devices(ubiquiti): the EdgePoint R6 and R8 could not power the radio they exist for #2206
No reviewers
Labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
jaylouisw/packetforge!2206
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "catalogue/1119-2022"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Refs #130 (does not close — the Dream Wall Pro itself is still without a vendor surface; see the thread).
What this increment found
Two members of the EdgePoint family in
devices/ubiquiti/could not power the airMAX radio they exist for. Same datasheet, two different failures:devices/ubiquiti/ep-r6.jsonpoe-pseclaimed, zero PSE ports, nopoeblockpoeOut: 16.8passive (eth0–eth4),poe.budgetW 65,poeIn: 7oneth0,10Mrungdevices/ubiquiti/ep-r8.jsonpoeOut: true, poeOutClass: "PoE"on eth0 and eth1 only —eth0is the vendor's PoE input port, and "PoE" is 802.3af, which this chassis has nowhere — whileeth2–eth7(five real PSEs) were dry; nopoeblock;poe-pdmissingeth0poeIn: 40and no output;eth1–eth2poeOut: 75.6passive (54V x 1.4A);eth3–eth7poeOut: 16.8passive (24V x 0.7A);poe.budgetW 284;poe-pddeclaredWith no
poeblock,engine/simulator.jscomputePoe()opened withif (!dev.poeBudget) continue;and skipped both boxes — not "unknown", skipped.show power inlinehad nothing to report and no radio on the tower could be powered by either router.Budgets are DERIVED, and say so in
poe.notewith the arithmetic and the PoE-input alternative:24V DC, 3A= 72 W less7W (Excludes PoE output)= 65 W (PoE-input path: 33.6 - 7 = 26.6 W)54V DC, 6A= 324 W less40W (Excludes PoE output)= 284 W (PoE-input path: 162 - 40 = 122 W)No 802.3af/at/bt capability is claimed on either: every output row the vendor prints is passive, so a 48 V 802.3 PD will not negotiate.
Evidence
https://dl.ui.com/datasheets/edgemax/EdgePoint_DS.pdf— HTTP 200,application/pdf, 6,571,516 B, 12 pp, text layer; page 6 = EP-R6 table, page 7 = EP-R8 table, page 9 = EP-S16 table.https://techspecs.ui.com/uisp/wired/ep-r6— HTTP 200, 577,753 B, rendered spec table (not the 609,911 B front page).https://techspecs.ui.com/uisp/wired/ep-r8— HTTP 200, 578,035 B, same rows.Vendor rows used verbatim: R6
Passive PoE Output (5) 24V / 0.7A, 2-Pair (+4, 5; -7, 8) Passive PoE, eth0 to eth4; R8Passive PoE Output (2) 54V or 24V / 1.4A, 4-Pair ... eth1 to eth2,(5) 24V / 0.7A, 2-Pair ... eth3 to eth7,Passive PoE Input (2) 54V / 1.5A ... eth0 and PoE InwithPoE In is DC Only, No Data.Deliberately not modelled
The R8 datasheet says two of its eight copper positions are RJ45/SFP combination ports but never says which two, and the QSG keys (
qig/ep-r8.pdf,guides/edgemax/EdgePoint_QSG.pdf) all answer 200 with the same 609,911 B front page. The pairing is left undeclared rather than guessed — ten cableable links into eight positions is exactly the bug a guessedcomboWithcauses. Recorded inresearch/ubiquiti-edgepoint-poe-census.md.Tests
test/edgepoint-poe.test.js(new, 31 assertions) pins both rows, both budgets, both PoE-input directions, and that a NBE-5AC-Gen2 cabled toeth4on the R6 andeth3on the R8 is powered.test/poe-inference.test.js: EP-R6 leaves thePD_ONLYlist — it was put there on the strength of its own description, not the vendor table — and is pinned by its own assertion instead, the way ER-10X was.Also corrects the EP-S16 row page citations (hardware table is page 9, not page 8), found while re-reading the same PDF.
Gates
syntax-check 466 files OK; audit-port-types clean; audit-devices no hard defects; hardware-invariants 26/0; spec-vs-description 3/0; audit-poe-budget silent PSEs 102 -> 101; run-tests 387 suites / 13,002 assertions, 0 failed; control-bytes clean; doc-citations clean; registry doc current (2,355 models, regenerated, no diff).