Boot time and idle resource usage get the benchmarking attention, but sleep and wake reliability is what determines whether a laptop works as a daily portable machine. A laptop that cannot reliably suspend and resume — or one that drains its battery overnight in sleep mode — is a machine you stop trusting. You start shutting it down instead of closing the lid, and the convenience advantage of a laptop disappears.
This page covers a one-week structured test of three refurbished business laptops running Windows 10, Windows 11, and Linux Mint. Each machine was put through three suspend/resume cycles per day for seven consecutive days, with overnight drain measurements and wake failure logging. The results reveal meaningful differences between operating systems and specific ACPI configuration issues worth knowing about before you deploy a refurbished laptop as a daily driver. For the broader performance picture, visit the Benchmarks hub. For boot and resource data, see our OS comparison benchmark.
Test Protocol and Hardware
Each laptop ran each operating system for seven days (21 days of testing per machine, 63 days total). The daily routine consisted of three suspend/resume cycles — morning, midday, and evening — each triggered by closing the lid. After 30 minutes of sleep, the lid was opened and we recorded whether the machine woke successfully, the time to reach an interactive desktop, and any error states (black screen, frozen display, kernel panic, or blue screen).
Overnight drain was measured by recording battery percentage at lid close (approximately 22:00) and at lid open (approximately 06:00) — an 8-hour sleep window. All machines started each overnight test at 100% charge to standardise the measurement. WiFi was connected and Wake-on-LAN was disabled in BIOS on all machines to establish a consistent baseline. Windows Fast Startup was disabled.
| Machine | CPU | RAM | Battery Health | Sleep Mode |
|---|---|---|---|---|
| Lenovo ThinkPad T440 | Core i5-4300U | 8 GB DDR3L | 72% (new replacement) | S3 (traditional) |
| HP EliteBook 840 G3 | Core i5-6300U | 8 GB DDR4 | 68% (original) | S3 (traditional) |
| Dell Latitude E5470 | Core i5-6440HQ | 8 GB DDR4 | 74% (aftermarket) | S3 / Modern Standby (Win 11) |
Battery health percentages were determined by powercfg /batteryreport on Windows and upower -i on Linux. The ThinkPad T440 had a replacement battery (common in refurb channels). The EliteBook retained its original 2016 battery. The Latitude had an aftermarket replacement of unknown manufacture date. These represent realistic refurb conditions, not laboratory ideal.
Wake Failure Rates
A wake failure is any event where opening the lid did not produce an interactive desktop within 30 seconds. This includes black screens requiring a hard power cycle, frozen displays, kernel panics, and blue screens. Each machine performed 21 suspend/resume cycles per OS (3 per day × 7 days).
| Machine | Windows 10 | Windows 11 | Linux Mint |
|---|---|---|---|
| ThinkPad T440 | 0 / 21 (0%) | 0 / 21 (0%) | 0 / 21 (0%) |
| EliteBook 840 G3 | 0 / 21 (0%) | 1 / 21 (4.8%) | 0 / 21 (0%) |
| Latitude E5470 | 1 / 21 (4.8%) | 3 / 21 (14.3%) | 0 / 21 (0%) |
ThinkPad T440 — Flawless across all three OSes
Zero failures in 63 total cycles. The T440's BIOS implements S3 sleep cleanly and Lenovo's ACPI tables are well-maintained. This machine is a reliable sleeper regardless of OS choice — a pattern we see consistently with ThinkPad T-series from 2013-2018.
EliteBook 840 G3 — One failure on Windows 11
The single Windows 11 failure was a black screen after overnight sleep that required a 10-second hard power cycle. Event Viewer pointed to a USB hub power state transition error. The same cycle pattern on Windows 10 and Linux produced no failures, suggesting a driver-level difference in how Windows 11 handles USB power states on this hardware generation.
Latitude E5470 — Problematic on Windows 11
Three failures in 21 cycles on Windows 11: two black screens and one blue screen (DRIVER_POWER_STATE_FAILURE). Windows 10 had one black screen on day 5. Linux had zero failures. The Latitude's ACPI implementation interacted poorly with Windows 11's Modern Standby attempts — the machine's firmware advertises S0 low-power idle support but does not implement it cleanly, causing the OS to enter an ambiguous power state.
Overnight Battery Drain
Battery drain during 8 hours of sleep (lid closed, WiFi connected, Wake-on-LAN disabled). Each value is the average of seven overnight measurements per OS. Battery health differences between machines make cross-machine comparison less meaningful than within-machine OS comparison.
| Machine | Windows 10 | Windows 11 | Linux Mint |
|---|---|---|---|
| ThinkPad T440 | 2% (S3) | 5% (Modern Standby) | 1% (S3) |
| EliteBook 840 G3 | 3% (S3) | 7% (Modern Standby) | 2% (S3) |
| Latitude E5470 | 2% (S3) | 8% (S0 idle attempt) | 1% (S3) |
Windows 11 Modern Standby overhead
Windows 11 drained 3-6 percentage points more per night than Windows 10 across all three machines. The cause is Modern Standby's design: rather than fully suspending to RAM (S3), the system enters a low-power idle (S0ix) that periodically wakes to check email, sync notifications, and maintain network connections. On machines with limited battery capacity, this adds up.
Linux consistent low drain
Linux Mint used traditional S3 suspend on all three machines, resulting in 1-2% overnight drain. The kernel's ACPI implementation defaults to S3 when available and does not attempt Modern Standby behaviour. For machines where you want to close the lid on Friday and open it Monday with battery remaining, Linux is the most reliable option.
The Latitude E5470 on Windows 11 showed the worst overnight drain at 8%. Over a weekend (approximately 60 hours), this machine would lose roughly 60% of its charge while sleeping. On Linux, the same machine would lose approximately 7.5%. For a refurbished laptop with a battery at 74% health, that difference means arriving Monday to a functional machine versus arriving to a dead one.
ACPI Issues and Configuration Fixes
Several of the issues we encountered have known fixes. If you are deploying a refurbished laptop and sleep reliability matters to you, these notes are worth checking.
Dell Latitude E5470/E5480 — Modern Standby conflict: These models advertise S0 low-power idle in their ACPI tables, but the implementation is incomplete. Windows 11 honours the advertisement and attempts Modern Standby, which fails intermittently. The fix on Windows 10 is to force S3 via a registry edit (CsEnabled = 0 underHKLM\SYSTEM\CurrentControlSet\Control\Power). On Windows 11, this registry key is less reliably honoured — a BIOS update from Dell (if available) is the more robust solution. On Linux, no fix is needed as the kernel defaults to S3.
HP EliteBook 840 G3 — USB hub power state: The single wake failure on Windows 11 traced to the Intel USB 3.0 hub losing its power state during suspend. Disabling "Allow the computer to turn off this device to save power" for the USB Root Hub in Device Manager resolved the issue in subsequent testing. This setting change had no measurable effect on battery drain.
ThinkPad T440 — No configuration required: The T440's BIOS correctly implements S3, does not advertise Modern Standby, and all three operating systems handled suspend/resume cleanly with default settings. This is the ideal outcome and the reason ThinkPads from this era remain popular in the refurbishment market.
powercfg /sleepstudy on Windows to see what is actually happening during sleep. On Linux, check journalctl -b -1 | grep -i suspend for the previous boot's suspend/resume log. Five minutes of diagnostics before deployment saves days of frustration. For a broader set of configuration tips, see our startup fix guide.What the Numbers Do and Do Not Prove
These results prove that sleep and wake reliability varies meaningfully by operating system on refurbished hardware. They prove that Linux's S3 suspend implementation is more consistently reliable and power-efficient than Windows 11's Modern Standby on older machines. They prove that model-specific ACPI issues exist and can be diagnosed and fixed with targeted configuration changes.
They do not prove that all ThinkPads sleep perfectly or all Latitudes have problems — our sample is three specific machines, and individual unit variation exists. They do not measure hibernate reliability (S4), which is a separate power state with different failure modes. They do not account for external USB devices, docking stations, or Bluetooth peripherals, all of which can introduce wake issues. And the battery drain figures are specific to these batteries at these health levels — a new battery would show lower absolute drain across all configurations.
The 7-day test period catches intermittent issues that a single overnight test would miss. The Latitude E5470's Windows 11 failures occurred on days 3, 5, and 7 — not on day 1. If you are evaluating a refurbished machine's sleep reliability, test for at least a full week before declaring it trustworthy.
Recommendations by Machine Type
| Machine Family | Sleep Reliability | Best OS for Sleep | Configuration Notes |
|---|---|---|---|
| ThinkPad T440-T480 | Excellent | Any — all three work cleanly | No special configuration needed |
| EliteBook 840 G2-G5 | Good to excellent | Linux or Windows 10; Windows 11 may need USB power fix | Disable USB hub power saving in Device Manager on Windows 11 |
| Latitude E5450-E5490 | Variable on Windows 11 | Linux or Windows 10 with S3 forced | Force S3 via registry on Windows 10; consider BIOS update on Windows 11 |
| Any refurb — general advice | Test for 7 days before relying on sleep | Linux for lowest drain; Windows 10 for best Windows compatibility | Run powercfg /sleepstudy (Windows) or check journalctl suspend logs (Linux) |
For the broader question of which refurbished machines are worth investing in, see Older Laptops Worth Saving. For the SSD upgrade that complements sleep reliability — because a machine that wakes fast from sleep also benefits from fast boot when sleep fails — see our SSD upgrade guide. For the complete boot time data across these machine generations, the 2026 Boot Time Index provides the reference. And for resource usage differences between the three OSes that affect how the machine behaves after waking, see our OS comparison benchmark.