Originally Posted by e_lm_70

How is possible people don't know what they do flash in their kk2.0 board?

Did LazyZero add your hacked firmware in his tool/database?

If people find only from the firmware from official source (that is a post here from mr KK) these ambiguities are not possible

Tchuss

E_lm_70

Ps: let remember that your hacked firmware is a nice experiment, but it is nothing else then code done by mr kk, that is uncompleted and consider not stable by mr kk, for this reason it is only in the source code , commented out, but intentionally not part of the official firmware ... as mr kk said about, the problem is a yaw input in special attitudes in autolevel on, may cause undesired results

I agree and I have never wanted to take credit for any of the code itself. It's all in v1.2; I am just selectively uncommenting parts. I don't know why it's in the LazyZero flash tool now. I wouldn't recommend putting in there because it's experimental.

I have continued working on it a bit, though, to try to fix one big problem that some people mentioned. If anyone is interested, here's an update on the small change I made:

http://scolton.blogspot.com/2012/09/...0-modding.html

There is also new firmware, but I won't post it here or ask anyone to test it. If you want to try it, it's in that post and you can do so completely at your own risk.

The problem I was trying to address is that of the controller getting "stuck" with bad angle estimates due to yaw while at steep angles. It's one thing for the angles to get temporarily messed up due to yawing or wind or other disturbance. It's much worse if they get stuck that way. I found with the original modification, they can even get stuck before you arm the controller! Here's a demo of what I mean:

KK2.0 Firmware Mod: Angle Latching Fix (4 min 37 sec)

I've made a small change to fix that problem, as you see in the second part of the video. Some people who mentioned that the modified self-level code would cause their multirotor to pull in one direction no matter what were probably seeing this scenario. (Either that or bad sensor calibration.) This could be a fix.

But, like I said, it's totally experimental. If you're not willing to risk your frame and whatever is attached to it, please use the unmodified v1.2 firmware.

can anyone recommend a good set of parameters for F450 Quad ?

Originally Posted by scolton

I agree and I have never wanted to take credit for any of the code itself. It's all in v1.2; I am just selectively uncommenting parts. I don't know why it's in the LazyZero flash tool now. I wouldn't recommend putting in there because it's experimental.

I have continued working on it a bit, though, to try to fix one big problem that some people mentioned. If anyone is interested, here's an update on the small change I made:

http://scolton.blogspot.com/2012/09/...0-modding.html

There is also new firmware, but I won't post it here or ask anyone to test it. If you want to try it, it's in that post and you can do so completely at your own risk.

The problem I was trying to address is that of the controller getting "stuck" with bad angle estimates due to yaw while at steep angles. It's one thing for the angles to get temporarily messed up due to yawing or wind or other disturbance. It's much worse if they get stuck that way. I found with the original modification, they can even get stuck before you arm the controller! Here's a demo of what I mean:

http://www.youtube.com/watch?v=xCLu202ZEzM

I've made a small change to fix that problem, as you see in the second part of the video. Some people who mentioned that the modified self-level code would cause their multirotor to pull in one direction no matter what were probably seeing this scenario. (Either that or bad sensor calibration.) This could be a fix.

But, like I said, it's totally experimental. If you're not willing to risk your frame and whatever is attached to it, please use the unmodified v1.2 firmware.

IMO I think your doing good work keep it going. I will flash one of my boards a test the new version

previous quote....
when i move rudder stick left or right my quad goes left or right BUT ALSO GOES UP i can not make pirouette without quad goes 1-2M UP

i write problem here before one month and answer is been mostly maybe its defective board

i got it another one and still i have same situation so what i can do

my setting is

roll/pitch P 55 I 45
yaw P 40 I 40

firmware is 1.2

Tighten the tension on your throttle stick. Sounds like when you use rudder, you are also pushing up on the throttle a little.

from my reading as a newbee and not much flying experience could another reason be that to yaw or spin the model some of the motors have to speed up or slow down..so that the torque rotates the model... and surely any changes in motor speed will also result in some slight change in lift.
or is my understanding of that incorrect as well..
What ever is the reason, the answer will be in the constant use of the throttle to control the height, just like when you tilt to go forward, you need more throttle to maintain height, as some of the previous lift is now pushing the model forward..
That is one thing I am learning, these things fly on throttle, being a plane pilot I tend to chop the throttle when I get in a poor situation... but doing that only makes my quad crash...and thats not so good...

Quote:
Originally Posted by zokizo
here go again!

when i move rudder stick left or right my quad goes left or right BUT ALSO GOES UP i can not make piruette withot quad goes 1-2M UP

i write problem here before one month and answer is been mostly maybe its defective board

i got it another one and still i have same situation so what i can do

my setting is

roll/pitch P 55 I 45
yaw P 40 I 40

firmware is 1.2
Tighten the tension on your throttle stick. Sounds like when you use rudder, you are also pushing up on the throttle a little.


i have experience with helicopters
i am pretty sure that i dont give throttle when i move rudder stick

this problem drive me crazy
last two months i try to make it work

i read a lots of manuals, watch a lot videos but still have that problem

thanks guys for try helping me

Try changing the P and I gains for yaw... start by halving your I on yaw ... they do influence the reaction on throttle and "can" cause an upward bump (had same experience with and openpilot board)

Originally Posted by scolton

I agree and I have never wanted to take credit for any of the code itself. It's all in v1.2; I am just selectively uncommenting parts. I don't know why it's in the LazyZero flash tool now. I wouldn't recommend putting in there because it's experimental.

I have continued working on it a bit, though, to try to fix one big problem that some people mentioned. If anyone is interested, here's an update on the small change I made:

http://scolton.blogspot.com/2012/09/...0-modding.html

There is also new firmware, but I won't post it here or ask anyone to test it. If you want to try it, it's in that post and you can do so completely at your own risk.

The problem I was trying to address is that of the controller getting "stuck" with bad angle estimates due to yaw while at steep angles. It's one thing for the angles to get temporarily messed up due to yawing or wind or other disturbance. It's much worse if they get stuck that way. I found with the original modification, they can even get stuck before you arm the controller! Here's a demo of what I mean:

KK2.0 Firmware Mod: Angle Latching Fix (4 min 37 sec)



I've made a small change to fix that problem, as you see in the second part of the video. Some people who mentioned that the modified self-level code would cause their multirotor to pull in one direction no matter what were probably seeing this scenario. (Either that or bad sensor calibration.) This could be a fix.

But, like I said, it's totally experimental. If you're not willing to risk your frame and whatever is attached to it, please use the unmodified v1.2 firmware.

Wow this is great how did you fix the problematic complementary filter...????
I thought yawing when your copter is banking steeply at some angle will cause bad angle estimates(i.e when your copter is armed and flying) because the accelerometer values will be rejected by the controller as the gyros dont sense reversal of angles on the roll axis.... but how is the controller latching to an angle even before it has been armed....this is very alarming ..can you just post what modifications you made to the code to solve this problem.....

Hi,its great to know that maybe a fix to the auto level is not to far away,your's Marty.

I wish the Stock FW was available, may have to buy a new kk2 and leave it stock. The FW has to be somewhere!

zokizo but what about adding expo to your yaw?
I think mine is 15-20

Originally Posted by Akcopter

Wow this is great how did you fix the problematic complementary filter...????
I thought yawing when your copter is banking steeply at some angle will cause bad angle estimates(i.e when your copter is armed and flying) because the accelerometer values will be rejected by the controller as the gyros dont sense reversal of angles on the roll axis.... but how is the controller latching to an angle even before it has been armed....this is very alarming ..can you just post what modifications you made to the code to solve this problem.....

Originally, I enabled the complementary filter code that is commented out in v1.2 and replaced the accelerometer-only angle with gyro+accelerometer angle from the filter in the self-level P controller. But, there's a conditional that says: "If filtered angle is higher than 26º (I think), ignore the accelerometer." I changed the conditional to: "If accelerometer angle is higher than 45º, ignore the accelerometer."

It doesn't solve the problem of yaw while steeply banked causing a bad angle estimate. If you yaw fast enough, before the accelerometer has a chance to correct it, it may still cause strange results. Slow, shallow banked turns seem okay to me, but your results may vary.

@Scolton
Very nice work mate. Looking forward for next stable update.

@scolton flashed DJI 330 latest mod very nice working very well

Originally Posted by Eastcoast78

I wish the Stock FW was available, may have to buy a new kk2 and level it stock. It has to be somewhere! lmao

zokizo i am to into Hili's!
this may of been said, or maybe you do, but what about adding expo to your yaw? as you know that well soften it up. My friend was running 0 expo and was giving thr input as he gave rudd input.
I think mine is 15-20

i didnt try to add expo on rudder

i noticed if i make slow pirouette quad does NOT go up but if i wanna make fast pirouette quad goes UP,
like i said before same experience with two different kk2 boards

Originally Posted by zokizo

i didnt try to add expo on rudder

i noticed if i make slow pirouette quad does NOT go up but if i wanna make fast pirouette quad goes UP,
like i said before same experience with two different kk2 boards

Are all your motors perfectly square to the horizontal and have you calibrated the throttle on all escs at the same time?

Originally Posted by zokizo

i didnt try to add expo on rudder

i noticed if i make slow pirouette quad does NOT go up but if i wanna make fast pirouette quad goes UP,
like i said before same experience with two different kk2 boards

An explanation I've seen before: Thrust goes as the square of rotor speed, so if all four rotors are at 4,000rpm, then two go up to 5,000rpm and two go down to 3,000rpm, the net thrust will go up. (2*3^2 + 2*5^2 > 4*4^2). It gets worse as the speed difference gets higher. This is assuming the KK2 mixer is linear, which I think it is.