VBD (Variable Buoyancy Device)¶

The VBD is the Seaglider's buoyancy engine and the reason it moves at all: a hydraulic system in the aft endcap that moves low-viscosity oil between an internal reservoir inside the pressure hull and an external bladder outside the hull (but inside the fairing). Pumping oil out increases the glider's displaced volume without changing its mass — it gets lighter than the water and climbs; bleeding oil back in shrinks it and it sinks. It is also the single largest energy consumer on the vehicle, so most piloting-for-endurance decisions are ultimately VBD decisions.

What's in the system¶

A few component details worth knowing¶

• Main pump — a commercial rotary piston pump (Hydro Leduc base) modified for the Seaglider (flats machined on the shaft for the motor coupling). Its shaft is sealed by a magnetically coupled shaft seal ("mag seal"); oil weeping from the top bearing means that seal needs replacing — see Oil leaks and the mag seal.
• Main-pump return line — carries the pump-body overflow from the low-pressure feed back to the reservoir. Air seen in it is cavitation from the pump pistons; it collects in the internal reservoir and comes out at the bleed screw (reservoir horizontal, bleed port tilted to the highest point).
• Skinner (bleed) valve — driven by a magnetically latching relay: one pulse opens it, another closes it, and it is not energized in between. The firmware has no direct feedback on valve state, so an electrical glitch that momentarily opens it goes unnoticed except through the potentiometers.
• Uncommanded-bleed detection — the firmware compares the average of the two linear-pot readings with the value after the last move; if it has drifted by more than a threshold it raises a VBD error, and the pilot's error settings decide whether the glider surfaces. One pot failing to an end stop (0 or 4095) drags the average away and triggers this falsely.
• Pot-to-pot difference — a steady gap of 200–400 counts between the two pots is normal on an older glider. The piston assembly develops a "memory" and tilts slightly as it moves, and since the full ~3400-count stroke is only about 7 cm of travel, a few hundred counts is just a few millimetres. The vendor's analysis template flags anything over 150, so judge it against the same glider's earlier tests rather than the flag.
• Pump retries in Rev E firmware (67.xx) — the Rev E code doesn't count pitch or roll retries (they always show 0 in the $ERRORS line), and it no longer watches the main pump. It only monitors the boost pump when it runs alone (shallower than $D_BOOST): if the boost rate drops below ~1 A/D count per second it logs a VBD retry and switches the main pump on to help. The 66.xx code for Rev B boards tracked retries differently.
• Oil volume — about 850 cc of oil is "movable"; the full fill is larger (typically ~1350 cc) because the tubes, pumps and valves also hold oil. The as-built amount is recorded on the glider's trim spreadsheet (Trim tab, aft-endcap assembly section).

Two plumbing layouts¶

Seagliders in the field carry two generations of hydraulic plumbing, and several pump-testing and repair decisions depend on which one you have:

Which one do I have? Operators often can't tell by eye. Two ways that work:

• The trim spreadsheet. If the enhanced reservoir section of the aft-endcap assembly is filled in with component masses, the glider has the enhanced, parallel-plumbed engine.
• Photos of the endcap plumbing. Send clear photos to the manufacturer. They can identify the layout from them. In one case two older gliders that looked parallel-plumbed to their operators turned out to be series-plumbed.

Positions are A/D counts — and the names are backwards¶

Like pitch and roll, the VBD position is read on a 0–4095 A/D count scale, with hardware limits found at assembly and tighter software limits inside them. The conversion is $VBD_CNV = −0.2453 cc per count (same for SG and SGX) — note the negative sign:

$C_VBD — the neutral position — is set for the densest water of the mission (the deepest part of the dive), and is one of the first things trimmed at sea: see Trim & Flight Model.

The VBD budget¶

A Seaglider has roughly 800–860 cc of usable volume change, and a mission spends it three ways:

The rule of thumb for what that remainder buys: about 70 cc per σ_T unit of density change for SGX (~50 cc for SG), so the 400 cc above absorbs ≈5.5 σ_T of stratification (SGX). If the mission's density range exceeds that, something has to give — shallower dives or less thrust at apogee. Driven flat out (−350 cc thrust, ~18 cm/s, full-range pumping every dive) a Seaglider can stem ~40 cm/s of depth-averaged current, but burns energy at roughly ten times the rate of a gentle mission where the VBD stays within half its range.

Energy: why the VBD dominates¶

Pumping at depth means pushing oil against full sea pressure — the pump accounts for about half the total energy budget of a Seaglider. The control scheme is built around this (no bleeding on descent, pumping only on the climb where the oil must be moved anyway), and the pump itself is optimized for efficiency near 1000 m — at shallow-water pressures it moves only ~2 cc/s, which is part of why shallow missions are hard on Seagliders. The most expensive single act is the big surface-maneuver pump to $SM_CC; reducing $SM_CC (where safe) and avoiding unnecessary deep pumping ($W_ADJ_DBAND, pitch-over-VBD) are the standard savings — see Trim & Flight Model.

The glider's cumulative $POWER summary in the capture file shows how lopsided this is. On one Seaglider at the end of a long mission, pumping at apogee accounted for about 84 % of all the charge drawn from the 24 V bus. Surface pumping was about 1 %, and Iridium (init, connect and transfer together) was about 13 %.

Let the boost pump do the shallow work (Enhanced Buoyancy engines only). The low-power boost pump can pump alone shallower than $D_BOOST. The vendor's advice for a glider running short of battery was:

• Set $D_BOOST to about 10 m at minimum, so that only the boost pump does the surface maneuver.
• Where the mission allows, set $D_BOOST to 120 m and limit dives to 100 m. The glider then pumps on the boost pump alone.
• Keep $SM_CC and $MAX_BUOY as small as is safe.

On engines without the upgrade the service-bulletin limits in the warning above apply instead: $T_BOOST,0 and $D_BOOST ≤ 5 m.

Lab: bleeding air out of the VBD¶

Air in the hydraulics makes VBD moves spongy and position readings untrustworthy. The system is bled in three stages, pushing air along the path lines → bladder → reservoir → out. The procedure below uses the glider's own electronics (a jog box — a manufacturer's tool that drives the motors and Skinner valve directly — makes it easier, but is optional). Setup: connect main/boost motors and both potentiometer leads to the tailboard, tailboard to mainboard (bench alongside the endcap is fine), bench supply at 10 V and 24 V, comms cable on port A, power on, wand the glider on, and go to the hw/vbd menu.

1. Lines — orient the endcap so the reservoir's "T" fitting (the pump supply line) is at the bottom: air in the reservoir floats away from the supply so the pump doesn't re-ingest it. Use the ad option to pump to a value ~200 counts lower than currently reported (pumping pushes any line air into the bladder). Confirm the lines look clear.
2. Bladder — reorient so the Skinner valve (silver cylinder with the blue coil pack) is at the top — it is the bleed port back to the reservoir. Shake/rattle the bladder gently to walk bubbles up to it. Use open to open the Skinner valve while squeezing the bladder by hand, driving air and oil back to the reservoir, then close while still applying pressure. Repeat a few times.
3. Reservoir — reorient with the Phillips bleed screw on top. Back the screw out slowly, a couple of turns only — the linear-potentiometer springs keep the reservoir pressurized, and trapped air is forced out. When oil (not air) starts to emerge, re-seat the screw fully.

Repeat any stage as needed; one full pass removes virtually all the air.

Lab: cycling the pumps ("cycletron")¶

Exercising the VBD through full-range cycles on the bench — main pump alone, main + boost, and boost alone — verifies pump health and produces a logged dataset (currents, rates) to compare against previous services. The community procedure runs from the glider's hw/vbd/pump menu with a terminal log capturing everything:

1. Start a terminal log (e.g. sgXXX_cycle_main_only_YYYYMMDD), wand on, and enter hw/vbd/pump.
2. Answer the prompts: specify A/D counts (not pressure); accept the software min; 5 s rest; accept the software max; 5 s rest; then the D_BOOST / "use boost to prime main" questions per the variant you are testing (both N for main-only; prime Y for main+boost).
3. Sample interval 1 s, display readings Y, 10 cycles, 900 s pump time. Close the log when done.
4. For boost-only, first set $D_BOOST,25 (saved to NVRAM) and re-zero the pressure sensor at sea level (hw/pressure/sealevel) so the glider believes it is deep enough to run the boost pump, then run the same cycle dialog answering Y to D_BOOST. Restore the operational $D_BOOST afterwards (see the boost-pump parameter warning above).

The logged HVBD lines can be bookmarked (e.g. in Notepad++), extracted, and pasted into a spreadsheet to trend pump rate and current draw over time.

Reading the output¶

Newer firmware (Rev E) prints one line per sample under the header cycle sec vbd0 vbd1 avg mA P psi rate effic motors volts:

The manufacturer's analysis spreadsheet has one tab each for pump and bleed in the main, main + boost and boost variants. It flags lines where the linpot difference is over 150 counts, the rate is under 1.2 counts/s or the current is over 500 mA, and summarises average, minimum and maximum current, rate and linpot difference for each variant. For scale, here is an older series-plumbed glider on the bench under vacuum, judged healthy by the manufacturer after a bladder change:

The main pump is also clearly louder than the boost pump, so you can hear which one is running. If a main-only or main + boost run shows the motor current staying around 40–65 mA and the A/D reading not moving on the pump lines, only the boost pump is running. After reassembly, check that the main-pump motor connector (J4) is seated. In one case that was the whole fault.

What matters is how current and rate compare with the same glider's earlier tests (or a healthy sister glider). A newly built vehicle that showed roughly double the usual current at about half the usual rate was judged unfit to deliver: it would still pump, but endurance would suffer badly and it pointed to a VBD fault. It went back for a rebuild.

Oil leaks and the mag seal¶

Oil appearing inside the aft endcap or the hull almost always comes from the main pump's mag seal:

• The mag seal "burps" — momentarily unseats and lets a little oil past — when the piston is over-pumped into the reservoir cylinder head at the extremes of travel, or when the main pump runs without a boost feed. It often reseats itself once oil starts flowing back. After a burp the software limits are commonly pulled in a little from the extremes (one example after a ~5 cc leak: $VBD_MIN 340 → 390, $VBD_MAX 3600 → 3550).
• A glider that shows oil in the endcap but then passes a pressure-chamber test (assembled, to ~1000 psi, no uncommanded bleeds) is probably fine — but a momentary loss of seal at the wrong moment can end a mission. Tell the owner. One operator pulled such a glider from a long deployment rather than accept the risk (and noted that a known, disclosed risk may also affect an insurance claim).
• A mag seal that keeps leaking should be replaced. The durable fix is APL-UW's mechanical shaft seal upgrade. The manufacturer also offered its own replacement seal for the legacy mag seal. Either swap is depot work, not a field job.
• The mag seal is finicky and only works if it is set perfectly. Running the main pump with air in the system strains it and can force the seal faces apart. That is why the manufacturer's advice is to change it whenever the VBD has been opened. If a leak is found and the pump hasn't been run since, the seal may still carry on working. When in doubt, replace it.
• Unexplained oil loss can also show up only in some tests — in one case main-only cycles were clean, but a boost-only run lost ~90 cc. Log where and when oil appears (check after every stage) before deciding what to replace.

A failing bleed path can take electronics with it: on one glider, parts of the main board around the VBD drive failed while the oil-return valve was misbehaving. After replacing the board components, the pump was cycled for hours under full vacuum to confirm the fix.

Before deployment¶

• Check the software limits in the glider, not just on paper. On one glider a temporary $VBD_MIN of 1400 (set during earlier trimming) was never reset to its normal ~600. The glider sat low at the surface with much less pumping range than intended, and comms suffered.
• If the glider has been opened, check that the internal cables (one case: the pitch cable) are routed clear of the mass-shifter gears before closing up.
• Treat a pot reading pegged at 0 or 4095 as a no-go. One glider logged 4095 on a linear pot during its first dive, recovered on the next, and was then lost after its second dive. The cause was never found, but in hindsight the pegged reading was the warning to act on.
• Anything mounted in the aft fairing must clear the fully inflated bladder. For example, a PAM recorder housing sits with its connector end toward the bladder, placed just aft of the bladder at full inflation so the two never touch. Route its cables clear of the bladder too. Don't overtighten the plastic cradle screws. A housing that works loose can also upset the glider's flight (in one case, uncommanded pitch changes on the climb).

See also¶

• Trim & Flight Model — $C_VBD trimming, volmax estimation in the tank, and the FMS vbdbias estimates that track volume through a mission.
• Dive Cycle & Control Files — where pumps and bleeds happen in the dive, and the parameters that bound them.