Good vibrations
Posted: Mon Mar 09, 2015 3:04 pm
Hello all,
spring is coming and it is time for tweaking and flying.
Above all, I find getting rid of vibrations or minimizing them as much as possible the most difficult part of tweaking.
I balance the props, in both axes, by using a static prop balancer. I use tape for the long axis and epoxy near the shaft (sometimes throwing in a nut or a bolt in severe cases).
I then balance the motors using dynbalancing build of MWII. Down to 2x2 millimeter patches of tape.
And I'm still getting the damn good vibrations.
There are lots of methods for measuring vibrations:
- By listening to the props
- Using a smart phone taped near the motors
- Watching the sensor outputs in Multiwiiconf
- ...
Perhaps a lot has been said on this topic; what I'm missing is a way of communicating the OBJECTIVE measurements of vibrations.
Thinking of RMS values. The graphics of sensor output are nice; but hard to compare. A single value here is best.
I am experimenting with modified code in MWI that measures RMS values of GYRO and ACC raw sensor outputs. I find after good vibration dampening, gyro is almost at zero, but ACC is still showing values.
If people are interested, I can post my RMS calc code (it outputs gyro and acc RMS values measured over 500 samples through DEBUG values in multiwii gui.
Since the values are what the MWII code sees, they should be of objective value to MWII users and directly comparable. I wonder how much vibrations you people get?
Of course the frame vibrations could differ; depending on FC board(sensor) vibration isolation.
But I think vibrations as acc sensor sees them are what counts.
If there is interest on this topic, I will post my code here and we can measure and compare.
The method:
RMSgyro = SQRT(SUM ( gyro(x)^2 + gyro(y)^2 + gyro(z)^2) [over 500 samples])
RMSacc = SQRT(SUM ( acc(x)^2 + acc(y)^2) [over 500 samples]) *** ignoring Z component of acc
Copter must be levelled and acc calibrated prior to running the test. I am not normalizing the values, so this is important.
Use soft pads under copter to decouple it from the table below.
The throttle for measurements is to be set at perhaps hover power.
What do you think?
spring is coming and it is time for tweaking and flying.
Above all, I find getting rid of vibrations or minimizing them as much as possible the most difficult part of tweaking.
I balance the props, in both axes, by using a static prop balancer. I use tape for the long axis and epoxy near the shaft (sometimes throwing in a nut or a bolt in severe cases).
I then balance the motors using dynbalancing build of MWII. Down to 2x2 millimeter patches of tape.
And I'm still getting the damn good vibrations.
There are lots of methods for measuring vibrations:
- By listening to the props
- Using a smart phone taped near the motors
- Watching the sensor outputs in Multiwiiconf
- ...
Perhaps a lot has been said on this topic; what I'm missing is a way of communicating the OBJECTIVE measurements of vibrations.
Thinking of RMS values. The graphics of sensor output are nice; but hard to compare. A single value here is best.
I am experimenting with modified code in MWI that measures RMS values of GYRO and ACC raw sensor outputs. I find after good vibration dampening, gyro is almost at zero, but ACC is still showing values.
If people are interested, I can post my RMS calc code (it outputs gyro and acc RMS values measured over 500 samples through DEBUG values in multiwii gui.
Since the values are what the MWII code sees, they should be of objective value to MWII users and directly comparable. I wonder how much vibrations you people get?
Of course the frame vibrations could differ; depending on FC board(sensor) vibration isolation.
But I think vibrations as acc sensor sees them are what counts.
If there is interest on this topic, I will post my code here and we can measure and compare.
The method:
RMSgyro = SQRT(SUM ( gyro(x)^2 + gyro(y)^2 + gyro(z)^2) [over 500 samples])
RMSacc = SQRT(SUM ( acc(x)^2 + acc(y)^2) [over 500 samples]) *** ignoring Z component of acc
Copter must be levelled and acc calibrated prior to running the test. I am not normalizing the values, so this is important.
Use soft pads under copter to decouple it from the table below.
The throttle for measurements is to be set at perhaps hover power.
What do you think?