build steps
In the water, weighed and balanced: neutral with the drop weight attached, positive without it, level fore and aft. The buoyancy budget stops being a spreadsheet.
The buoyancy budget is mostly [EST] and [WEIGH] rows. This is where they become numbers. Do it BEFORE the vehicle goes in the water — once it is assembled and floating, individual masses are unrecoverable without taking it apart again.
[EST] or [WEIGH] marker with [MEAS] and the date. Keep the estimate alongside it in brackets.WARNING: A big discrepancy here is good news, not bad. Finding out on the bench that the vehicle is 3 kg heavier than budgeted costs an afternoon of re-planning foam. Finding out at the shoreline costs a launch, and finding out at depth can cost the vehicle.
WARNING: Weigh the drop weight separately and clearly. It is the one mass that is *supposed* to leave, and every buoyancy calculation has a with-it and a without-it answer.
Every [EST] and [WEIGH] row in BUOYANCY-BUDGET.md carries a real measured number, and the pre-foam total is written down.
The foam is the entire buoyancy solution and it is the one material whose properties were measured rather than assumed — 0.297 g/cm³, giving +0.703 kg of lift per liter, considerably lighter than the 0.45 originally assumed.
WARNING: Epoxy coating adds mass and no displacement. The BOM plans to coat the foam for depth margin — budget roughly 5% off the net lift, about 5.4 kg per block rather than 5.53. Small, and it goes the wrong way.
WARNING: Do not cut all the foam at once. Cut what the arithmetic says, float the vehicle, and cut the rest against what the water tells you. The water is the authority, not the spreadsheet.
Foam is cut, coated, measured, and its actual mass and displacement are recorded.
The first time the assembled vehicle floats is the moment the spreadsheet meets reality. Do it somewhere you can reach in and grab it.
WARNING: Fresh water only, and know your water temperature. Lake Michigan is fresh; salt water is ~3% denser, which is a *large* offset on a neutrally-trimmed vehicle. BUOYANCY-BUDGET.md sets water type with trim weights per deployment, not with the VBS.
WARNING: Have the vehicle on a line, and have a second person. A sub that turns out to be 5 kg negative goes to the bottom of the test tank at a speed that surprises people.
The as-built buoyancy offset is measured in kilograms, and the delta against the budget is written down.
This step produces one of the five numbers the VBS decision rests on (see TEST-PLAN.md Stage 1). Not just "is it neutral" — how close can you actually get it, and how repeatably?
WARNING: This number sets how hard the thrusters have to work for the entire mission. ±100 g is easy work for them; ±500 g is continuous work from the surface down. Per BUOYANCY-BUDGET.md, the VBS covers the dynamic range only and *cannot fix a gross trim error* — without a VBS that is even more true.
WARNING: Trapped air is not buoyancy, it is a lie. Work every pocket, shroud and free-flooding volume until bubbles stop. Trapped air compresses at depth, so a vehicle trimmed with air in it gets heavier as it descends — exactly when you can least afford it.
She hangs neutral with the weight attached, the residual offset is recorded in grams, and the figure is repeatable across a reassembly.
BG is how far the centre of gravity sits below the centre of buoyancy, and it is the pendulum stiffness of the whole vehicle. It appears nowhere in the project's documents, and the cruise-efficiency argument for pitch-based depth control rests entirely on it (see VBS.md).
The easy way — roll period: 1. Floating neutral, roll her a few degrees and let go. 2. Time ten full oscillations and divide. A short period means a stiff vehicle (large BG); a long, lazy period means a soft one.
The direct way — known offset mass: 3. Hang a known mass a known distance off the centreline. 4. Measure the steady heel angle it produces. 5. BG follows from the righting moment: W × BG × sin(θ) = m × g × d.
At cruise, the fore/aft vertical thrusters can pitch the hull so the HULL does the lifting — far cheaper than heaving directly. But holding that pitch means fighting the righting moment:
| BG | force to hold 2° | vs 4 N direct heave | |---|---|---| | 10 mm | 0.08 N | 47× cheaper | | 20 mm | 0.17 N | 24× cheaper | | 40 mm | 0.34 N | 12× cheaper |
WARNING: A stiff vehicle is expensive to hold off-level; a soft one is cheap but wanders. Both are workable, but you cannot tune the depth controller sensibly without knowing which one you built.
BG is measured, written into BUOYANCY-BUDGET.md, and the roll period is recorded alongside it.
Fore/aft balance matters as much as total buoyancy. BUOYANCY-BUDGET.md: *"A nose-heavy vehicle will not hold depth, and nose-up surfacing (which lifts the LTE antenna out of the wave shadow) is a trim decision."*
WARNING: Submerged trim and surfaced trim are different problems and can fight each other. Surfaced attitude is set by where the waterline cuts the hull; submerged attitude is set by BG and mass distribution. Get submerged level first — that is the one that affects every hour of the mission.
WARNING: The drop weight's position affects trim, and it is designed to leave. Check the attitude with it and without it. A vehicle that surfaces nose-down after dropping weight cannot get its antennas out of the water — which defeats the entire recovery chain.
Level submerged, slightly nose-up surfaced, and the same in both cases with the drop weight released. Weight positions photographed and recorded.
This is the failsafe of record. FAILSAFES.md builds the entire recovery ladder on the drop weight making her positively buoyant *even if thrusters, VBS and both computers are dead*. Until it is measured in water, that is an assumption.
WARNING: A marginal ascent is a failed test. She may be positive in a warm test tank and negative in 4 °C lake water, which is denser only slightly — but she will also be carrying a fouled hull, a full sample basket and possibly a flooded free-flooding volume. Reserve buoyancy needs real margin, not a couple of hundred grams.
WARNING: Test with the sample basket loaded. A 200 g rock is near the limit of a 0.1 L VBS swing, and it is also mass the drop weight has to overcome.
With the drop weight released she rises on her own with measured reserve buoyancy, surfaces nose-up, and does so with a loaded sample basket.
The point of this milestone is not a trimmed vehicle — it is a buoyancy budget that stops being a spreadsheet. If the numbers stay on a notepad, the next person to touch this (including you, in six months) is back to estimates.
[EST] and [WEIGH] with [MEAS] and the date.WARNING: Record what you did NOT measure, too. An unmarked gap reads as a measured zero to whoever reads it next.
BUOYANCY-BUDGET.md contains no unmeasured rows for anything that exists, BG is written down, and TEST-PLAN.md Stage 1 carries the VBS-decision numbers.