Hi everyone! I am new to my APM 2.6 and have been doing some simple flights with an Ansley Peace Drone to learn my way around the technology. I am trying to configure/tune the APM, but am not currently using an airspeed sensor, and can't figure out how the APM handles airspeed in this case. I've found very little online.

First, I am wondering how the APM uses the ARSPD_FBW_MIN and ARSPD_FBW_MAX parameters in the absence of an airspeed sensor. Does it use these parameters, and if so, does it have an algorithm for estimating airspeed?

Second, I am not sure the flying characteristics of my plane, and am trying to determine its range of flying airspeeds from telemetry logs of a MANUAL flight (so I can set ARSPD_FBW_MIN and ARSPD_FBW_MAX). Can anyone tell me what the APM is actually measuring in the "airspeed" field? I can't decipher the plot below. I was flying racetracks around the pattern in strong winds, but as you can see in the plot, airspeed consistently indicates well above groundspeed for the entire duration of the flight.




You need to be a member of diydrones to add comments!

Join diydrones

Email me when people reply –


  • You have knower the method about apm2.6 airspeed without airspeed sensor?
  • To close this discussion out, I finally figured out was causing the strange airspeed readings. I had ARSPD_ENABLE set, despite not having an airspeed sensor installed. I have no idea why that would lead to these high ghost readings, but when I disabled that parameter, they went away.

  • I've been doing more testing, trying to figure out what vfr_hud airspeed is actually indicating with no airspeed sensor installed. The attached plot shows three engine runs on the bench, with the airplane fixed in place. As you can see, vfr_hud airspeed is not actually indicating airspeed, but is somehow linked to throttle movement.

    Can anyone tell me if this is a bug or a confusing feature? If a bug, how should I be reporting it?


  • Thank you for the replies, as these are quite helpful! However, I still don't understand the numbers indicated in the log file. Groundspeed oscillates between 10-20 m/s, which makes sense given that I was flying patterns in wind. If the groundspeed is accurate, then airspeed should be somewhere around 15 m/s. Instead, the log shows an airspeed almost consistently above groundspeed, with peaks of 50-70 m/s. It's hard to see how these numbers are anything but garbage. Am I missing something?

    Actually, now that I'm staring at this plot again, a light bulb just went on... I noticed earlier in the log that "throttle" indicated zero the entire flight. Is the APM reporting a throttle percentage setting in the airspeed_vfr_hud_t field instead? That is almost what the plot looks like.

    • T3


      Your plots do not make any sense to me, either.

      Hopefully someone from the APM community can help you.

      Best regards,


  • The only way that I have found to change the average cruise speed is using elevator trim and cruise throttle. I have several airplanes with no airspeed sensors and they fly amazingly well. The latest firmware does an excellent job of computing airspeed so I fond no need to go to the complexity of using and tuning an airspeed sensor.


    I have also found that the min and max parameters do nothing.



  • T3


    Early versions APM used the direction cosine matrix and this method.

    The basic idea is that when a fixed wing aircraft turns, the magnitude of the change in IMU velocity vector is equal to the magnitude of the air speed vector times the magnitude of the change of the column of the direction cosine matrix that represents the fuselage axis as seen in the earth frame of reference.

    The original version of the algorithm used GPS velocity. The latest version uses IMU velocity.

    Recently the APM community has implemented extended Kalman filtering, I am not sure what method is being used in that case.

    Best regards,


  • This is referred to in the aviation business as PFM, Pure F'ing Magic.
    The magicians (Tridge and Paul) use accelerometers to calculate the approximate wind velocity. It's amazing how well it works, although an airspeeds sensor is certainly much better.
This reply was deleted.


Hiroki Tanaka liked Hiroki Tanaka's profile
Oct 13
Derrick Davies liked lisa TDrones's profile
Sep 23