title: Compass / Attitude Sensor description: The Slocum compass and attitude sensor (True North TNT Revolution / legacy TCM3, the attitude_rev device): what it measures, magnetic variation, when and how to calibrate it on shore with RevolutionTest, the in-situ at-sea calibration, the post-cal compass check, and field troubleshooting for heading errors and interference.

Compass / Attitude Sensor¶

A precision navigation compass and attitude sensor continuously measures the glider's heading, pitch, and roll. These are the inputs the glider uses to dead-reckon its position underwater between GPS fixes, so a compass that is off by a few degrees turns directly into navigation error — the glider thinks it is flying one way while the water carries it another.

On modern G2/G3/G3S gliders the device is a True North Technologies "TNT" Revolution attitude module, addressed in the flight code as the attitude_rev device. Very early gliders (roughly serial numbers below ~200, G1-era) instead carried a PNI TCM3 compass with its own tcm3 cal routine. This page focuses on the TNT/attitude_rev device that the vast majority of the fleet uses.

Source

Paraphrased from the Slocum Glider Operators Manual (Rev. 1, "Attitude Sensor") and the Slocum G3 Glider Maintenance Manual (Rev. A, "Calibrating the True North Compass"), masterdata.txt, the Teledyne Webb Research user forum, and the UG2 community Slack (including in-situ calibration user notes contributed by NorGliders). For the calibration software itself, refer to True North Technologies' Revolution Compass User's Guide. This is a condensed field reference — always defer to the official Teledyne documentation and the readme.txt shipped with the calibration tool for your specific glider.

When to calibrate¶

The compass should read true within a few degrees in the configuration you will deploy in. Recalibrate (or at least re-check) whenever that magnetic environment changes or the readings drift:

New or swapped battery packs. Battery packs — especially aft alkaline packs, which sit closest to the compass — carry residual magnetism that shifts the calibration. Teledyne has shipped alkaline sets that needed degaussing before they would calibrate. Pitch and nose packs sit far enough away to matter much less.
After adding, moving, or removing hardware near the aft section (sensors, hydrophones, recovery hardware, anything ferrous).
Heading errors in the water — the glider not tracking to waypoints, or the classic spiral signature (see Symptoms of a bad calibration).
Moving to high magnetic latitudes (Arctic, high-latitude deployments), where the horizontal field is weak and a cal done elsewhere may not hold (see High-latitude operations).
As a routine pre-deployment check — Teledyne suggests a quick heading check before every deployment and recalibration as required.

Calibrate, then verify

A calibration is only as good as the compass check that follows it. Don't treat "the software applied new coefficients" as success — always spin the glider against a hand compass afterward and confirm the error is within tolerance for your battery chemistry. See The compass check.

What it measures¶

The attitude sensor publishes the glider's orientation. The headline sensors:

Sensor	Units	Meaning
`m_heading`	rad	Measured heading. Magnetic, not true (see below).
`m_heading_deg`	deg	`m_heading` converted to degrees.
`m_pitch`	rad	Measured pitch; > 0 is nose up.
`m_pitch_deg`	deg	`m_pitch` in degrees.
`m_roll`	rad	Measured roll; > 0 is port wing up.
`m_roll_deg`	deg	`m_roll` in degrees.
`m_hdg_error`	rad	`m_heading − c_heading` (how far off the commanded heading the glider is).
`m_gps_mag_var`	rad	Magnetic variation (declination) read from the GPS at the surface.
`m_attitude_rev_mode`	nodim	Command state of the `attitude_rev` device.
`m_attitude_rev_measure_state`	nodim	What the device is doing with its output (streaming vs. sample mode).

Heading is magnetic, not true

m_heading is the magnetic heading. When you fly to a commanded heading with c_heading, that is also magnetic. The glider only converts to true heading when it works in LMC (local mission coordinates) for waypoint navigation, using the magnetic variation.

Magnetic variation (declination)¶

The glider corrects between magnetic and true heading using m_gps_mag_var:

mag_heading = true_heading + mag_var
mag_var > 0  ==>  variation is West (like on Cape Cod)

m_gps_mag_var is read from the GPS receiver at the surface and published automatically — for normal operations you do not set it yourself. A typical value might be, e.g., −16.7° without you ever touching it.
The on-board Garmin GPS carries a magnetic-variation lookup table in its own firmware. That table is static: it is only as current as the GPS firmware. If the GPS firmware is years out of date, its declination values can be stale — a consideration for high-precision work and for regions where declination is changing fast.
There is also a default s_mag_var in masterdata used before a GPS fix is available.

Arctic declination changes fast

Near the magnetic poles, magnetic variation changes quickly with position and over time, and published models (and the GPS lookup table) have had to be revised for the Arctic. At high latitude, declination error compounds with the weak-horizontal-field problem below.

Standard (shore) calibration with RevolutionTest¶

The standard calibration is done on shore with the glider physically rotated through 3-D space while the TNT RevolutionTest Windows application records the magnetic field and computes new hard-iron/soft-iron coefficients, which it stores on the compass itself.

Get away from magnetic interference

The compass is very susceptible to ferrous material and magnetic fields from electronics, vehicles, rebar, and buildings. Calibrate outdoors, away from any of these, suspended from a non-ferrous structure. A cal done in a steel-framed building or next to a metal cart can be worse than no cal at all. (Calibration can be done from the cart in a pinch — see the note on coverage below — but a clean hang is the gold standard.)

Equipment¶

The glider, hung so it can spin, pitch, and roll freely (a bridle under a non-ferrous hoist; some teams manage from the cart with extra strapping).
A laptop running TNT RevolutionTest (RevolutionTest.msi).
A Freewave modem programmed to your glider, plus power supply — most teams do the cal over Freewave so no cable tethers the hanging glider. (See the Freewave page.)
A serial terminal program (to issue the talk command and then release the port).

Procedure¶

Step-by-step: shore calibration

Set up Freewave comms to the glider (CD light green) and escape the startup into the console.
Drop to the device shell and start streaming attitude data:
G3 / G3S: talk attitude
G2: exit pico then talk att

This streams the raw attitude data and, in TWR code from roughly the last several years, closes the air-pump solenoid valve to put the glider in the same magnetic state it is in while diving/climbing (see the solenoid note below). 3. Close / disconnect your serial terminal so the COM port is free — RevolutionTest needs that same port. 4. Start RevolutionTest. When prompted, choose the wireless/serial connection, the same COM port, and baud 9600 (the attitude_rev comms rate). Optionally open the compass display (top-left icon) just to confirm you are connected. 5. (Two-step / vertical reference — recommended.) In Tools → Capture Vertical Reference, hold the glider level and steady, click Start, then Use New. This captures a vertical reference outside the worst of the vehicle's influence so the subsequent 2-D rotation can solve all three hard-iron components. 6. Tools → Calibrate Magnetics. Set the number of samples to the maximum (3000) and click Start/Begin. After a ~20 s countdown the software begins acquiring samples. 7. Exercise the glider through every reachable orientation in 3-D until it reaches ~3000 samples — flat spins through N-E-S-W, then repeated with the nose pitched up, pitched down, rolled to port, rolled to starboard, and combinations of those. The bar graphs show coverage; the goal is to fill in all the bins. The roll/z bins are the hardest to fill and need lots of rolling (the far ends fill only at extreme roll — even inverting the glider). Have someone call out the sample count so the people turning the glider can pace it. It stops on its own at ~2999. 8. The screen switches to a Results page. The New column's "Mag Total 3 sigma" should be lower than the Old — aim for < 1% for a solid cal. Click Use New / Apply to write the coefficients to the compass. 9. Close RevolutionTest, reopen your terminal, and confirm the glider is back at the console.

Take Iridium out of service during the cal

An Iridium call mid-calibration can inject a magnetic transient and corrupt the data. Many teams use - iridium (and avoid Freewave/Argos keying where possible) while running the cal. This is the same RF/electrical coupling family that causes false digifin leak detects — see Fin / Digifin.

Why the solenoid matters

The air-pump solenoid is a strong, switchable magnetic source right in the glider. During diving and climbing it is closed, and talk att/talk attitude closes it so the cal is done in the flight state. If you later do a compass check from GliderDOS (report ++ m_heading) instead of from the talk state, the solenoid is open unless you inflate the air bladder and wait for the air pump to shut off — otherwise the check can read up to ~12° off versus the actual flight calibration.

The compass check¶

After applying new coefficients, verify before you trust them:

Open the compass display in RevolutionTest (or report ++ m_heading from GliderLAB/GliderDOS — but if you do it that way, inflate the air bladder first so the solenoid is closed, as above).
Using a hand compass, sight the direction the glider is actually pointing and compare it to what the glider reports.
Take readings around the full circle — every 30° (or 45° if short on time) — and repeat at roughly 26° nose-up (climb) and 26° nose-down (dive) pitch, since the error varies with pitch.
Record the glider-vs-hand-compass error at each point (a spreadsheet helps). The errors typically trace out a rough sine wave around the circle, often with one bearing (frequently west) worse than the rest.

Target error by battery chemistry (community rules of thumb):

Battery chemistry	"Good" heading error
Lithium-ion (rechargeable)	< ~5°
Lithium primary	~3–5°
Alkaline	< ~8–10°

With the solenoid correctly closed (air bladder inflated), well-covered calibrations have reached ~2–3° peak-to-peak precision and ~1.5° accuracy across all headings on all chemistries. If the solenoid is not locked, 5–10° peak-to-peak is more typical, most of it an uncorrected hard-iron offset.

In-situ (at-sea) calibration¶

If the glider is already deployed — or headed somewhere remote where a clean shore cal isn't possible — there is an in-situ calibration that runs entirely from a mission, collecting cal data while the glider yos in the water. It uses the factory attcal.mi mission (and a community-tuned attcal2.mi variant) plus the GliderCal.exe desktop tool.

In-situ is a fallback, not a replacement

Teledyne's guidance: an in-situ cal is not a substitute for a proper shore calibration. It is genuinely valuable when you have no other option — notably high-latitude deployments where it has rescued badly behaving compasses — but results are mixed, and some gliders still fly with a significant heading offset afterward. Treat it as "better than nothing."

How it works, in outline:

The mission puts the attitude_rev device into CCD sample mode via the compass_cal behavior and collects field data over a long surfacing window (the community attcal2.mi raises when_secs to ~7200 s, sets a longer overtime abort, deepens the yo to ~50 m, and disables low-power settings so they don't trigger a different abort). It needs at least 300 lines of data for GliderCal to compute an accurate offset.
It generates a *.cal file on the glider. You send that file off (send *.cal), open it in GliderCal.exe, and Calculate Results; as with the shore cal, the new Mag Total 3 sigma should be much lower than the old.
You apply the result with compass_cal set_offsets X Y Z (X/Y/Z offsets — include negative signs!), and confirm with compass_cal get_offsets. compass_cal ? lists the available sub-commands.

The attitude_rev-stays-offline / abort caveat

A known quirk: once the compass_cal behavior has put attitude_rev into CCD sample mode, it stays there after the surfacing condition is met and simply appears offline — which trips a device abort. The mission collects a valid *.cal file regardless, but the glider needs to be exit reset to bring attitude_rev back online (the use commands generally won't recover it). Plan for that abort: set the mission's overtime and num_samples so you know roughly when to be on console, turn off the GliderTerminal xml script, and have a sensors.mi ready to reload your flight settings (c_dive_bpump, c_climb_bpump, u_alt_min_depth, low-power sensors, …) after the reset. This same flow has been confirmed to work on G3S as well, despite the abort.

Why it matters at high latitude — a field result

One Arctic G2 (software 8.4, ~69° N) was flying north on dives and south on climbs while trying to make headway south. An in-situ cal moved the offsets from (−1676, 1720, 3207) to (1023, 1950, 117) and dropped Mag Total 3 sigma from 55.9% to 0.67%, restoring normal flight. Operators in Baffin Bay have had less luck — at the most extreme latitudes even the in-situ cal can struggle.

Avoiding and managing magnetic interference¶

Degauss suspect battery packs. A magnetized aft pack can make a clean calibration impossible and can throw the compass off in the water. Teams degauss aft alkaline packs (CRT/tape degaussers, bulk degaussers) — pass the pack slowly over the degausser, starting and ending well clear of it. You can gauge magnetization before and after by running a hand/hiking compass along the pack and watching the needle deflect.
Soften the attitude warning levels if the compass keeps going out of service. The factory attitude_rev warning thresholds were found to be overly stringent; Glider Service Bulletin 011 reduces them so the compass drops out of service (and aborts the mission) far less often near magnetic fields. Make sure that bulletin's settings have been applied — an unusually high rate of attitude_rev errors often means they haven't.
The compass goes out of service near strong fields by design — steel buildings, electronics, high-iron ground. That is expected on the bench; it is a problem when it persists in the water.

attitude_rev errors and aborts

When the compass cannot produce a heading, the glider raises attitude_rev device errors and can ultimately abort with MS_ABORT_NO_HEADING_MEASUREMENT (compass busted). There is no good in-water fix for a genuinely bad compass — the durable answer is to recalibrate after recovery. See the Aborts page for handling device errors at sea.

High-latitude operations¶

Near the magnetic poles the horizontal component of Earth's field gets weak while the vertical component dominates, so the compass is increasingly "looking at" the vertical field instead of the horizontal field it needs to find heading. The practical threshold the community has converged on: as the horizontal field strength drops below ~10,000 nT, compass performance degrades. Symptoms include the glider reading wildly different headings on dive vs. climb and spiraling. Mitigations:

Calibrate in situ at the deployment latitude rather than relying on a cal done further south.
Keep the GPS firmware current so its declination table isn't stale.
Expect larger residual errors and design the mission to tolerate them.

Symptoms of a bad calibration¶

The hallmark of a compass problem in flight is the glider spiraling — m_heading sweeping through full circles, often on the climb but sometimes the dive, while c_fin/m_fin slam from hard port to hard starboard as the autopilot keeps flipping which way it thinks it must turn. Commanded heading looks fine; the glider just can't hold it.

Not every spiral is the compass

Spiraling and dive/climb roll asymmetry have several causes that mimic a bad compass — confirm the compass before tearing into a recovery:

Biofouling on the rudder / hull seams disrupting flow (often shows up 1–2 months in, near shore/warm water) → see Fin / Digifin.
A lost or pivoting wing / broken wing rail — imbalanced roll on dive/climb, sometimes without an obvious mass change.
Center of gravity too close to center of buoyancy — small transverse imbalances produce large, variable roll (a stable glider holds roll standard deviation < ~0.25°).
A loose internal component (e.g., a shifted nose hydrophone) changing the roll between dive and climb.

A useful tell that the compass itself is good: run the in-situ cal and confirm it agrees with the shore cal — if it does, look mechanical.

Troubleshooting¶

Symptom	Likely cause / fix
"Mag out of range" (red) in RevolutionTest, no points collected	Too much magnetic interference at the location, or the attitude sensor's serial stream isn't really getting to the software. Move well away from metal/electronics; confirm `talk att`/`talk attitude` is actually streaming; take the sensor in/out of service and retry.
RevolutionTest won't connect / icons don't change color	Another program still holds the COM port — fully close your serial terminal so the port is free, then start RevolutionTest on that same port at 9600 baud.
`talk attitude` shows only one line of output (often G3S)	The compass is stuck in TNT sample mode, so it isn't streaming continuously. From GliderDOS confirm `m_attitude_rev_mode = 0` and `m_attitude_rev_measure_state = 0` (defaults), then re-issue the compass command to return it to continuous output.
Calibration "inadequate data points" / won't finish	Coverage is incomplete — rotate much more aggressively through pitch and roll extremes (the roll/`z` bins need the most). Also check for a magnetized battery pack (degauss it).
Compass good on bench, ~12° off in a post-cal check	The solenoid is open during your GliderDOS check — inflate the air bladder and wait for the air pump to stop so the solenoid closes, then re-check.
RevolutionTest won't run / "no permissions" / fails to launch	Historically version- and Java-sensitive — some users only succeeded on Windows XP (ideally with a real serial port) and struggled on Windows 7. Run as administrator; try a known-good machine.
Compass keeps going out of service near fields, frequent `attitude_rev` aborts	Expected near strong fields; if persistent, verify Service Bulletin 011 warning-level settings are applied. Recalibrate after recovery.

Quick reference¶

Command	What it does
`talk attitude` / `talk att`	Stream raw attitude data (and close the solenoid); G3/G3S use `attitude`, G2 uses `att` after `exit pico`.
`report ++ m_heading m_pitch m_roll`	Watch heading/pitch/roll from the console while rotating the glider (inflate air bladder first for a valid check).
`report clearall`	Stop the live reports.
`compass_cal ?`	List the `compass_cal` sub-commands.
`compass_cal get_offsets`	Read the current X/Y/Z hard-iron offsets.
`compass_cal set_offsets X Y Z`	Write new X/Y/Z offsets (keep the negative signs!).
`use` / `use - iridium`	Check device status / take Iridium out of service during a cal.
`exit reset`	Reset the glider to bring `attitude_rev` back online after an in-situ cal.
`run attcal2.mi`	Run the (community-tuned) in-situ calibration mission.
`send *.cal -num=1`	Send the generated calibration file off the glider.