I am asking about the internal design of the autopilot of a CRJ200 (no autoland).

I was flying an ILS approach when the AHRS heading started to drift rapidly by 30 degrees. The autopilot somehow was not able track the ILS and started to to fly an "S" turns. After this incident I had a discussion with TRE. He insisted that the ILS has nothing to do with heading. I think that it's impossible or very difficult to design an ILS tracking autopilot without internal heading reference.

How can an autopilot know that a plane is turning and how much without some internal heading reference? Can anybody provide any info on this topic and how the autopilot is implemented?

  • 4
    $\begingroup$ What aircraft/autopilot? There are huge differences in design. Most small aircraft A/P just blindly track the ILS, but an air transport approved for autoland has a very complicated A/P system to ensure safety. $\endgroup$
    – Gerry
    Feb 17, 2021 at 12:45
  • $\begingroup$ CRJ200 no autoland $\endgroup$
    – Andrius
    Feb 17, 2021 at 14:33
  • $\begingroup$ Yeah on an older system like Collins Proline once the loc is captured the ap is just responding to the loc deviation, and if you lose the loc it goes into heading hold. $\endgroup$
    – John K
    Feb 17, 2021 at 16:13
  • $\begingroup$ Oh that problem. That's related to a loc signal broadcast antenna issue at certain airports. I forget the details. Jazz was complaining about incidents at Pearson where their 200s were doing that (overshooting the loc and than turning back to crisscross it) and the capt had to disconnect and hand fly. $\endgroup$
    – John K
    Feb 17, 2021 at 16:16
  • $\begingroup$ Your conclusion does have merit and makes sense. Although, logic can be programmed into the system to know how to navigate the LOC based on the Hertz received (90 vs 150). The FD/AP could then navigate by reference to equalizing these two signals within a certain distance from the center. Receiving a stronger signal would indicate which side you are on. While the weakening or strengthening of the other signal would indicate whether you are traveling toward or away from the center. You can determine which way to turn from this data. How much to turn would be a factor of ground speed. $\endgroup$
    – Dean F.
    Feb 17, 2021 at 17:03

2 Answers 2


I'll give an answer with the caveat that I haven't used the system in over 10 years. So maybe do some additional research.

The Collins Proline Flight Director used on the 200 (it's not the Autopilot per se, it's the Flight Director telling the AP what to do), and most similar ones I would expect, needs heading, bank angle and turn rate data to anticipate and judge correction heading offsets, intercept angles, etc, to track navigation solutions whether on "white needles" (FMS navigation) or "green needles" (direct radio signal navigation like VOR or LOC).

If you think about it, it has to; the Flight Director is substituting for your brain, and it needs the same data, about where the airplane is headed and how much it's turning, that you would need to hand fly an approach on raw data.

If you're hand flying and drifting to the left of the loc, you mentally estimate the heading change needed to re-intercept the loc, fly the new heading and wait. Without compass information you would have a pretty tough time doing it with any efficiency. If your heading indication did the same thing while you were hand flying, you would probably end up doing something similar if you kept trying to fly the LOC by ignoring the heading indication.

So if the AHRS gyrocompass unit the Flight Director is slaved to starts to slew on its own, it'll get confused and go into a kind of software loop with the resulting S turns. I would assume that selecting XFER to select the opposite FD/AHRS system fixed the problem.


Most autopilots in small aircraft blindly track the ILS (same for VORs) without any heading reference, which is actually simpler because heading can be substantially different from ground track, especially in crosswind. Technically speaking, the autopilot is a PD-controller, performing two tasks simultaneously:

  1. When the needle wanders left, turn left, and when the needle wanders right, turn right. When the needle stops, the autopilot has found the correct heading. The actual position of the needle does not matter, only that it stops wandering. This is the loop that tracks heading, the differential (D-) part of the PD-controller.

  2. When the needle is left of center, turn left, and when the needle is right of center, turn right. This is the loop that brings you back on the localizer, the proportional (P-) part of the PD-controller.

Both loops run simultaneously and the commanded turn rate is the sum of both. If the needle is left of center but wanders right, both loops cancel and the autopilot continues straight ahead. As the needle gets closer to the center, the proportional loop gets weaker and the differential loop wins, with the result that the aircraft turns slightly to the right and the needle stops wandering. The needle is now in the center and not wandering anymore, which means that the autopilot has successfully intercepted the localizer.

Note that while the autopilot does not use any heading reference, it still commands a turn rate and uses the turn coordinator and ailerons to fly the commanded turn rate. If your AHRS served as a turn coordinator for the autopilot, then the aircraft will veer off course if the AHRS fails.

  • $\begingroup$ Modern inertial navigation systems have a set of rate gyros (laser-ring or MEMS (vibrating)) and integrate the heading and attitude from the rates. So heading drifting means rate of turn did as well, because they are fed from the same gyro. $\endgroup$
    – Jan Hudec
    Apr 28, 2021 at 16:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.