Energy · off-grid making
That is the experiment: take a real making operation, the 3D printers and tools, into the garden, unplug it from the wall, and run it on stored solar. This page is the working log, and it leads with the real numbers, including the unflattering ones.
This is an early experiment, not a proven off-grid workshop. The hardware is real and charging today. But the first dawn measurement came in at ~21% of the panels' rating, and a sustained all-day print run on solar alone has not been demonstrated yet.
Every number below is either measured, tagged with the hour it was taken, or labelled est. If you came for a finished blueprint, this isn't one yet. It is the honest middle of one, published as it happens.
The chain is short. Folding solar panels feed a portable power station; the station stores the energy and serves it back as ordinary AC; the printers and tools plug into that instead of a wall. No roof install, no grid tie, no electrician. The whole rig carries out to the garden by hand and packs away the same way.
UB-200 folding panels. Each is rated 200 W, puts out DC at 20 V / 10 A max, runs transfer efficiency ≥ 23%, and weighs 4.6 kg. It folds to 54.5 × 58.3 cm and opens to 204 × 58.3 cm, carried out and packed away by hand. Several combine into the array. Bought from Overland GCC, a UAE overland-and-outdoor-gear retailer, named so the rig can be copied part for part.
The station is a Pecron E2400LFP. It puts out 2400 W of AC, holds 2048 Wh on its own, and expands to ~8 kWh with two 3000-series extension batteries. The panels feed its main solar port (GX16MF-5): 32–95 V MPPT, 15 A, 700 W max.
3D printers and workshop tools. Everything the operation runs is meant to come off this battery. That is the goal being tested, not an achievement being reported.
Solar arithmetic on paper is generous. The first measurement was taken at the least generous hour instead: 7:43 in the morning, sun still low, panels lying nearly flat on the ground. Two wirings, measured back to back into the station:
Divide through and both wirings land in the same place: ~42 W from each 200 W panel, about 21% of rating.
The wiring didn't change the harvest. The sun did. At dawn the array is irradiance-limited. Topology only starts to matter near solar noon, when current is high.
That per-panel figure held still while the topology changed, so no rearrangement of cables was going to fix it. The limit at 7:43 was the sun's angle on flat glass. Which is good news, in a way: the missing watts aren't lost to the wiring. They are still in the sky, waiting on the two cheapest levers there are.
Getting from a 21% dawn to a working day is not one heroic fix. It is three ordinary levers, in cost order:
Aim the array at midday sun and it should make est. 550–650 W, around four times the 167 W dawn reading. That covers a heated-bed print's draw of est. 300–500 W, with the 2048 Wh battery buffering the peaks. In the morning, and after the 3 PM shade line, it cannot. Bed-heavy work runs midday, or off the battery. Every number in this paragraph is a projection until it is measured.
Already in force on the station:
A compressor is the classic high-inrush load. Those get staggered, so a start-up surge doesn't trip the 2400 W ceiling.
The station itself stays out of direct sun. The panels want the light; the battery doesn't.
Everything else on this page is a lever. This is a wall. The E2400LFP's main solar port (GX16MF-5) accepts 32–95 V, and each UB-200 sits at roughly 24 V open-circuit. Three in series is about 72 V, safely under the 95 V ceiling. A fourth would push roughly 96 V into the port: past the ceiling, and enough to damage the unit.
This is a hard voltage limit, not a preference. Past three, panels can only combine in parallel, as the measured four-panel array (two parallel pairs) already does, inside the port's 15 A / 700 W input budget.
It listed cold- and low-bed printing as "a strategy, not yet a proven workflow", and it said the plan was to print cold or low where possible. That was wrong in the flattering direction. A cold bed here is not unproven. It was tried and it failed. The correction stays on the page rather than being quietly edited away.
The reasoning underneath it was sound and still is. On limited solar the 350 mm bed is the largest continuous draw in a print, far larger than the nozzle, so the cheapest watt is the one the bed never spends. Bed 0 was the obvious floor and became the default. It printed spaghetti twice on 2026-07-30, no first-layer adhesion on the bare plate. The mechanical press-weld this shop leans on for adhesion does not, by itself, hold PLA to a cold plate here.
The floor moved to 60 °C, the middle of this filament's rated 50 to 70, picked for reliability rather than for the last watt. What that buys is worth stating plainly, because it is not measured either: bed draw follows the lift over ambient, so on a 35 °C day 60 asks for roughly 25 degrees of lift against 120's 85. Around a third of the load, by arithmetic off the temperatures, with no meter reading behind it yet.
So the direction survives and the number on the page did not. A low bed, yes, and 120 was always the expensive habit. A cold bed, no. It was tested and it lost.
The standing rule on every senku page: the stage must be obvious at a glance, not after you have read carefully. So here it is, split honestly.
Nothing on this page is for sale. There is nothing to buy. It exists to be copied: one person, real hardware, real numbers in public, so the next person starts from 42 measured watts instead of 200 hoped-for ones.
The rest of the operation is published the same way: measurements first, stage always visible.
If you build the same rig and beat these numbers, that is the page working as intended.