Schrödinger’s Charge: The $14 Battery That Refuses To Admit It Exists

- by

This is the companion piece to my article about putting Fedora on an Acer Chromebook C771T. That one covers the firmware surgery; this one is about the battery that started all the trouble — and the one that ended it, sort of.

A laptop held hostage

My wife’s old work Chromebook had been sitting in a cupboard for years: an Acer C771T, the ruggedised education model with a touchscreen and a proper Core i5-6200U inside. When I finally charged it overnight and tried to switch it on, all I got was a flashing orange and blue light show on the left-hand side. The screen would try to come on for a split second, in tandem with the blinking, and then give up. Over and over.

Here’s the thing that made no sense at first: the machine would happily sit in ChromeOS Recovery Mode for minutes at a time — driving the screen, running the CPU, even writing a full recovery image to the eMMC. So the hardware was clearly fine. It just refused to complete a normal boot.

That pattern has a name: a power-rail brownout loop. The machine tries to power up, the voltage sags at the moment of peak draw, the embedded controller (EC) kills the attempt, waits, and tries again. Blink. Blink. Blink.

And the prime suspect was the known-dead battery. A failed lithium pack isn’t always electrically passive — a pack with a collapsed cell or a tripped battery-management controller can actively drag the power rail down. Recovery mode squeaked through because its early power profile is gentler; a full boot did not.

The five-minute diagnosis

The test costs nothing but a screwdriver session: disconnect the battery and boot on mains only. Most laptops of this era run perfectly well without a battery installed.

I took the bottom cover off (thirteen screws and a ring of snap clips — the scary-sounding kind, if you’ve ever opened a Samsung NC10 you know the drill), unplugged the battery connector at the top-right of the pack, plugged in the charger, and pressed power.

Instant, boring, normal boot. Case closed. The battery wasn’t just dead — it was actively holding the machine hostage. That also reframed the economics: this wasn’t a “shorter runtime” problem, it was an unbootable with battery connected problem. Replace or remove; no middle ground.

The search-term lesson that saved me $48

When I first looked for a replacement battery under “Acer C771T battery”, I found exactly one listing: $50 plus $12 shipping. More than the old machine deserved, frankly.

Then I opened the case and read the label on the pack itself: AP13J4K.

Search for that instead, and the picture changes completely. The AP13J4K (and its sibling AP13J3K) is the same battery Acer used across the enormously mass-produced Chromebook C720, C720P and C740 series. Replacements are everywhere — Amazon, Newegg, eBay — in the $20–45 range. I paid $32, minus some unused reward points: $14 delivered.

The general lesson, and it applies to every laptop: batteries belong to battery part numbers, not laptop models. The sticker inside the machine always beats the compatibility lists outside it. Manufacturers reuse packs across whole families of devices; resellers indexing by laptop model only capture a fraction of the market.

Waking a battery from deep sleep

New pack in, first surprise: pressing power did nothing at all. No lights, no boot.

This is normal, and worth knowing before you panic and start a return: aftermarket packs often ship in deep-sleep protection mode. The battery’s own management chip won’t respond until it has received some charge. The wake-up ritual:

  1. Connect the mains. The charge LED should come on (steady orange on this Acer).
  2. Boot the machine on mains power.
  3. Shut down, disconnect the mains, and try booting on battery alone.

Step three worked — the pack had woken up and was happily powering the machine. If it doesn’t for you, leave it charging (machine off) for an hour or two and try again.

The battery that works but doesn’t exist

And now the twist that gives this article its title. The battery charges. It runs the laptop for hours. And yet Fedora shows no battery at all — no tray icon, no percentage, nothing. As far as KDE is concerned, I’m permanently on mains power.

This isn’t a KDE problem or even a kernel problem. The evidence trail:

ls /sys/class/power_supply/
# AC  CROS_USBPD_CHARGER0
# ...and no BAT0

sudo dmesg | grep -i batt
# ACPI: battery Slot BAT0 absent

That ACPI line is the telling one: the firmware declares a battery slot exists, but the embedded controller answers “nothing in it” when asked. The kernel isn’t failing to find the battery — it is being explicitly told there isn’t one. I confirmed this across three different kernels (the two installed by Fedora plus an older one from a Clonezilla stick): all agree, because the answer is decided below the operating system entirely.

The architecture explains the split personality. The battery connector carries fat power pins and a pair of thin data lines speaking the SMBus “smart battery” protocol. Charging and discharging don’t need the data lines at all. My pack’s power side is flawless; its fuel gauge chip is either silent or speaking gibberish the EC won’t dignify with a BAT0 entry. Budget aftermarket packs sometimes get the cells right and the telemetry wrong.

Things I tried that ruled out everything else:

  • EC reset — the proper technique on a Chromebook is: hold Refresh, tap Power once, keep holding Refresh until the machine restarts itself. (Holding both buttons for ten seconds doesn’t work — the power button’s force-off behaviour fights the reset.) The instant restart confirms the reset took. Multiple confirmed resets: still no BAT0.
  • Reseating the battery connector — several times across several evenings, firm and straight. No change.

So: functional pack, mute gauge. The fault sits in a few cents’ worth of telemetry silicon inside the replacement battery.

The verdict: Schrödinger’s Charge, embraced

I could have emailed the vendor for a replacement, and gauge duds tend to be per-unit rather than per-design, so a second pack would probably report properly. I’ve decided not to. The pack does its actual job — it powers the machine for hours and charges reliably. The gauge is a luxury, and shipping another lithium battery across the planet to fix a cosmetic ailment felt like the wrong trade. Better for the planet, and honestly, better for the story.

So the machine runs on 2013 rules. The battery exists in a superposition of full and empty until the laptop collapses the wave function by switching off. There is no low-battery warning and no emergency hibernate; when the pack runs dry it’s a hard cut, like yanking the cord in the old days.

And here’s the field report after living with it: Btrfs genuinely does not care. I’ve now taken several unannounced hard power cuts mid-session, and the filesystem has come back clean every single time — no corruption, no lost files, no fsck drama. Copy-on-write means there is never half-written filesystem state to corrupt; the worst case is losing whatever wasn’t saved. That’s not a theoretical property from the documentation any more; that’s field-tested on my desk.

The wristwatch policy: Acer claimed up to 12 hours for the touchscreen C771T under ChromeOS. With a 45 Wh pack and a 15 W Skylake i5 sipping 5–7 W under KDE, I budget 6–8 hours of writing-and-terminal use and start glancing at the clock around hour five. Two mitigations worth setting up on any gauge-blind machine:

sudo dnf install powertop
sudo powertop --auto-tune

…flips the various device power-saving toggles (every saved watt is margin for error), and KDE’s screen dimming and suspend-on-idle still work on the output side even though the input side is blind — an abandoned laptop drains slowly rather than at full tilt.

Somehow the mystery makes the machine more charming, not less. It has no battery gauge, no F-key labels, no keyboard backlight. It simply works, silently, and tells you nothing. A little monk of a laptop — and I’ve decided to find not knowing how much juice is left very cutting edge.

Further Reading



If you enjoy my content, please consider supporting me on Ko-fi. In return you can browse this whole site without any pesky ads! More details here.

Leave a Comment!

This site uses Akismet to reduce spam. Learn how your comment data is processed.