Skip to content

Crafty's Old Board — Cleaning Flux and Ink Stopped the Runaway Heating

The board pulled from Crafty after the unresolved temperature warning went back on the bench to see whether it could still be used. With no airflow it overheated past 85 °C and was still climbing. After two isopropyl cleanings it levels off at 72 °C. It has not been flown, and nothing was tested under battery or motor load.

The VTX must not run without an antenna. The bench had none fitted. The board is now saved at vtx_power = 1 (25 mW), with the AUX6 power mapping cleared and VTX PIT MODE (mode 39) always on (aux 4 39 0 900 2100 0 0). That is not “off”: Betaflight has no VTX off. Unplug the board between tests.

The board

BETAFPVG473_V2 on the BETAFPV dev firmware 2026.6.0-alpha, craft_name = Crafty, vcd_video_system = AUTO, osd_displayport_device = MAX7456, gyro ICM42622P at 8 kHz, 47 % CPU. The gyro separates it from the replacement board, which has a BMI270. The MCU ID was not read.

Photos showed the battery-lead pads (+ and –) coated in black baked flux, with uneven solder around the R001 shunt. The pink tint is marker ink used to colour parts, which dissolved into the flux. It is not damage.

Bench temperatures

USB only, no battery, no motors. Core temperature from status, sampled every 12–15 s.

Core temperature, no fan, old Crafty board

Cleaning the flux and ink slowed the climb and lowered the plateau. USB only, no battery. Hover for values.

Not cleanedAfter first cleaningAfter second cleaningSecond cleaning, repeat run
304050607080900100200300400500Seconds since first reading (USB power only, no battery, no fan)Core temperature, °COSD alarm 70 °CReplacement board63–65 °C plateauNot cleanedstopped at 85 °C cutoffFirst cleaning75 °CSecond cleaning72 °C, two runs
Show as a table
TimeNot cleanedFirst cleaningSecond cleaningRepeat run
30 s77 °C58 °C52 °C53 °C
60 s83 °C67 °C56 °C58 °C
120 s71 °C64 °C66 °C
180 s73 °C68 °C69 °C
240 s74 °C70 °C71 °C
300 s74 °C72 °C72 °C
360 s75 °C72 °C
420 s74 °C72 °C
500 s75 °C
ConditionCore temperature
Not cleaned, no fan66 → 85 °C in ~2 min, still rising (stopped at the 85 °C cutoff)
Not cleaned, fan on, pid_process_denom 445 °C, flat
After first cleaning, no fan37 → 70 °C in ~85 s, then 74–75 °C for ~7 min
After second cleaning, no fan45 → 68 °C in 3.4 min, then 72 °C flat (73 °C last read)
After second cleaning, repeat from cold, 6 min48 → 72 °C at 5 min, then flat (73 °C last read)
Replacement board, no fan (23 Sep)63–65 °C at 3.5 min

The 17 September bench reading of this board, a 72 °C plateau, matches the second-cleaning result. Only the uncleaned run looks like a fault. Each cleaning lowered the plateau and slowed the climb, which points at the residue, though the starting conditions differed from run to run.

Halving the PID rate (pid_process_denom 2 → 4) changed nothing: CPU stayed at 47 %. The 8 kHz gyro loop cannot be lowered on this board.

What is and is not known

Known: the runaway heating went away after cleaning. The board is stable with the fan on (45 °C) and plateaus at 72 °C without it. Config, gyro and OSD are intact.

Not known:

  • Whether flux and ink leakage was the whole cause. It is the likeliest reading, not proven.
  • Which cleaning did what: both were done between runs.
  • Current draw, motor outputs and DShot telemetry. No battery was connected.
  • Flight behaviour. On 17 September the alarm tripped in flight at low VTX power.

72 °C is still above osd_core_temp_alarm = 70, so on the bench without airflow the warning still fires.

Outcome

Usable as a bench and programming board. Before flying it:

  1. Fit a VTX antenna, remove the always-on pit mode (aux 4 0 0 900 900 0 0) and restore the AUX6 power mapping.
  2. Run a props-off battery test for current draw and all four motor outputs.
  3. Fly a short pack and read the temperature on the OSD. If it warns, raise osd_core_temp_alarm (about 85) rather than chasing it. The replacement board’s 63–65 °C shows the airframe is not the problem.
IPA cleaning is the cheapest test for a board with heavy flux on it, and it should come before a replacement is ordered. It changed the result here, though it is not proven to be the whole cause.