Deep Pump¶

The deep buoyancy pump (350 m and 1000 m, ~900 cc) is a rotary displacement (hydraulic) design. Instead of moving seawater, it moves oil from an internal reservoir to an external bladder to change the vehicle's buoyancy. A rotary valve controls the flow of oil from the bladder back into the reservoir.

Vacuum requirement¶

The deep pump can retract at any depth (unlike the shallow pump, which is limited to its rated range). Even so, deep gliders are rarely inflected at the top for sustained operation below ~75 m.

The G3 HD pump & 3-way valve¶

The deep pump was redesigned for the G3 to be more robust and easier to maintain. It uses the device name hd_pump (must be in the installed-devices list in autoexec.mi) and adds a 3-way ball valve controlling oil flow from the external bladder back into the reservoir.

• In normal operation the valve opens at the surface to retract oil and begin the dive, then stays closed for the whole dive (including the bottom inflection, where oil is pushed out from the reservoir to become positively buoyant). It typically never has to open against high pressure.
• A drift-at-depth mission (becoming positively then negatively buoyant at depth) can push higher pressure on the valve. The 3-way valve offers a restricted and an unrestricted port: the unrestricted port allows much faster inflections, making deep gliders far more efficient in shallow water.

Storage & shipping¶

Bladder care: store clean and free of foreign matter using clean water and a clean cotton cloth. Do not use solvent-based cleaners (acetone, alcohol) — they dry out the bladder and shorten its life. No storage lubrication is required, but a thin layer of silicone oil or silicone spray (e.g. Molykote 316) is recommended.

How to test¶

From lab_mode:

1. wiggle on
2. report ++ m_de_oil_vol
3. Confirm the pump completes a full extension (m_de_oil_vol = +430 cc) and full retraction (-430 cc) without errors.
4. wiggle off

To confirm the pump is active: on every power-up the pump extends buoyancy to full displacement; report ++ m_de_oil_vol shows the position moving. With a properly ballasted glider, positive cc → climb, negative cc → dive.

Field troubleshooting¶

Small oil leaks in the forward section¶

A recurring G3 issue: a small amount of oil (a few mL / about a teaspoon) appears in the forward section after a mission, often smeared by the pitch battery, and frequently cannot be reproduced in the lab.

• TWR traced many cases to the ribbed oil tubing developing slow leaks over time and has moved to an upgraded tubing — have it replaced at service.
• A leaking pump often shows up as a reduced total available oil volume (one operator saw it down to about −311 cc). In reported cases up to ~9-month deployments it didn't degrade performance or compromise the O-rings, but you should not knowingly deploy a glider in that state.
• TWR strongly recommends letting them re-seal/re-service the oil system. If you do inspect it yourself (e.g. before sending it back anyway), don't over-torque hose clamps/fittings, and photo-document the teardown.
• A bench check some operators use: with the glider open but electrically connected, attach a vacuum pump to the open barb on the oil reservoir, clean up all old oil, lay shop towels around the pump, then manually command c_de_oil_vol out to +500 and use the vacuum to pull it back to −500, watching for where fresh oil appears. (Running the oil pump without vacuum is acceptable only in this limited way.)

"Oil flux too slow" / oil volume out of deadband¶

A dive that suddenly can't move oil — the pump appearing unable to shift past a fixed volume (e.g. stuck near −97 cc) — typically logs:

de_pump: oil flux too slow: -0.068382 (cc/sec), minimum flux limit: 0.100000
DRIVER_ODDITY:de_pump:410:de_pump_safety_check(): oil flux oddity
DRIVER_WARNING:de_pump:410:Error from pump safety check
DRIVER_ODDITY:de_pump:0:oil volume out of deadband
:OOD:de_pump_ctrl:OUT OF DEADBAND: M_su:-91.8 C_su:-260.0  delta:168.2 limit:20.0 ...

de_pump_safety_check() compares how fast oil is actually moving against the minimum flux limit (~0.1 cc/s); "out of deadband" means measured (M_su) and commanded (C_su) oil volume have diverged by more than the allowed deadband. Two common causes:

• A torn / lacerated oil bladder (most often the real cause). Water gets in and the oil–water mix pumps poorly, so the pump can't reach commanded volume. Operators seeing persistent slow-flux / wrong-direction messages have repeatedly recovered to find a laceration on the oil bladder — frequently from the nose recovery spool retaining clip whose edges were not sanded after replacement (a sharp edge nicks the bladder when it contacts in the same orientation dive after dive). High inflection counts raise the odds.
• A glitching oil potentiometer. A few bad position readings can mimic the same messages while the pump actually moves oil fine. Cross-check the m_de_oil_vol trace: a real problem shows commanded volume never reached over a whole dive; a sensor glitch shows isolated bad samples.

An oil flux in wrong direction message alongside the above strengthens the bladder diagnosis. Either way it is not fixable at sea — plan a recovery and have TWR inspect/replace the bladder and oil system.

Pump fault / stuck¶

• DRIVER_ODDITY: ... Buoyancy Pump is FAULTED! with MOVE ERROR Error reading position and the pump repeatedly taking itself out of service usually points to a failed position potentiometer (the board reading pump volume) — typically a TWR rebuild, not fixable at sea.
• A pump fault closely followed by a dropping leak-detect voltage can mean water intrusion shorting the pump electronics (e.g. a torn/perforated bladder weeping into the nose). Plan a recovery.
• Field stopgap when the pump is stuck at a fixed volume: setting f_ballast_pumped_safety_max to the stuck value can let you re-enable the pump (held in place) and limp toward recovery on the thruster.