HOLD, which differs from the PRD's coast-out STOPPING. See ADR-0005. |
| B-05 | Open | Open | Supervisor role went to the MIB, not ABC. Revisit ABC's scope in ADR-0002. |
Decision queue
Proposed decisions, in the order to grill them
| ADR | Decision | Pillars | Status |
|---|---|---|---|
| 0001 | Rename PACE to SPACE | SPAC | Accepted |
| 0002 | Zonal architecture | PC | Proposed |
| 0003 | Physical link chosen by criticality, not bandwidth | SAC | Proposed |
| 0004 | Two power domains, logic independent of actuation | PA | Proposed |
| 0005 | Stop sequence across node and bus bar | PAC | Proposed |
| 0006 | Keep a hardwired e-stop loop alongside the network | PC | Proposed |
| 0007 | Wire protocol on 10BASE-T1S segments | SC | Proposed |
| 0008 | Time-synchronised timestamps in the telemetry schema | SC | Proposed |
| 0009 | Node role from topology, identity from the MCU | C | Proposed |
| 0010 | MCU and PHY platform for new nodes | SAC | Proposed |
| 0011 | Single-pair connectors and harness standard | PC | Proposed |
| 0012 | Zone switch silicon, build or buy | C | Proposed |
Known issues
High and medium severity, PRD v0.1 §14
| # | Issue | Severity |
|---|---|---|
| 1 | Hung LM75/I2C read silently blocks the stream task → thermal guard blind | High |
| 2 | RTPS delivery collapses to 0 above ~485 Hz with pub_fail = 0 | High |
| 3 | Phase-B high-side intermittent VDS faults near transients | High |
| 4 | DRV8353 VDS latch survives ESP reset; requires VM power cycle | Medium |
| 5 | USB-JTAG console wedge; only USB re-enumeration clears | Medium |
| 6 | Current-sense conversion factor is a placeholder in the BSP | Medium |
| 7 | Phase-C channel condemned; p auto disabled | Medium |
| 8 | Hall lines on ADC1-adjacent GPIO3/9 | Medium |
| 9 | late% 4.9 % at 500 rpm | Medium |
Roadmap, next quarter
Written before the SPACE direction. Re-plan once the decisions settle.
| Milestone | Exit criteria |
|---|---|
| R1: Thermal-guard robustness | Guard demonstrably trips with an LM75 unplugged mid-run |
| R2: Command ingress + heartbeat | Host publishes a 100 rpm setpoint over DDS; pulling the cable stops the motor within the bound |
| R3: Calibration persistence | Cold boot → drive in hall mode without hcal |
| R4: Gate-drive review | Root cause recorded; envelope extension approved or design change filed |
| R5: Speed campaign | Report §9 with FET plateau temperature and continuous-power rating |
The PRD was frozen on 2026-09-14. Work has moved on in other repos since then. This page records what changed so the status view is current without rewriting the PRD.
Firmware and network
- The Racer got a production app,
atos-drive(branchfeat/atos-driveinpace-racer-fw). It is hall-only, has a hard speed cap in every mode, and implements the ATOS motor API. - The wire contract moved into a shared repo,
rammp-org/rammp-rtps. It holds topic names, XCDR1 structs, and the axis table (front_caster_left,front_caster_right,drive_left,drive_right). This answers most of PRD open question 3: the project already chose its own IDL-as-C++-structs over ROS 2 standard messages. - espp rtps moved from 1.2.0 to 1.3.3 on 2026-10-09. 1.3.3 fixes a reader-proxy leak on MIB restarts. The MIB-restart case has not been re-run on the bench since.
- A switch with IGMP snooping and no querier broke SPDP discovery in one direction. The MIB now discovers boards by unicast from its DHCP lease table. This matters for the managed-switch decisions.
Nodes the PRD did not list
- MIB (MEBot Interface Board), ESP32-P4,
RAMMP-mib-fw. It is the supervisor: it runs DHCP, commands each wheel, and owns the deadman. Its README lists "control of PACE RACER motor controllers over PoE" as planned work. - HMI / joystick, M5Stack Tab5 (ESP32-P4) on the RAMMP HMI PCB, W5500 Ethernet,
pace-hmi-fw. Its UI follows RAMMP UI spec V2 and is in daily use. - Power bus Gen 1.5,
RAMMP-ElectronicsbranchGen1.5_powerbus(Ariana, 2026-10-07). It is a passive distribution board: XT90 in, four XT60 out. - Brake resistor,
pace-brake-r. It is a KiCad design for protecting the battery from regen on high-power controllers. The bus bar brief makes each Racer responsible for its own brake resistor. - Dyno torque bench,
dynopace. A MotorGo Plink reads a DY-N-200 torque sensor over RS485.
New direction (not yet in the PRD)
- The project renamed PACE to SPACE (ADR-0001).
- Smart bus bars, managed switches, and single-pair Ethernet with power over data lines. The proposed decisions ADR-0002 through ADR-0012 cover these.