Hello,

since a few flights, my Spark is having a very slow IMU heatup time.

Before that, it was "ready to fly" right after power up when the app and RC are also ready.

Now, it says "calibrate IMU", every beginning of flight sessions. After looking into more carefully, I understood that this is the IMU "heating time" which is at play :

if start a "cold" (ambiant temperature) aircraft, even with a fresh calibration, then the gyro readings are all red at the beginning, and they slowly decrease to green values allowing flight.
if I start an already warmed up aircraft, then everything is Ok.

This is 100% repeatable, but what I do not understand is why this behaviour change suddently without any action on my side (no firmware change, no app change)

At least, now I know I do not have to recalibrate, just wait (about 1.5 minutes) until the bird is warm and IMU gives decent readings.

Searching around, I saw that some users of P3 had similar issue and some recommand to do a "cold" calibration i.e. calibrate the bird after lowering it's temperature (e.g. in fridge). I've been trying this but it do not solve anything...

Any clue why this is happening and if there is a way to return to "immediate" starting time ?

I would advise to recalibrate IMU , leave Aircraft in cold room minus the battery for about an hour then calibrate, test to see if it warms up quicker.
From time to time and vibration can effect IMU .

IMU "Inertial measurement unit" .

As far as I see it, IMU calibration on a level surface updates a table of values the flight controller software uses as a reference for a level stationary hover. From there the craft responds predictably to flight commands.  It also likely measures any sensor noise and thermal drift so that these technical imperfections are accounted for when using IMU sensor data in flight. - this is likely why there must be no vibrations during the calibration process.


Bad IMU calibration could cause drift and attitude issues as the flight controller fights to hold the craft in what it thinks is the correct attitude as opposed to the correct physical attitude.


The IMU usually has 2 types of sensors – angle and acceleration and in turn 3 sensors of each type measuring in the X,Y and Z axis. These sensors can, through vibration, aging, impact etc, drift in their response over time and thus an IMU calibration will establish a new reference for the IMU’s level/stationary state that the flight controller can work with to restore stable flight.


Think of calibration as the bringing back into line the measured craft attitude with the true physical craft attitude.


As to how often an IMU calibration is needed, Refer to your manual ; certainly after any impact or if there are

Hi,

I already calibrated the IMU many (really many) times since this slow warm up issue appears (not resulting of any impact or any hard landing)
First calibration was because I had this "calibrate" message... Then I noticed that it was asking for calibration when flight sessions were not close to each other...
Then I realized and observed that this was temperature/warm up related...

I always perform the calibration an a flat, hard leveled surface, far from any metalic objects and vibrations free. Just to say, i'm IMU/INS specialist (not those kind of low-end ones but still have good notions on the physics at play with those cheap devices)

I guess the most important part of the calibration is the bias calibration (reading of the sensors at 0 roll/yaw and 0 angular velocities).
When I look at the 2 readings displayed on the sensor status page, he accelerometers part is always green and pretty unsensitive to temperature. The gyro (angular velocities) part is too high at power up. It's red saturated at 0.05 (don't know the value...) then it decreases when temperature increases and eventually becomes green with < 0.005 finale value (I think it turn greens below 0.015)

You says that the calibration is tabulating the values with temperature, are you sure about that (this is common to do so with IMU and would indeed be a good idea...) do you have info if DJI actually doing that ?
When first chilling the Spark (not the battery) I indeed noticed that stage 1/6 last much longuer than same step when starting with a bird a room temperature or with a hot bird, as if it where indeed tabulating across the temperature readings that are increasing from start values after power up to some kind of minimal / stable temperature value.)

Still this is quite long to test/sort out, because after each calibration, I have to wait long enough for the bird to chill to room temperature again... otherwise the next startup is immediatly "green" because IMU is already warmed-up.