Electronics¶

The main electronics board runs everything on a Seaglider: the motors, the VBD, navigation, comms, the pressure sensor, and every sensor port. Two generations are in service:

• Rev B — built around a TT8 processor with a Persistor CF2 and a CompactFlash card. Its firmware lineage is 66.x.
• Rev E — an ARM-based board with an SD card. Its firmware lineage is 67.x. Existing gliders can be upgraded to it.

Most day-to-day differences between the two (firmware files, menus, parameters, basestation versions) are documented by APL-UW IOP. For firmware versions, release notes and update procedures, use the IOP pages rather than this wiki:

• Seaglider firmware (seaglider.pub)
• Rev E software notes (APL-UW IOP, PDF)

This page covers the hardware side and the lessons from operating and servicing both boards.

Where the settings live¶

A glider's configuration is split between two places, and mixing them up is the most common mistake when a board is replaced:

Moving the card to a new board does not move the settings. They are only on the card if someone saved them there deliberately.

Back up before you change anything¶

• Save the parameters to a file (param/save from the main menu) before any firmware change and before any board work. Different firmware builds can change NVRAM contents, and parameter sets differ between versions. A saved file lets you restore everything with param/load.
• Keep a parameter dump with every glider's records. A self-test capture or a dive log also lists every parameter. It can be edited down to a load file if the board dies without a backup.
• Try new firmware without committing to it. On Rev B, APL-UW's advice for running different firmware temporarily (for example, to update the GPS) is to upload it under another name, such as glider.run, and start it from the top-level PicoDOS> prompt. Leave main.run untouched. A reboot returns to the original firmware. Restore the saved parameters afterwards.

Replacing a main board¶

When a board is swapped, work through the list rather than trusting that "everything carried over":

1. Load the parameters from the saved file, or from an edited log. Compare them with the last known-good parameter set, line by line.
2. Re-enter the utility settings: password, telnum, altnum.
3. Pressure sensor: check $PRESSURE_SLOPE and the gain, then redo the sea-level zero. See Pressure Sensor.
4. Internal pressure: zero it at atmospheric pressure before pulling vacuum.
5. Sensors: make sure the .cnf files are in the board's library and every sensor slot is configured. On Rev E, none of the serial sensors are built in. Every one needs its .cnf. See Sensor Configuration.
6. Compass: confirm the compass model and port. After some firmware updates the compass has to be reinstalled (param/config/compass).
7. Fuel gauges: check that CURRENTS and BATTERY are present and that consumption is being counted. See Batteries.
8. Run a full self-test and compare it, section by section, with the last good self-test from before the swap.

The same list applies when only the TT8 is replaced on a Rev B board. The parameters and the utility settings have to be copied across.

Rev B: the TT8 and its failure modes¶

The conductivity and temperature frequency signals from the CT sail go directly to timer (TPU) pins on the TT8. Damage to those I/O lines gives the classic board-level CT fault:

• the CT reports zero counts, or values that jump between sensible, wildly wrong and zero, with nothing moving;
• a new or freshly calibrated CT shows the same fault;
• the fault disappears when the main board is swapped with one from another glider.

In one case the manufacturer couldn't reproduce the fault on the bench, so the board went back to the owner as untrustworthy. In another, every other cause was ruled out: firmware, ribbon cable, CT boards and supply voltage. Damaged TT8 I/O lines have been seen before, sometimes with no known root cause. Before blaming the board, APL-UW suggests powering the C and T boards directly, disconnected from the tailboard, and checking for a clean frequency output from each.

TT8s are no longer available from the manufacturer. A failing Rev B board may therefore mean a Rev E upgrade rather than a repair. Weigh that cost early when planning a refurbishment of an older glider.

Rev E upgrades in practice¶

Upgrading a Rev B glider to Rev E is a board-and-wiring job plus a software migration. Field upgrades have run into:

• Kit differences. Upgrade kits have arrived without some adapter cables, including the altimeter adapter and the internal shorting-plug connector, and without some jumpers. Some cables have been too short to reach their new headers. Check the kit against the procedure before the glider is opened.
• Port assignments. The transponder/altimeter UART and power assignments were not documented in one upgrade procedure. The glider pinged nothing until the right port was found. Ask APL-UW for the port and mux map.
• Basestation. Rev E needs a newer basestation than most Rev B installations run. Check this before the first sea trial. See the IOP links above.
• Sensor integration changes. Every serial sensor needs a .cnf file. The factory Rev E builds don't use SciCon. Some .cnf strings that worked on Rev B can crash Rev E firmware. See Sensor Configuration.
• Pressure sensor. A Paine sensor does not work on a Rev E board. Fit a Kistler as part of the upgrade. See Pressure Sensor.

10/24 V to 15 V conversions¶

Converting a 10/24 V glider to the univolt 15 V architecture changes more than the battery. APL-UW's list of what has to change:

• Motors: boost pump, main pump, pitch and roll.
• Solenoid coil (the Skinner valve).
• Phone DC-DC converter and RF relay. These may be fine, but need checking.
• Optional: a resistor change so that the external-power relay picks up reliably, and 15 V-specific transponder magnetics. APL-UW has done conversions without either and saw no real difference.

The VBD assembly itself is built for one architecture. See VBD and Batteries.

Bench power and brown-outs¶

• Give the bench supply enough current. A supply limited to 1.5 A let the battery bus sag below 4 V while the main pump ran. The processor browned out while writing to the card and left corrupt, undeletable files that filled it. At about 3 A the same test only sagged to about 11 V. Low supply voltage also makes the pumps look slow, which can send you hunting for a VBD fault that isn't there.
• Run at least one self-test and a simulated dive on the glider's own batteries before launch. External power hides problems and creates others.
• The powered comms cable has over-voltage protection. It stops the glider powering up if the 10 V and 24 V leads are swapped. On some gliders this circuit made start-up on external power slow and erratic, sometimes dropping into the low-level TOM8 prompt, while start-up on the batteries was normal. The manufacturer said this was harmless. The real rule is: never reverse the supply leads.

Rev B CompactFlash card¶

• Watch the used space. Corrupt or runaway files can fill the card. Once the card is full or the file structure is damaged, files may not delete or transfer with the normal tools. A glider in that state may keep flying as long as main.run is intact, but it stops writing new data.
• Reformatting means opening the glider. Before you do, gather the files you'll need to put back:
• the firmware (main.run);
• capvec;
• every .cnf file;
• the tcm2mat compass file;
• BATTERY and CURRENTS;
• bathymetry maps;
• SciCon files such as scicon.ins, if used.

Then reinstall the logger and sensor configuration. The sensor configuration lines in an old self-test capture show how it should look. After one reformat, the science sensors returned empty files until the SciCon files were put back. - Restore the BATTERY file too. Without it the fuel gauge starts from zero and the consumption to date is lost.

Reboots and $RELAUNCH¶

$RELAUNCH decides what the glider does after an unexpected reboot. The default, 0, puts it into recovery. With 1 it carries on diving. The manufacturer's view is that a random reboot at sea deserves investigation before the glider dives again, hence the conservative default. The value changes by itself: it becomes 4 while the glider is in recovery, and goes back to the original setting when it leaves. So seeing 0 → 4 → 0 in capture files, for example while trimming at the start of a mission, is normal.

Handling the boards¶

• Disconnect the battery from the main board first, before you lift, probe or unplug anything. Removing the shorting plug or external power is not enough on its own.
• Keep the insulating film under the main board in place. In one case the film shifted during testing and the underside of the board touched the chassis. It gave off a puff of smoke and never booted again. That happened even with external power and the shorting cable removed.
• Check the fasteners on the small board under the main board. One was found completely loose, free to come off and rattle around the forward section during a mission.
• Connectors:
• The 0.1-inch MTA connectors are reliable when made correctly. They are insulation-displacement types, so don't strip the wire first.
• Never tin the ends of crimped wires. The solder makes a hard spot where the strands work-harden and break.
• Poorly mated connectors on older gliders have caused intermittent pitch-sensor dropouts as cable tension changed.
• A bad Molex connection caused GPS failures on another group of gliders.
• If a fault reproduces in the field but not on the bench (at depth, in the cold, at certain attitudes), a bench poke-and-prod test proves little. Log the symptom against depth, temperature, pitch and roll from the dive data before replacing parts.

See also¶

• Pressure Sensor — the analog front end on the main board and what to redo after a board swap.
• Sensor Configuration — serdev/logdev .cnf files, the sensor library, and Rev E specifics.
• Comms, GPS & Basestation — modem, antenna and GPS faults.
• Batteries — pack voltage measurement shares the board's relay and ADC.