docs(power): split power into its own guide and cover the DC routes - #275
Merged
Merged
Conversation
10 tasks
HuggeK
force-pushed
the
docs/power-guide
branch
17 times, most recently
from
September 22, 2026 22:23
53f0dfd to
078297a
Compare
HuggeK
marked this pull request as ready for review
September 22, 2026 22:23
Power decisions were scattered across two tables and three blockquotes in the parts list. That hid the thing that matters most: picking a power route decides about six purchases at once, and one of those routes can destroy the build if it is wired backwards. All of it moves to a new docs/get-started/power.md built around three routes, a battery variant of one of them, and two add-ons, and parts.md keeps the accessories every build needs whichever route you pick. The routes: 1. The official 27 W supply. Its cable is captive to the brick, and there is no panel connector to unplug at either, because a USB-C pass-through carrying a genuine 5 A is not a practical part to buy. So it stays tethered to the case. The page says plainly why it is still the right default: nothing to wire, nothing extra to buy, and no unkeyed barrel plug near the boards. 2. A Geekworm X1202/X1206 UPS HAT. Leads with the trap, which is that a UPS HAT is not one purchase: it arrives with neither cells nor any way to get power into it, so the DC adapter, the panel jack and the leads that join them have to be bought at the same time. 3. A Geekworm Pi5-5V5APD, for a detachable lead without batteries. Flags that it needs more than 30 W in, so the official 27 W brick will not drive it, and that the case has no mount for it. A USB-C PD charger or power bank through a VFLEX or a PD trigger board, and Power over Ethernet through an 802.3bt splitter, are not routes and are no longer numbered as though they were. Each only changes what plugs into the barrel jack in the back of the case, with the UPS or the converter behind that jack unchanged, so both now sit under a "what plugs into the barrel jack" table next to the wall adapter they stand in for. PoE suits this case in particular, because the shell already has an Ethernet opening beside the DC hole. There is also a 2b, for anyone who can live with a supply that does not detach. The barrel jack stays the default for a UPS build, since being able to unplug at the case is most of the reason to fit a panel jack, and none of that route is removed. But both UPS boards take 5 V 5 A on their own USB-C socket, and 5 V 5 A is exactly what the route 1 supply exists to make, so the official captive brick can feed the UPS directly, and that is the only way to have batteries with no barrel plug anywhere in the build. Geekworm list the X1202's USB-C input as "Compatible with Raspberry Pi USB-C Power Supply", and on the X1206 V1.1 they "strongly recommend" it over the DC jack, so on that revision it is the recommended input rather than a workaround. It drops the adapter, the panel jack and the lead into the board, which is about $25 of parts and every step where a polarity can be got wrong. The price is that nothing detaches at the case and charging is slower: 25 W in barely covers a Pi that can pull 25.5 W at peak, so the pack tops up when the rig is idle rather than under load. The page says which of the two to build rather than leaving it as a preference. The shell does not stand in the way of that. There is no USB-C panel connector in it, which is not an oversight — a pass-through rated for a genuine 5 A is not a practical part, and the repo's own USB-C rear shells are EOL for "USB-C spec compatibility" — but the plug passes through a rear opening: the official supply's plug measures 12.0 mm across at its widest, which clears the 12.5 mm DC hole, and the 16.0 x 14.0 mm Ethernet cut-out has room to spare. Measured on the supply itself rather than derived from the USB-C envelope, so the page states it rather than working through it. Adds the barrel-jack polarity section JedS asked for on open-flight#273. A 5.5 x 2.1 mm plug has no keying, so a centre-negative supply mates perfectly with a centre-positive jack and reverses the rail into the Pi, the UPS and both radars at once. The X1202 and X1206 are both centre pin positive; since Geekworm's wiki does not state it either way, the page cites the exchange where Geekworm confirmed it rather than sending readers to a page that will not answer them. Adds wire sizing for the DC run, the only part of this build carrying real current: 20 AWG or heavier, 22 AWG as the floor on a sub-100 mm run, and not the 26 AWG lead whose connector would otherwise fit. The two ways into the UPS are reordered so the shorter one leads. A screw-terminal barrel plug takes the panel jack's stripped leads directly and goes into the board's own jack, and that is the whole run: no Wago splices, no XH lead, nothing crimped or soldered. Wagos only ever existed to join two leads, so they are needed for the XH header option, or for a barrel plug that arrived pre-wired, and the parts table says that rather than listing them as unconditional. On lithium safety, Geekworm's guidance is about charging rather than storage: "When charging the Battery Pack, please place the battery in a fireproof container." There is no Geekworm guidance about storing cells and no mention of a metal or airtight container, so the page says what they actually say, and notes that the battery-handling site their template links to no longer resolves. Corrections found while moving things: - X1206 V1.1 takes USB-C 5 V, or 5-6 V on its DC input, and wide voltage will burn it out. The old text implied the 12 V route applied to every X1206. - The v3 case wants a 12 mm power button, not the 16 mm one the cost lines named. - Geekworm states flat-top for the X1202's 18650s but only "unprotected" for the X1206's 21700s. - A long-reach hex key is no longer called required. The case screws sit deep in the shell, so a driver needs about 90 mm of reach, but the head type follows whichever screws you buy. The cost summary is rebuilt as three staged tables with running totals, so you can see what it costs to stop at each point instead of reading one number off a flat list that mixed required parts, optional extras and mutually exclusive alternatives in a single column. Doing the arithmetic turned up three errors: - The Core line read $355 while its own rows sum to $425. Checked rather than assumed: Adafruit 5812 is a Pi 5 4 GB at $130, so the rows were right and the summary understated every build by $70. The README carried the same gap. - The micro-USB cable was counted twice. There is one cable and the layout decides which board it plugs into: the OPS243 when it is the only radar, the IWR6843LEVM once Layout A moves the OPS243 to the GPIO UART. - The angle radar's GATE-to-BCM17 jumper was counted twice, being one wire from the pack the Sound Trigger line already buys. The battery line also costed only the HAT, the cells and the button, when a UPS cannot be powered in a closed case without a DC adapter, a panel jack and the leads between them. Those are now in it, itemised, and expressed as a net add because a UPS replaces the 27 W supply rather than adding to it. Base build, OPS243 in its case 534 + angle radar 684 + inclinometer and camera 729 Everything, X1202 battery power 819 Everything, X1206 instead 831 The enclosure sits in the base rather than under optional, because you need one whichever radars you fit and the same printed set covers both builds: the radar front carries mounts for the OPS243 and the IWR6843, so an OPS-only build prints the same parts and leaves the IWR mounts empty. Nothing measures repeatably until the boards are held in a fixed arrangement. The $52 assumes you print it yourself. build-order.md follows, moving the enclosure out of Optional extras into a step of its own, and pointing that step at the openflight-enclosure repository rather than the IARC case: choose a variant, required hardware, print, assemble. The IARC page is kept as a note for existing builds only. Also in the same files: - The firmware hex for the internal-trigger route can be requested on OmniPreSense's Discord or through the contact page on their site, not only by email. Sandy at OmniPreSense offered both. - Explains why the internal trigger is worth wanting beyond the parts it saves: a microphone cannot distinguish your strike from the rest of the room, so a noisy range is what the sound trigger handles worst. Marked as the expectation rather than a measured result. - Drops the claim that this build needs no soldering, which contradicted the R17 resistor and the heat-set inserts. - Trims the J-Link EDU Mini row to what is in the box. - Shortens five cable rows that re-derived their lengths in the row text when the Cable Lengths section already covers it. - Names the rear DC jack and power button in Enclosure Hardware, since they fall between that list and the enclosure repository's. README's hardware table gains the enclosure, so its headline figure is a unit you can pick up rather than a pile of boards, and both glance tables say the 27 W supply's cable is captive, since that is the tradeoff the route page spends a section on. The documentation list links the new page, and it is in the site nav and the Get Started index. Co-authored-by: HuggeK <48095810+HuggeK@users.noreply.github.com> Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
HuggeK
force-pushed
the
docs/power-guide
branch
from
September 23, 2026 06:57
078297a to
9acdef1
Compare
Contributor
Both my question and Geekworm response is on their wiki (in the FAQ of https://wiki.geekworm.com/X1206), so the wiki can be treated as source. |
jewbetcha
approved these changes
Sep 23, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Power decisions were scattered across two tables and three blockquotes inside
parts.md. This moves them into a new Powering OpenFlight page and leavesparts.mdholding the accessories every build needs whichever route you pick.Companion PR for the rear hole sizes this page links to: open-flight/openflight-enclosure#13.
Pi5-5V5APD— wide DC in, USB-C outA USB-C PD charger and an 802.3bt PoE splitter are add-ons rather than routes: each only changes what plugs into the barrel jack, leaving the board behind it unchanged.
Why was this required?
mainare wrong, two of them expensive. The 12 V route is presented as applying to every X1206 — on a V1.1 that burns the board ("use the correct power supply, or the board may burn out"). The cost lines name a 16 mm power button that does not fit the v3 case's Ø12.5 mm holes. And the Core line read $355 against rows summing to $425, so every headline build total was understated by $70.Automated tests
None — docs only, nothing under
src/,ui/ortests/, so there is no code path a test could reach. Verified by script instead: every internal link and heading anchor resolves, all 19 tables have consistent column counts, every external URL returns 200, and the cost tables re-add to their totals — which is how the $70 error surfaced.Manual (human) testing
Template:UPS_Safety_Warning— establishing that it states neither barrel polarity nor any cell-storage guidance, so the page sources both elsewhere or scopes them honestly.No hardware was powered for this PR. No UPS wired, no jack metered, no splitter or PD trigger bought. Every electrical claim is vendor-sourced and cited inline; the polarity section is instructions to follow, not something proven here.
🤖 Generated with Claude Code