Left/Right Motors Diverge At Rest?!

[musclera]

http://www.youtube.com/watch?v=0i3zYn8MRUw

I've built a few Multiwii quadcopters in differing sizes but basically I have not gotten the darn things to fly with any amount of stability. IN the video above I have the copter level on a table. When I arm the motors and throttle up I see that within a few seconds the left and right motors are at different pulse widths. When I do this with the props on I can tell that all the motors are not spinning at the same speed despite the copter being level and there being no inputs from my transmitter.

What is going on? I can't imagine this is normal...

My setup:
Arduino nano 328p
Genuine WMP
Turnigy 9X w/ accompanying receiver

[Pyrofer]

Calibrate acceleromter.
Trim the controls.
Calibrate the escs.

Try again.

[tovrin]

I agree with pyrofer, especially ont he ESC calibration. i had a similar issue with my hobbyking SS esc's, and it was related to my min/max not being the same on all 4 ESC's. I had to calibrate them together at the same time.

[musclera]

That's the thing - I'm not using an accelerometer.
And as you can see in the video the controls are nearly perfectly trimmed.
This is an issue with the output PWM - ESCs are not even attached.

Think of it as just the arduino and a WMP. With no input from the gyro or the sticks (which are centered), the left and right motor pulse widths are drastically different when they should (all 4 should) be the same.

[bill516]

If you are using only a WMP your cycle time should be about 6000 with nunchuck it is about 3000. How can you you get the motors to spin up if you havn't got the esc's attached. The GUI is not a calibrated piece of software it is only a visual guide, stability is finally obtained through flight testing, the GUI will give you a ball park figure to start with, which will be somewhere near where it needs to be. Dont pay too much attention to the numbers, you are looking that the numbers are approx the same, once you get flying the motor rpms will not be stable they will be fluctuating as they respond to gyro, but you probably wont hear it.

Your motors are not trimmed the right motor is running faster than the other three use Tx trim to balance motor rpms, this is only an approx set up. You will need to do final trim with a flight test, have a read of the threads by Tovrin and AndyM in general discussions, I have explained to them how the setting up works. BTW what esc's and motors are you using there may be an incompatability.

[musclera]

If you are using only a WMP your cycle time should be about 6000 with nunchuck it is about 3000.

Why would the cycle time be higher when you are only taking gyro readings? It is higher when you have a nunchuck + WMP as it interleaves them:

Cycle time

With a WMP only configuration, the cycle time should be something like 2000 at 400kHz, and 3300 at 100kHz.

With a WMP+NK configuration: the cycle time should be always around 6000 whatever the I2C speed.
If the cycle time is different, you must increase INTERLEAVING DELAY.
The cycle time is good when it is 2x INTERLEAVING DELAY.

How can you you get the motors to spin up if you havn't got the esc's attached. The GUI is not a calibrated piece of software it is only a visual guide, stability is finally obtained through flight testing, the GUI will give you a ball park figure to start with, which will be somewhere near where it needs to be. Dont pay too much attention to the numbers, you are looking that the numbers are approx the same, once you get flying the motor rpms will not be stable they will be fluctuating as they respond to gyro, but you probably wont hear it.

Your motors are not trimmed the right motor is running faster than the other three use Tx trim to balance motor rpms, this is only an approx set up. You will need to do final trim with a flight test, have a read of the threads by Tovrin and AndyM in general discussions, I have explained to them how the setting up works. BTW what esc's and motors are you using there may be an incompatability.

The ESCs or motors do not need to be attached here - they are merely consumers of the pulses generated by the arduino and only have an affect on the system if they induce a change in gyro readings.

My concern here is that the arduino is telling the right motor to spin faster than it is telling the left motor to spin. This when there has been zero input to the system by way of stick input or a change in gyro readings.

[Pyrofer]

"My concern here is that the arduino is telling the right motor to spin faster than it is telling the left motor to spin. "

Wouldn't trim fix this? Isn't that what its for?

[bill516]

Yes Pryfor thats what trim is for, if the world was perfect we wouldnt need trim as everything would be spot on.

Musclera

Yes your correct on cycle time got my numbers mixed up with only wmp it should be about 3000 or less if you can get it to 400kHz.

The esc's dont control the gyro, the gyro is affected by the attitude of the machine which in turn affects the signal to the esc's

Your arduino is telling the motors what it has recieved from the Tx and the gyro. With the machine is stationary use the Tx trim to balance the motor rpms in the GUI, then leave it alone and watch what happens to the rpm and gyro readings. Just because your Tx trims are neutral doesnt mean that the Tx is sending a neutral signal. The stick pots are just variable resistors and as such what is middle for one is not likely to be the middle for another. To prove what I've just said get the thing running and move a Tx trim from one extreme to the other Yaw/Pitch/Roll your choice and watch the motor rpms, repeat on the other two if you want.

Dont try to make this more complicated than it already is, get your CofG in the middle of your machine seeing as its a quad, and get the balance the motor rpms in the GUI to all approx the same, then do a test fly. So long as you dont have any vibration issues you should be able to get into a hover, trim as required to get an hands off or almost hands off hover.

[lukibob]

I watched your video, and listened to your description. It seems normal to me. The data from the motor channels in the gui, with the quad on the ground, sometimes doesn't show all levels what you would expect. If you move the quad while the motors are running and the appropriate motor output increases, it should be fine.

The small difference in the left to right channel should not matter once the quad is in the air. The numbers will not be exact.

Try moving the sticks and moving the gyro with the gui active and it will show you how the quad should behave in the air.