[kevinybh]

Hi, I'm new here and was trying to open up OP forum but couldn't and now i know why.
But anyway, all the questions and issue were raised in that forum and i believe the solution that was there went with it as well.
I am guessing this is an issue faced and raised by some people and at some point was solved.
But for me, I still couldn't solve it.

Setup:
It's a CC3d EVO i got from a seller in Aliexpress.
Hooked it up with: (if these information is needed)
1) DYS BLHeli 30A OPTA ESCs
2) SunnySky X2207S 2700kV
3) 3S 2800mAH
4) Walkera DEVO 7CH Rx
5) Chinese ZMR Frame (exact clone)

Issue:
After hooking up the battery and letting it arm for about 10secs, I tried to hover it and it drifts to the right.
Of course I tried counter drifting with my sticks and eventually it fights my counter and it rolls to the right with higher angle slowly. As i let go of my stick and positioned it middle, my quad rolls to the right with much higher angle. It resets after I landed it and takes off again and of course the problem is still there.
I've been using CC3D and this is my 2nd build. Didn't have this problem with my previous CC3D.
I tried a few steps like:
1) Re-calibrating the gyro and accel again and again (even tried to calibrate it while letting it sits at an anlge to the left)
2) Setting all PID settings to default value
3) Disabling Pirouette compensation
4) adjusting the attitude filtering 0.00, 0.05, 0.10, 0.15, 0.20

nothing seems to work. and the problem still persists.
I'm running out of idea, any help would be great.

-Kevin

[RayRegan]

When you plug your quad into LP GCS, click on "flight data" in the lower left corner. When your quad is plugged in and not moving, do you see the model shown in the lower left corner moving?

[kevinybh]

Hey Ray,

Well it doesn't move. But if u look at it closely and zoom in, its actually yawing to the right very very slowly turning.
But if u look at the horizon monitor, there's a compass value and the value increases overtime... on board compass fault?

[RayRegan]

In the Vehicle Wizard there is a sensor calibration that seems more critical than the one that is in Configuration tab>Attitude. I have been told that it is very important to do this right. You may try to re do the vehicle set up wizard. I would also recommend you check the box to "Zero Gyros While Arming Aircraft" also in the Configuration tab>Attitude. Be sure to give the aircraft about 15-30 seconds after you plug in the flight battery and then 10 seconds or so after you ARM so the gyros can zero.

[kevinybh]

I've tried ur advice and redo the vehicle setup as well as waited 30secs after plugging in the battery. the problem still persists.
it seems like it's not fighting my counter, i got it wrong.
what it really seems like was as i push more throttle, the roll angle increases and the drift was more vigorous...

[RayRegan]

Is it possible that for whatever reason you are creating more lift on one side of the aircraft than the other? If you are flying in a flight mode that is ATTITUDE/ATTITUDE/AXIS LOCK than I agree that your CC3D is the problem, but if you are not flying in a mode setup this way than there is still hope. Can you be sure that the firmware and setup is the same in all four of your ESC's? Also when you setup the neutral point for each of your motors in vehicle setup wizard, you did each motor one at a time right? I know that this seems silly, but all your motor direction and prop directions are correct right? Are you using a Power Distribution Board? I have heard of some PDB's limiting current because of poor manufacturing to some of the ESC's more than others and cause an issue like this.

If you could be sure that the aircraft will always roll to one direction when throttle is added by holding it with your hand and increasing the throttle. You should be able to feel that it is being powered over to one side. If this is true, I would look for a difference either in the physical setup on the quad or in the software setup. I had to make several changes to the setup with my quad to make it react the way it should. I have my board orientation different than stock which meant I had to set this up in the software. My motor output channels are also different and needed to be reassigned. I was very carful to setup each one of my ESC's exactly the same in their manufacturers setup software. I also at first setup all the motors at the same time in vehicle setup, thinking "how different could they be right". After arming I realized that some motors would start spinning well before others. This could cause an imbalance or delay in the correction I thought, so I went back and re did this setup one by one.   

[jull]

Hello it must be a vibration problem. On mine I balanced propellers if not its me like that.

[TheOtherCliff]

First of all, the quad must be sitting on firm surface (not held in hand) during gyro calibration that happens right after plugging battery in.

Second of all you should never use your transmitter trims.  If you have changed them and don't know how they were set when you did transmitter wizard, you should center trims and run tx wiz again.

It sounds like you are using ATtitude or Rattitude mode.  If so, and it still drifts, you need to adjust rotate virtual roll in GCS -> Configuration -> Attitude => Settings

If I recall correctly you need to increased rotate virtual roll number.  Try adding about 5, save, power off and on, see it that helped.  Beware that I might have that backwards and you would need to subtract about 5.  Adjust till it hovers without drift.

[RayRegan]

Just to clear things up:
Is your CC3D mounted parallel to your propellers or on a plane that is parallel with your propellers?
If it were not mounted flat or parallel to the same plane as the motors and propellers than try adjusting the virtual roll as Cliff said. If it was mounted parallel or on the same plane as the motors and propellers and then calibrated and the mounting had not been adjusted or changed in any way since calibration it should accept whatever angle it was calibrated at as ZERO. That is why you calibrate it. If for any reason you have removed or altered the mounting of the board after calibration than you will have to redo calibration. If you calibrated the quad on a surface that was not parallel to the earth (within reason) than you will need to redo calibration. If your CC3D thinks that it is flat but the aircraft is not flat in a hover ( a condition that would exist if your board was mounted a few degrees out of level with your motors and propellers), and you add power it will continue to accelerate in the direction that it is tilting until you add compensation with your radio. Any time you calibrate or set up your radio in the setup wizard or any other model anywhere you should always  have all your trims centered to give yourself the maximum  amount of adjustment in the future. The" model in hand thing" would have been after your aircraft was armed and had selfcalibrated. It was suggested only as a test to determine if your quad was actually producing more thrust in one direction as a product of something other than a bad CC3D. 

[TheOtherCliff]

Some copter designs do not hover level, even if built perfectly (e.g. tricopter, helicopter).  They must be a little unlevel to stay in one place.

Even in a quad, if any motor is not perfectly vertical it will not hover level.

Although it helps to use a spirit level, you just can't guarantee it will be perfect.  The way to fix this is with rotate virtual.

[RayRegan]

The FC Board will continue to apply power to the respective motor until it determines that it is level per the calibrated angle. If the aircraft was not built with all four (in this case ) of its motors pointed up (within reason) than it will drift. Drift does not seem to be what he is talking about. He is building a ZMR 250, like mine. I have crashed and tweaked one of my arms. I have not replaced it yet, although I can see that it puts one of my motors out of alignment. It still hovers just fine. He seems to be complaining about a ROLL tendency upon lift off that he can counter with his radio but that get more aggressive with power. He says that on a flat surface when plugged in to GCS on the FLIGHT DATA tab, he cannot see a drift or angle. So the board thinks its level when it is level. Unless the arm is warped or damaged badly enough for it to be seen visually, I can't think of a way to mount a motor with a machined flat metal case directly onto a flat carbon plate without it coming out flat. On a ZMR all four arms mount on the same flat carbon plate, so all four arms are on the same plane. Now if the FC is on a parallel plane and calibrated on a flat surface parallel with the ground, then the physical build is correct. This leads me back to one of the four corners being somehow different than the others ( motor, ESC, Firmware, Neutral Setting, Prop, ect.). If his board is acting funny when it is mounted flat or parallel with all four motors being the same and the calibration was done correctly, I think it is safe to say that his board may be damaged. Maybe virtual roll will help, but it should not be needed if all other things are as stated above. Maybe he has a mismarked motor that is actually 2300Kv instead of 2700Kv. He could check that with a Tachometer.

[kevinybh]

hey guys,

wow!! thanks for all the advice given!
I'm still ought to try all the recommendations.

Cliff: I will try the virtual roll when I have time to do so. Been caught up with work lately and didnt have much time to spend on my quad. My TX trims are all at centre position, no trims at all.

Ray: yeah it is mounted parallel with my motors on 4 nylon spacers of the exact same height. And yes, I am using the new SunnySky 2700kV motors.

This issue only happens in Attitude mode. When I switch it to rate mode, it's perfectly fine and no rolls or whatsoever when I pump the throttle. Still yet to test it outside. Been only able to test it in my room. Didn't wanna fly it until it's perfectly tuned.

Maybe it's better if I take a video of the flight and post it here?

[TheOtherCliff]

Weak or different motor or ESC then and certainly must verify that ESC calibration is done correctly.

ESC calibration is where you (take props off) power the ESC up at full throttle, wait for the first beep, immediately reduce throttle to min, wait for beep, power off.  It sets a calibration inside the ESC to expect that length of pulses.

Details moved to wiki:
https://librepilot.atlassian.net/wiki/display/LPDOC/ESC+Calibration

[kevinybh]

wow this is a long step!!!
but i think i fixed it by taking your advice using virtual roll. I've set it to 3 and it seems like it's quite stable and flat now. 

The ESC calibration step, doesn't CC3D do it for you during the vehicle setup wizard step whereby u connect ur battery and press start and ur ESC starts beeping like crazy.

But according to your last post. could this be the answer for this issue that im facing for my other quad build?
https://www.facebook.com/kevinybh/videos/10153414624184495

[TheOtherCliff]

Generally speaking, even if the ESC calibration is incorrect (but not horribly so like "motors don't start till half throttle"), it should hover level and correctly without drift.  That is because the PID 'I' term handles it.  You can have a badly unbalanced quad still have it hover correctly.  Just like cruise control on your car can handle a hill.  Your quad may lean a little when taking off or hitting that gas really hard though.

For these acceleration problems, it can be caused by not timing your clicks during the ESC calibration correctly.  You might get 3 of 4 motors done correctly.  After you have done this a few times, you can tell by the beeping "calibration done" beeps.  You can tell that the beeps are all the same (good) or some are different (bad).