Moderator

ArduIMU V2 flat suggested/reccomended mod?

It was suggested here, by Chris, that a mod to the ArduIMU V2 (flat), similar to that made on the APM Oilpan, would help eliminate (or at least tame) the linear acceleration problems being experienced by some. The mod involves removal of filter capacitors on the board.
  1. Are the capacitors shown in the picture below the correct ones to be removed?
  2. Do you leave the circuit open here or bridge the connecting?
  3. Does anyone have any experience having made these mods already?


Are these the correct capacitors to remove, 3 x 0.1uf ? Can anyone comment on their experience making this mod on an ArduIMU V2 (flat)? Isthis now a recommended mod?

Nathaniel

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

Join diydrones

Email me when people reply –

Replies

  • Developer
    I have discussed this suggestion with Chris after he made the suggestion. IMO this mod will not help with the linear acceleration issue at all.

    The linear acceleration issue is due to the high gain needed to correct the pitch/roll drift in the gyros. When you have linear acceleration the resulting total acceleration vector (linear + gravity) is interpreted as pitch and the drift correction algorithm "corrects" accordingly. Since the gain is high the time constant is very short - on the order of 1 second.

    On ArduPilot Mega which has better gyros I have tested this and seen that lower drift correction gains relieve this problem. Unfortunately the gyros on ArduIMU are not good enough to significantly lower the gain and keep the drift in check in real world conditions.

    With the gyros on this board I think the only real strategy for dealing with this problem is to be more aggressive with the gain deweighting. Line 153 in the DCM.pde file is:

    Accel_weight = constrain(1 - 2*abs(1 - Accel_magnitude),0,1); //

    This produces a gain deweighting factor which is 1 (no deweighting) when the Accel_magnitude is 1 (meaning 1g, or just gravity), and which decreases to 0 as Accel_magnitude either increases to 1.5 or decreases to 0.5. The idea is that vibration or linear acceleration will cause the acceleration magnitude to vary from 1g, and the more it varies the more we deweight the gain. This has been successful in increasing the vibration tolerance.

    You could substitute another equation that would deweight more aggressively. For example you could take the deweigting factor to 0 at Accel_magnitude of 0.8 / 1.2
This reply was deleted.

Activity