devices(ubiquiti): the EdgePoint S16's sixteen PoE ports could light nothing (refs #130) #2202

Merged
jaylouisw merged 1 commit from catalogue/1119-1952 into main 2026-11-19 20:08:34 +00:00
Owner

Refs #130 (blocked on the Pro; this is the family's next real defect, not a close).

devices/ubiquiti/ep-s16.json carried sixteen poeOut ports and no poe block, so engine/simulator.js opened computePoe() with if (!dev.poeBudget) continue; and skipped the box: the ports were not "unknown", they could power nothing, and show power inline printed "budget not modelled".

Two more faults came out of the same census:

  • The class split was inverted. Ports 1-4 were poeOutClass: "PoE" (802.3af, 48 V negotiated); the vendor describes them as (4) 54V or 24V /1.4A, 4-Pair ... Passive PoE, Ports 1 to 4. Ports 5-16 are the ones that actually negotiate: (12) 802.3af/at or 24V/0.7A, 2-Pair. So the four 75.6 W ports were filed as the weakest, and a 48 V 802.3af PD on port 1 was told it would work on hardware with no 802.3 detection at all - the UISP Router defect again (test/uisp-r-poe.test.js).
  • Ports 1-2 were not PDs. (2) 54V / 1.5A, 4-Pair passive PoE input, so poeIn: 40 (the vendor's own Max. power consumption 40W) on eth0/eth1; cabling the switch to a PoE source now powers it instead of reporting an unpowered box.

Budget, derived and labelled as such. Neither surface prints a PoE-budget row. Both print (1) DC terminal block, 54V DC, 6A input = 324 W and Max. power consumption 40W (Excludes PoE output), so 324 - 40 = 284 W leave the ports on the stated DC-input configuration; poe.note also records the PoE-input alternative (2 x 54V/1.5A = 162 W in, so 122 W out). Same arithmetic shape as uisp-r/er-6p.

l2-forward added: the datasheet calls the EP-S16 "Layer-2 switch with some layer-3 capabilities" and lists Static/PBR routing, so l3-forward stays.

Sources (read 2026-11-19):

Trap recorded in research/ubiquiti-edgepoint-poe-census.md: EdgePoint_S16_DS.pdf, EP-S16.pdf, EdgePoint_S16.pdf and datasheets/edgemax/EP-S16_DS.pdf all 302 into the 609,911 B tech-spec front page - a 200 check reads that as a document.

Gates: syntax 463 OK; audit-port-types clean; audit-devices no hard defects; hardware invariants 26/0; census/description 3/0; audit-poe-budget silent PSEs 103 -> 102; run-tests 385 suites / 12,944 assertions, 0 failed; control-bytes clean; registry doc current (2,355 models).

Siblings left for their own increment, census written up in the research note: ep-r6 (five passive outputs, none modelled) and ep-r8 (seven outputs, two modelled, both mislabelled; ports 7-8 are combo cages with no comboPorts).

Refs #130 (blocked on the Pro; this is the family's next real defect, not a close). `devices/ubiquiti/ep-s16.json` carried sixteen `poeOut` ports and no `poe` block, so `engine/simulator.js` opened `computePoe()` with `if (!dev.poeBudget) continue;` and skipped the box: the ports were not "unknown", they could power nothing, and `show power inline` printed "budget not modelled". Two more faults came out of the same census: - **The class split was inverted.** Ports 1-4 were `poeOutClass: "PoE"` (802.3af, 48 V negotiated); the vendor describes them as `(4) 54V or 24V /1.4A, 4-Pair ... Passive PoE, Ports 1 to 4`. Ports 5-16 are the ones that actually negotiate: `(12) 802.3af/at or 24V/0.7A, 2-Pair`. So the four 75.6 W ports were filed as the weakest, and a 48 V 802.3af PD on port 1 was told it would work on hardware with no 802.3 detection at all - the UISP Router defect again (`test/uisp-r-poe.test.js`). - **Ports 1-2 were not PDs.** `(2) 54V / 1.5A, 4-Pair` passive PoE *input*, so `poeIn: 40` (the vendor's own `Max. power consumption 40W`) on eth0/eth1; cabling the switch to a PoE source now powers it instead of reporting an unpowered box. **Budget, derived and labelled as such.** Neither surface prints a PoE-budget row. Both print `(1) DC terminal block, 54V DC, 6A input` = 324 W and `Max. power consumption 40W (Excludes PoE output)`, so 324 - 40 = **284 W** leave the ports on the stated DC-input configuration; `poe.note` also records the PoE-input alternative (2 x 54V/1.5A = 162 W in, so 122 W out). Same arithmetic shape as `uisp-r`/`er-6p`. `l2-forward` added: the datasheet calls the EP-S16 "Layer-2 switch with some layer-3 capabilities" and lists Static/PBR routing, so `l3-forward` stays. Sources (read 2026-11-19): - https://dl.ui.com/datasheets/edgemax/EdgePoint_DS.pdf - 200, `application/pdf`, 6,571,516 B, 12 pp, "Models: EP-R6, EP-R8, EP-S16"; page 8 is the EP-S16 hardware table, page 4 the bottom-panel port map, page 5 the EP-R6 and page 6 the EP-R8 (three models, three adjacent tables). - https://techspecs.ui.com/uisp/wired/ep-s16 - 200, 577,761 B (not the 609,911 B front page), rendered spec table repeats `(12) PoE/PoE+ or 24V/0.7A`. Trap recorded in `research/ubiquiti-edgepoint-poe-census.md`: `EdgePoint_S16_DS.pdf`, `EP-S16.pdf`, `EdgePoint_S16.pdf` and `datasheets/edgemax/EP-S16_DS.pdf` all 302 into the 609,911 B tech-spec front page - a 200 check reads that as a document. Gates: syntax 463 OK; audit-port-types clean; audit-devices no hard defects; hardware invariants 26/0; census/description 3/0; audit-poe-budget silent PSEs 103 -> 102; run-tests 385 suites / 12,944 assertions, 0 failed; control-bytes clean; registry doc current (2,355 models). Siblings left for their own increment, census written up in the research note: `ep-r6` (five passive outputs, none modelled) and `ep-r8` (seven outputs, two modelled, both mislabelled; ports 7-8 are combo cages with no `comboPorts`).
devices(ubiquiti): the EdgePoint S16's sixteen PoE ports could light nothing (refs #130)
All checks were successful
CI / Classify the change (pull_request) Successful in 11s
PR rebase / Rebase onto main and arm auto-merge (pull_request) Successful in 10s
CI / Server smoke test (pull_request) Successful in 10s
CI / Docker build (pull_request) Successful in 22s
CI / Test (Node 22) (pull_request) Successful in 1m0s
CI / Test (Node 20) (pull_request) Successful in 1m1s
CI / Test (Node 18) (pull_request) Successful in 1m4s
CI / Arm auto-merge (pull_request) Successful in 3s
4e18df0581
forgejo-actions scheduled this pull request to auto merge when all checks succeed 2026-11-19 19:58:24 +00:00
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
jaylouisw/packetforge!2202
No description provided.