Expand description
Entacore’s AIM flight computers’ logs, as the AIM XTRA and AIM BASE software exports a flight with Export to Excel…: a CSV file (AIM XTRA/BASE v4.x.x Manual, Entacore, 2 February 2025, p38).
The AIM XTRA samples its sensors at different rates: its barometer every 20 ms, its 100 g accelerometer, 16 g three-axis accelerometer and gyroscope every 10 ms, its magnetometer every 100 ms, its temperature every 500 ms and its GPS’s position every 200 ms (manual p2), each written at a whole multiple of its period that the owner sets for each phase of the flight (p21). So the export gives each group of channels its own clock: “there are multiple time columns. Each time column applies to the columns to its right until another time column is reached” (p40). One header line names the columns; each row below holds the next sample of every group, so a row’s cells for different groups were taken at different times, and a group that has run out leaves its cells empty:
time,Gyro X,Gyro Y,Gyro Z,time,pressure,Pressure MSL,Pressure AGL,time,acceleration,…
-0.52,0.07,-0.14,0.21,-0.51,97700,302.87,0.17,-0.52,1.0039,…The AIM BASE ground station’s software writes the same layout of the telemetry it received
(p28), with fewer channels at slower, irregular rates (p21, p38) and the radio’s RSSI.
Recognition. The first line that isn’t blank is comma-separated, starts with a time
column, names at least two, and names at least one of the AIM’s own channels, such as Pressure AGL, lat. X accel. or eject. bat. voltage (the reader’s MARKS); names are compared
trimmed, without trailing points and without regard to case. A Raven’s export starts with
Time@… columns and names none of them.
Channels. Each name the reader knows, what it is read as, and where its unit comes from. The column keeps the file’s name, trimmed. A header naming a column twice is refused:
| Column | Quantity | Unit, and from where |
|---|---|---|
time | the clock of the columns to its right (p40) | seconds or milliseconds, from GPS Time of Week or the step (below) |
pressure | Quantity::Pressure | Pa, kPa or inHg, from the values (below); the manual’s sample export writes 100182 (Table 2, p40) |
Pressure MSL | Quantity::AltitudeMsl | m or ft, from the pressure (below): the software’s altitude from “Standard Atmospheric table data” (p22) |
Pressure AGL | Quantity::AltitudeAboveGround: the primary track | as Pressure MSL |
acceleration, lat. X accel., lat. Y accel. | Quantity::Acceleration | g, × 9.806 65 m/s² (standard gravity): the manual plots accelerations “in G’s” (p23–24) |
Gyro X, Gyro Y, Gyro Z | Quantity::AngularRate | degrees per second (p24) |
Mag X, Mag Y, Mag Z | Quantity::Unrecognized, as written | below |
eject. bat. voltage, comp. bat. voltage, A Voltage … D Voltage | Quantity::Voltage | volts, from the values (below) |
temperature | Quantity::Temperature | °C or °F, from the altitudes’ unit and the manual (below) |
lat, long | Quantity::Latitude, Quantity::Longitude | decimal degrees, as the manual’s sample export writes them (Table 2, p40); a value beyond ±90° or ±180° is a gap |
GPS MSL | Quantity::GnssAltitude | the altitudes’ unit, if it agrees with Pressure MSL (below) |
GPS Time of Week | Quantity::Unrecognized, as written | milliseconds, where the clock is decided by it (below) |
SVs | Quantity::Number | the satellites used in the GPS’s solution (p25) |
RSSI | Quantity::Number, as written | dBm, the strength of the radio signal the AIM BASE received (p29), not an SI unit |
launch detect, apogee detect, A cont. … D cont., A on. … D on | Quantity::Number, 0 or 1 | flags; see Events |
A column the reader doesn’t know is kept as written, Quantity::Unrecognized, with a note.
Each unit gets a note on the record saying how it was read and on what evidence; a unit no
evidence settles is kept as written, Quantity::Unrecognized, and the note says why.
The clock. The manual states no unit for time. GPS Time of Week counts milliseconds: its
values stay below a week’s 604,800,000, and the manual’s sample export counts it up by 5,000
over 5 of its time in seconds (Table 2, p40). The reader takes the median, over the
channel’s consecutive samples, of its rise over its clock’s: within 10% of 1,000, time is in
seconds; within 10% of 1, in milliseconds, and every clock is read × 0.001. As of 2026-10-13
the median is 969.8 to 1,002.4 in the 16 AIM files among the project’s sample logs (the BASE’s
clock, the time a packet came in, is the less regular). Without such a channel, the unit comes
from the step, as the Eggtimer reader’s does: the AIM writes each sensor every 10 to 500 ms
(p2) times a whole multiple the owner sets (p21), so the fastest clock’s median step reads
0.009 to 2 in seconds (the 10 ms period less 10%, to 2 s; 0.0097 to 0.773 in the sample
files) and 9 to 2,000 in milliseconds. A step in neither range, or no clock with two samples,
is refused (LogError::Unit). The time’s zero is the software’s: the start of the
recording, or the time the owner chose to make zero, such as the launch (p33).
The pressure. The manual’s software shows the pressure in kPa or inHg (p25), but its sample
export writes pascals (Table 2, p40). The reader takes the unit from the first value that is
a launch site’s pressure in one of them, on the pad, so a placeholder ahead of it is passed
over: pascals within 50,000 to 110,000 (the bounds crate::rrc3 argues), kilopascals
within 50 to 110, inches of mercury (3,386.389 Pa, NIST SP 811) within 14.8 to 32.5;
otherwise the column is kept as written. Every sample file is in whole pascals.
The altitudes. Pressure MSL is the pressure’s altitude in the standard atmosphere, and
Pressure AGL the same less the logger’s ground altitude: over the 16 sample files Pressure MSL is within 0.06 m of the U.S. Standard Atmosphere’s pressure altitude of pressure, and
Pressure MSL less Pressure AGL is constant to 0.02 m in each. The software offers metric or
imperial units (p33); the reader decides between meters and feet as crate::rrc3 does: from
the pressure’s pad sample to the one of lowest pressure, the climb the pressure implies in the
U.S. Standard Atmosphere, 1976 (hpr_atmos::ussa76), against the column’s climb read in
meters and in feet; the unit within 25% is taken. A climb under 2 m (where the pressure’s
1 Pa step, 0.08 m at sea level, is 4% of it), a file without a pressure of known unit beside
the altitudes on their clock, or a climb in neither unit is refused (LogError::Unit): the
primary track has no unit to be read in.
The temperature is read in °C beside meters and °F beside feet: the software sets one unit system, metric or imperial, for all its displays (p33), and its sensor reads °C or °F (p25). Its first value must also be one an electronics bay can have in that unit, −40 to 60 °C (−40 to 140 °F), or the column is kept as written; the sample files’ first values, 27.4 to 50.6, are in °C.
The voltages are read in volts where every value lies within 0 to 20: the computer’s battery is 3.5 to 8.4 V and the ejection battery at most 10 V (p10), and the lines’ readings sit between, where millivolts would be thousands. A column with a value outside is kept as written. The sample files’ values are 0.05 to 8.33.
GPS MSL is read in the altitudes’ unit if it agrees with Pressure MSL: over the GPS
samples within 1 s of a barometric one, the median difference is within 300 m in that unit
and beyond 300 m in the other. 300 m is the error a day’s weather gives a standard-atmosphere
altitude when the sea-level pressure is 36 hPa off the standard 1,013.25 (8.4 m a hPa near sea
level); the manual’s own sample export reads 283 m by GPS beside 95.6 m by pressure (Table 2,
p40), and the sample files’ medians are 28 to 173 m. Otherwise, as at a site too low for the
units’ factor of 3.28 to show, the column is kept as written.
The magnetometer. The manual gives tesla (p24), but the field the sample files’ three columns make, 0.0011 to 0.0022 in the 9 with them, would be 17 to 34 times the Earth’s at its strongest, about 65 µT, so no source bears a unit out; the record has no quantity for a magnetic field either, and each file’s note gives its own range.
The primary track. The record’s time and altitude are Pressure AGL’s, on its own clock:
the AIM’s barometric altitude above the ground it was calibrated to (p22), the altitude above
the pad that the logger itself states. A file without it is refused: working one out of the
pressure is a reading, not this reader’s to make. The primary track takes its samples in time
order, the first at each time, leaving out any sample without a time; a note counts the
samples put in order or left out (the channel keeps them all, in the file’s order). Debrief’s
sample, a thinned copy of a flight with one row out of order at its end, has one. Where the
altitude above ground reads more than 100 m off zero at the track’s first sample, as one of the
sample BASE files does where its ground reference came from an altitude far from the pad’s, a
note says so. No column is a vertical speed, so FlightLog::vertical_speed_m_s is None;
the temperature and comp. bat. voltage fill FlightLog::temperature_k and
FlightLog::battery_v only on the altitude’s clock, which none of the sample files has.
Events. launch detect, apogee detect and the four outputs’ A on. … D on are flags
the AIM logs every 100 ms (p2); each time one goes from 0 to 1 is an Event named by its
column, at the sample’s time, in time order. A flag already 1 at its first sample is no event.
The AIM detects the launch some time after leaving the rail, once its filtered velocity has
stayed above 30 m/s for 0.5 s (p36), so launch detect is no liftoff time.
Rows one cell too wide. In 11 of the 13 sample files with GPS columns, every row after the
GPS group (time,lat,long,GPS MSL,SVs) runs out holds one cell more than the header: the group
is written as six empty cells, not five, so the groups to its right sit one cell right of
their names. A row one cell wider than the header whose GPS group’s cells and the one after
them are all empty has that one after dropped, counted in a note, if every group’s cells are
then all filled or all empty, as they are in every such row of every sample file; otherwise
the extra cell is somewhere else, and the row is refused, naming its line. Any other row of
another width is refused too, so the record never holds more values than the file has cells.
Cells. A group with any cell filled in a row is a sample: an empty time or value in it is a
gap (NaN), counted in a note; a cell that isn’t a number is refused, naming its line. Lines may
end in CR, LF or both, blank lines are skipped, and a byte-order mark is dropped.
Where this comes from. The layout follows Debrief’s reader (lib/parsers/entacoreAim.ts,
MIT, the project owner’s own; see THIRD-PARTY-NOTICES.md), which reads each channel on its
own clock and recognizes the file by Pressure MSL and Pressure AGL; the units are the
manual’s, and the files’, as above. Where this reader departs from Debrief’s: it keeps every
column on its own clock where Debrief resamples a few onto the densest; it takes the altitude
above ground from the export’s own column where Debrief works one out from the pressure; it
drops the extra cell of a run-out GPS group, which Debrief reads shifted; and it states the
flags’ rises as events.
Functions§
- read
- Reads an Entacore AIM CSV export from its text.