The XSens MTi-1 series modules can provide those kinda rates IIRC, but they're unfortunately constrained to 16g... Mind you, that's with all the inertial processing work done for you
TP
TP
"should". Do you have confidence in this and how have you determined the roll rate?? Making HPR rockets is a sometimes not quite optimal science, especially with the build. It is a good philosophy to design for off-nominal situations, to ensure flight success.The rocket should be less spins than maximum range gyro.
"should". Do you have confidence in this and how have you determined the roll rate?? Making HPR rockets is a sometimes not quite optimal science, especially with the build. It is a good philosophy to design for off-nominal situations, to ensure flight success.
Also be sure not to use accelerometer g reference to compensate for gyro drift in your karman filter. That is a common implementation for terrestrial application but not is valid for a rocket application.The rocket was stable on the flight but I have not recovered my data on last flight.
Do You have any materials how to estimate orientation in rockets? I saw a lot of articles where the kalman filter is compesated by magnetometers and accelerometers also in flying objects.Also be sure not to use accelerometer g reference to compensate for gyro drift in your karman filter. That is a common implementation for terrestrial application but not is valid for a rocket application.
Do You have any materials how to estimate orientation in rockets? I saw a lot of articles where the kalman filter is compesated by magnetometers and accelerometers also in flying objects.
I know The madgwick algorithm but i thought that ekf is implementing on navigation board in The rockets.You can use the magetometer to compensate for gyro drift but not the accelerometer once in flight. The Madgwick algorithm (you can search) is the current gold standard for an IMU implementation. The accelerometer is used to correct the attitude by using the gravity reference vector. This can be used to determine your rocket orientation while the rocket is sitting on the pad. But once the rocket is flying the direction of the gravity reference vector is no longer available. If you don't turn this part of the filter off then the algorithm will confuse the thrust and drag vector with the gravity vector and your compensation will be totally wrong, unless the rocket flies a perfectly vertical trajectory (thrust and drag pointing in the same direction as the undetectable gravity vector).
Yes you can formulate an ekf for that task. But you asked about orientation. Fortunately rocket flight duration is short relative to modern gyro drift rates, if you have a good one with low acceleration sensitivity.I know The madgwick algorithm but i thought that ekf is implementing on navigation board in The rockets.
Are you familiar with the Mahoney algorithm/paper?I know The madgwick algorithm but i thought that ekf is implementing on navigation board in The rockets.
I think your accelerometers and their readings will be fine. I also would like you chance change your terms. Real rocket vibrations transferred to the accelerometer is not noise. It is real motion. Existing integration to state, velocity and position will handle those just fine.Yes I know but i have never implemented on procesors. I will try compare this 3 methods and I will be know which of these is the best. But I hope the noises from accelerometers will be Small
Real rocket vibrations transferred to the accelerometer is not noise.
Yes. The device that the op is using appears to be a decent Novatel imu. The DS doesn't mention an anti alias filter but I have to believe thats standard practice on the ADC of these packages. Analog Devices builds these in on their chips and the filter corner points options are configurable. Perhaps the OP should contact Novatel and ask them.Not noise, but once folded back due to aliased sampling it becomes noise because it's now spectrally in the wrong place.
You almost ALWAYS need an anti-alias filter.
All is well as long as the vibration is not tested too strong, otherwise it may have negative effects on the navigation algorithm. each subsequent integration generates an error, I would not want to increase it by the noise from the rocket's natural vibrations. therefore there will be a number of filters, but it would be good to know the vibrations and introduce estimate to the filter that will be prepared for a possible correction. hence my question regarding the mechanical damping of rocket vibrations.I think your accelerometers and their readings will be fine. I also would like you chance change your terms. Real rocket vibrations transferred to the accelerometer is not noise. It is real motion. Existing integration to state, velocity and position will handle those just fine.
Enter your email address to join: