Autoballast¶

Autoballast is the Slocum software feature that automatically tunes how much buoyancy drive the glider uses, so a pilot only has to specify the total drive they want and let the glider find the rest. It does two things:

Reduces the drive used on dives and climbs to the smallest amount that still flies the glider acceptably — saving energy and reducing speed.
Centers that drive so the dive and climb are balanced (a dive/climb speed ratio near 1:1) and the pump does not have to fully extend at the surface every cycle.

It is especially valuable on G3 gliders, whose high-displacement pumps move a lot more volume — running them at full ±1000 cc drive "significantly increases energy consumed" and can cause fast, steep, uncontrolled dives in shallow water. See Power Saving, which recommends autoballast as a core energy-conservation lever.

Source

Paraphrased from the masterdata autoballast/speed-control sensor block and the Slocum G3 Glider Operators Manual (the public doco/how-it-works/autoballast.txt), with field practice from the Teledyne Webb Research user forum and the UG2 community Slack. The concept — the glider "adjusts the center point of the drive to provide a dive/climb ratio of one" given a user-defined total volume — is described in TWR's UUST 2013 Slocum paper. Defaults and sensor names vary by firmware; confirm against your masterdata and simulate before flying.

What it does¶

Without autoballast you command a fixed buoyancy: full drive is ±1000 cc, but a glider flies fine on far less (often ~300 cc total, sometimes less). Picking that reduced drive by hand — and re-centering it as ballast, water density, and payload change — is fiddly. Autoballast does it continuously:

You give it one number: the total drive volume you want (the spread between the dive and climb buoyancy). There is no separate "climb" value — the single total is split around a center point the glider chooses.
The glider starts at full drive (c_dive_bpump −1000, c_climb_bpump +1000) and steps the drive down on each yo as it converges.
If vertical speed falls below your minimum, it adds drive back until the glider is fast enough again.
The converged, reduced drive is maintained at the surface too, so the pump isn't fully extended every surfacing (more surface stability, less energy).

What \"total drive\" means

If you ask for a total of 300 cc and the glider is perfectly ballasted, it will settle on roughly −150 cc on dives and +150 cc on climbs. The total is recorded as c_autoballast_volume in the data files; the live split is c_dive_bpump / c_climb_bpump.

Key sensors¶

Sensor	Role
`c_dive_bpump`	Live ballast (cc) used on the dive (negative). Watch in the surface dialog
`c_climb_bpump`	Live ballast (cc) used on the climb (positive)
`c_autoballast_volume`	The total drive you requested (read from the yo `b_arg` `d_bpump_value`)
`c_autoballast_state`	Current state of the autoballast routine (see below); writable to force a state
`f_min_ballast`	Floor on total drive (default 250 cc) — a total below this aborts the mission
`f_min_pump`	Minimum delta ballast used for a climb or dive (default 10 cc)
`u_autoballast_abort`	If autoballast fails to converge: `1` = abort, `0` = go to state 3 and keep flying on the last amounts
`u_autoballast_end_on_converge`	`1` = stop adjusting once converged; `0` = keep running autoballast
`c_speed_min` / `c_speed_max`	Slowest / fastest allowed depth rate for speed control (set per dive/climb)
`f_depth_rate_method`	Which filtered depth rate drives speed control (default `3` = running average, `m_depth_rate_avg_final`)
`c_time_ratio`	Climb/dive time ratio that must be maintained (default 1.1)
`c_wait_for_pitch` / `c_wait_for_ballast`	Let pitch and ballast settle after an inflection before enabling speed control
`u_diveclimb_msg_print`	Verbosity of autoballast messages (`-1` none, `0` errors, `2` basic, `99` all)

`c_autoballast_state` values¶

State	Meaning
`0`	Uninitialized — set this to reset/reinitialize autoballast
`1`	Initialized, still converging
`2`	Converged successfully
`3`	Converged unsuccessfully — a dive/climb amount hit the pump's max (`X_BALLAST_PUMPED_MAX` / `X_DE_OIL_VOL_MAX`)
`4`	Converged unsuccessfully — the climb−dive spread is below your requested total
`5`	Converged unsuccessfully — more complex; examine the `.mlg`

Setting it up¶

Autoballast is configured through behavior arguments, and it must be specified in two places:

The yo (or drift) behavior — the diving behavior, via the b_arg: d_bpump_value(X) total drive (and the speed-control args such as the minimum depth rate).
Every surface behavior — so the reduced drive is used on the surface too, not just while diving.

The TWR software release ships a standard autoballast mission (historically astock.mi, using yo14.ma and the surfac2x.ma surface files) — the easiest starting point is to base your mission on that template rather than wiring it up from scratch.

Don't set the total below f_min_ballast

If d_bpump_value is smaller than f_min_ballast (default 250 cc), the mission aborts. Review the masterdata description before lowering f_min_ballast itself.

Shallow water: trim the first dive by hand

Autoballast starts at full ±1000 cc drive and works down. In shallow water with a small target altitude, that first full-drive dive can drag the bottom before autoballast converges. Reduce c_dive_bpump before the initial dive to avoid bottom-sampling on the way to a converged solution.

Monitoring¶

Watch these in the surface dialog every piloting day:

c_autoballast_state — is it at 2 (converged successfully)? A 3/4/5 means it could not converge on the drive you asked for.
c_dive_bpump and c_climb_bpump — the live drive amounts. A healthy solution is roughly symmetric around zero; a lopsided pair means the glider is out of trim (ballast, density, or a stuck pump).
Dive and climb speeds vs. your c_speed_min — if the glider keeps bumping the minimum it will keep adding drive (and energy).

Raise u_diveclimb_msg_print to 2 (basic) if you want autoballast to narrate what it is doing.

Resetting autoballast in the field¶

Autoballast converges to the water it converged in. If you move into water of very different density (or it just isn't doing a good job), you can make it reinitialize without exiting the mission by forcing the state back to uninitialized from a surface dialog:

!set c_autoballast_state 0

It then re-converges from full drive over the next several yos.

Reinitializing costs a little energy and re-runs the motors

When you reset it, the buoyancy (and pitch) motors work to re-find the solution, and the glider briefly flies on full drive again. That's usually fine, but in shallow water remember it restarts at ±1000 cc — re-apply a reduced c_dive_bpump for the first dive if bottom clearance is tight.

A glider reset reverts to masterdata defaults

If the glider resets during a deployment (power cycle, exit reset, watchdog), c_dive_bpump / c_climb_bpump go back to the masterdata defaults of ±1000 cc and autoballast starts converging from scratch. After any reset in shallow or busy water, check the drive before the next dive.

If autoballast seems stuck or sluggish, operators nudge it by changing the total drive a little and back — e.g. raise the total to ~450 cc, then bring it back to 400 or 375 — and re-checking the dive/climb speeds. And be aware that autoballast has limits: if the surface density is far from where it converged, the glider can reach the end of its ability to compensate and fly poorly until you reset it (or adjust trim).

Simulation caveat¶

Simulate with full buoyancy, not a converged solution

Run simulations at full drive (d_bpump_value −1000, c_bpump_value +1000). If a simulated dive or climb stalls ("WE GOT STUCK not moving vertically"), confirm full drive is set in the yo dive/climb behaviors and check the simulated bottom (s_water_depth_avg) and altimeter-on depth (u_alt_min_depth). Autoballast's drive-reduction is meant to converge in real water, not the simulator.