Offline Vault Power Consumption Estimates
A first look at the Offline Vault’s estimated power consumption, from the display and Raspberry Pi to standby draw and the clock battery. We’ll test the components and share the results as we work toward a more efficient build.
Power draw and energy consumption are key factors in the development of The Offline Vault. If this device is going to be useful when power and internet access are unavailable, we need to understand not only how long it can run, but also how well it preserves its charge while it’s sitting unused.
Before we begin testing, I want to document the components we’re considering and what we expect them to consume. These are starting estimates, not measurements of our finished system. We’ll test the actual draw over time, use the results to make the build as efficient as practical, and share what we learn.
The components we’re starting with
Our first prototype is a wall-powered workbench build using a Raspberry Pi 5, a 15.6-inch non-touch display, solid-state storage, two wired controllers and a USB hub. The core parts are now ordered. It will let us test the proof of concept before we commit to a final enclosure or battery system. We’re holding off on the media remote for now; it is not required to get this first build running.
Ordering a part is not the same as approving it for the finished Vault. Our approved product list starts empty. I’ll add parts only after testing and approving them, then reconcile that list with the living bill of materials as the design develops.
Raspberry Pi 5, 4GB, with the official Active Cooler. Both are ordered, along with one official 27W USB-C power supply. Raspberry Pi lists approximately 4 watts for an active bare board. Actual draw will change depending on what the Vault is doing and what is connected. Video playback, games, storage activity and cooling will all need to be measured. The 27-watt power supply is rated for what it can provide; it does not mean the Pi constantly uses 27 watts.
Raspberry Pi Monitor, 15.6-inch, 1080p, non-touch. We’ve ordered this in place of the Waveshare touchscreen. We don’t need touch for the first proof of concept, and power consumption is one of the things we want to evaluate. Raspberry Pi’s documentation lists 4.5–6.5 watts during use and less than 0.1 watt at idle. Those are manufacturer figures, not our measurements. At that operating range, the display alone would use roughly 135–195 watt-hours over 30 hours, before external power-supply losses. The old 7.6-watt touchscreen estimate no longer describes our selected screen. These figures are not a controlled comparison at equal brightness and volume, so we’ll test before claiming an efficiency improvement. The ordered bundle includes a separate 5.1V/3A USB-C supply and a 3-foot micro-HDMI-to-HDMI cable.
TEAMGROUP MP33 1TB NVMe SSD and Pimoroni NVMe Base. Both are ordered. The MP33 replaces the unavailable Raspberry Pi-branded 1TB SSD. We haven’t established a reliable power figure for this drive in our Pi setup. We’ll measure it while the Vault is idle, playing media and copying files, and check what it draws when it isn’t actively being used.
Raspberry Pi USB 3 hub and two Logitech F310 wired controllers. These are ordered; the F310 pair replaces the previously planned 8BitDo controllers. The gamepads are USB-powered and don’t need separate batteries, but they still add to the main system’s load. We’ll measure the hub and controllers individually and together, and test cold startup, reconnecting, button mappings and two-player games on the Pi 5. The hub’s power rating is a supply limit, not its continuous consumption. A second 27W supply for the hub remains optional, depending on what testing shows.
Setup and recovery accessories. A SanDisk 64GB Extreme microSD card and UGREEN card reader are ordered, along with a mini wireless setup keyboard. The card and reader are for installation and recovery; the keyboard is a workbench tool rather than a commitment for the finished product. We’ll distinguish accessories left connected during normal use from tools used only during setup.
Cooling and audio. The Pi’s fan will use power when it runs, and the monitor’s speakers will add more depending on volume. We’ll measure both rather than assume they’re negligible. We’ll also avoid counting speaker consumption twice when using a whole-monitor measurement.
What is still open. We’ll check the included cables and mounting hardware when the orders arrive, then add insulating workbench supports, standoffs, screws or extra cables only where needed. The media remote is deferred. The main battery, charging electronics, shutdown controller and final enclosure remain later design work, not purchased or validated Phase One components. I already have a Kill A Watt P4400 for the initial wall-power measurements.
These estimates don’t yet give us a reliable total. The complete system will also have power-conversion losses, and the final battery and charging system have not been validated. Each watt used continuously adds up to 30 watt-hours over a 30-hour session, so even small improvements can matter.
What does “off” really mean?
A device can appear to be off and still draw power. The Raspberry Pi 5 can consume around 1 to 1.4 watts in its default shutdown state. Raspberry Pi documents a much lower-power shutdown configuration, but the Pi is only one part of the system. The display, power converters, USB accessories, charging electronics and battery-management system could also consume energy while the Vault is stored.
For a device intended to be stored for a long time without an external power source, that matters. We’re considering a power controller that would let one button request a clean shutdown, wait for the software to finish, and then disconnect power from the Pi and other nonessential components. The same button would restore power when it’s time to use the Vault again.
We’ll also evaluate whether a separate storage disconnect is necessary. Either way, the battery’s protection circuitry must remain intact. Even with the rest of the device disconnected, the battery-management system can use some power and the battery itself can lose charge over time. We won’t know the real shelf life until we choose and test the complete power system.
The clock battery
The Raspberry Pi 5 includes a real-time clock. We’ve ordered its small rechargeable backup battery so the clock can retain the date and time when the main power is disconnected.
But a specialized battery should not become a single point of failure for an offline device. If the clock battery dies or is unavailable, the Vault should still run and its offline content should remain usable. After a complete loss of power, the user may need to set the date and time manually. We’ll test that scenario as part of the build.
Storage life versus runtime
The battery system currently being considered uses four 3.2V, 30Ah LiFePO4 cells in series: 12.8V and approximately 384Wh nominal.
For an illustrative operating estimate, assume 90% of nominal capacity is usable and 85% of that energy reaches the loads after conversion losses. That leaves about 294Wh at the loads. At an average 15W, that would be about 19.6 hours; at 20W, about 14.7 hours; at 30W, about 9.8 hours. Actual workload, volume, screen brightness, temperature, age, cutoffs, and converter efficiency will change the result. At those same assumptions, 30 hours at 15W would require about 588Wh nominal, before aging and other reserves.
Storage is a different calculation. With the main loads disconnected, the battery-management electronics and battery itself can still lose energy. As a planning scenario only, if the pack starts at 70% charge and loses 2–3% of its remaining charge per month, it would retain roughly 49–55% charge after twelve months. That assumes the stated loss rate covers all relevant losses; it is not a verified specification for the selected pack. Temperature, BMS/charger/gauge draw, cell self-discharge and age can change it substantially. We have not measured whole-system storage draw or validated a 12-month ready-to-use claim.
Under that hypothetical twelve-month scenario, approximately 187–211Wh nominal would remain. At a 15W load and an assumed 85% conversion efficiency, that represents roughly 10.6–12 hours of operation, before any additional operating reserve. The assumed 2–3% monthly loss is the total loss in this illustration, so BMS consumption is not deducted a second time. These figures are not measurements or safe-storage guidance.
The separate Bioenno BLF-1230A battery considered for testing has a manufacturer manual that advises not leaving it unattended for more than six months. It must not be presented as validated for a year of unattended storage. The 384Wh figure above is a candidate capacity used for illustration, not a finalized battery selection. Manufacturer manual: https://cdn.shopify.com/s/files/1/0219/0154/files/BLF-1230A_User_Manual_rev_04_23_2019.pdf?209=
We will measure the complete pack and system under real operating and storage conditions before treating these estimates as performance targets.
What comes next
Once the workbench is assembled, we’ll use my existing Kill A Watt P4400 to measure the complete setup at the wall during startup, idle, video playback, gaming, file transfers and shutdown. Those readings include the plug-in power supplies’ losses; they are not direct battery-runtime measurements. We’ll compare screen brightness and volume levels. Later, with the protected battery system in place and suitable low-current instrumentation, we’ll measure the true storage-off draw over time.
Those results will help us decide where to optimize, whether to change components, and how much usable battery capacity we actually need. I’ll document the results here as we go.
Sources: Raspberry Pi Monitor documentation · Raspberry Pi hardware and power documentation · Raspberry Pi RTC battery
