SMALL - espritmodel.com SMALL - Telemetry SMALL - Radio
Reply
Thread Tools
Old Jul 03, 2012, 12:06 PM
Registered User
United States, FL, Miami
Joined Jan 2012
194 Posts
Help!
I don't know what else to do

Problem (same two frames):
  • At stock PID settings, I get a fast wobble and the copter is hard to control...it tends to shoot up and down quickly also.

Attempted solutions:
  • Pitch/roll P settings from 1.5 to 6.5 with relative changes in ID
  • The two totally different frames, all brand new parts
  • Balancing props
  • Using Naze32 housing with and without rubber pads inside
  • Calibrating ESCs several times (although they all work perfectly together)
  • Calibrating acc, acro gyro, mag before each attempt
  • Different baseflight.hex versions

Comments/Specs:
The only way I have been able to find to get rid of the wobble is to cut the roll/pitch PID settings pretty much in half. I understand these values are wrong and too low (makes it a bit sensitive). But I don't know what else to do.

I'm really at a loss as to what I'm doing wrong. I've tried two different frames now with the only common factor being the FC.

First frame:
  • Naze32
  • 220mm carbon fiber
  • 1510 2200kv motors
  • Turnigy 10A ESC
  • HK power distribution board
  • Orange mini 4 ch Rx
  • DX8 Tx

Second frame:
  • Same Naze32
  • Kindjal style G10/Acrylic frame
  • KDA 20-50S 1088kv motors
  • HK 10A blueseries ESC
  • a different HK power distribution board
  • Orange 6 ch Rx
  • JR9303 Tx

Here are a couple of clips, the first from the 220mm frame and the second from the larger kindjal-style frame.
(Note: the 220mm frame oscillates faster than the kindjal so it's harder to see in the video)

Kindjal-style frame (at stock settings - 4P and then 1.8P):
http://www.youtube.com/watch?v=2l25nwojNOA

220mm Frame (at stock 4P):
http://www.youtube.com/watch?v=Ut0r7C6i3BE&feature=plcp
DaMaDo is offline Find More Posts by DaMaDo
Reply With Quote
Sign up now
to remove ads between posts
Old Jul 03, 2012, 12:12 PM
Emress
sirbow2's Avatar
United States, SD, Rapid City
Joined Nov 2011
1,936 Posts
naze32 uses the MWC code, so a MWC FC would behave similarly with the same PIDs. PIDs were annoying to me at first but now i can manage fairly well.

Quote:
Pitch/roll P settings from 1.5 to 6.5
why would you increase P? P makes oscillations worse. that would make it worse even if you raised the other values to compensate. try going back to the default PIDs, decrease P a bit and then increase I and D. naze32 uses the MWC gui so the defaults should be like P:4, I:.030,D:20?
sirbow2 is offline Find More Posts by sirbow2
Last edited by sirbow2; Jul 03, 2012 at 12:18 PM.
Reply With Quote
Old Jul 03, 2012, 12:18 PM
Registered User
United States, FL, Miami
Joined Jan 2012
194 Posts
Default P is 4 on the baseflight.hex for the naze.

It wobbles down to 1.8. Only there is it completely wobble-free, but of course that's too low for good control.

After going so low I only tried going up because I just wanted to try anything.
DaMaDo is offline Find More Posts by DaMaDo
Reply With Quote
Old Jul 03, 2012, 12:20 PM
Emress
sirbow2's Avatar
United States, SD, Rapid City
Joined Nov 2011
1,936 Posts
try increasing I and D by .01 and 10 respectively from defaults and then see how low P will go
sirbow2 is offline Find More Posts by sirbow2
Reply With Quote
Old Jul 03, 2012, 12:35 PM
Registered User
United States, FL, Miami
Joined Jan 2012
194 Posts
Many thanks for your help and time in responding sirbow2. I greatly appreciate it.

Changing I to .040 and D to 33, it only slightly wobbles when it's at P = 3 now, 2.5 it gets almost perfect.

I guess I should have added in my OP that I had it fundamentally ingrained in my head ( not sure why or from where) that PID all had to go up or down together but not opposite directions. I guess I was terribly wrong there.

At I = 0.050, D = 43 it's good at 3.

Is it ok to keep going up on I and D to get P to the target goal of 4?

Or is this undesired - similar to going down to 1.8 P?

Since I'm just hovering over my bed, I thought the 1.8P was great until I was told it was wrong. I'm sure I would realize it if I flew outside in more space. So this is why I'm asking if going up too much on I or D would be wrong too.

Actually nm, nothing will make it stable at 4 it seems.

Yeah P=3 is the limit with only slight wobble in hover with I >=0.045 and D >=43
DaMaDo is offline Find More Posts by DaMaDo
Last edited by DaMaDo; Jul 03, 2012 at 01:19 PM.
Reply With Quote
Old Jul 03, 2012, 01:43 PM
Emress
sirbow2's Avatar
United States, SD, Rapid City
Joined Nov 2011
1,936 Posts
yes, too much I or D could be bad too. its like a balancing act lol. i think if you increase I and decrease D you should be able to increase P, too.
This is pretty much what you need to know:

"Increasing value for P:
It will become more solid/stable until P is too high where it starts to oscillate and loose control
You will notice a very strong resistive force to any attempts to move the MultiRotor

Decreasing value for P:
It will start to drift in control until P is too low when it becomes very unstable.
Will be less resistive to any attempts to change orientation


Increasing value for I:
Increase the ability to hold overall initial position and reduce drift, but also increase the delay in returning to initial position
Will also decrease the importance of P.


Decreasing value for I:
Will improve reaction to changes, but increase drift and reduce ability to hold position
Will also increase the importance of P.

Increasing value for D: (remember, that means a LOWER number as it is a negative value)
Improves the speed at which deviations are recovered
With fast recovery speed comes a higher probability of overshooting and oscillations
Will also increase the effect of P

Decreasing value for D: (remember, that means a HIGHER number as it is a negative value - i.e. further from zero)
Reduces the oscillations when returning any deviations to their initial position
Recovery to initial position becomes slower
Will also decrease the effect of P"

links about PID:
http://www.rcgroups.com/forums/showthread.php?t=1375728 (my favorite :P)
http://www.rcgroups.com/forums/showt...&#post17216625
http://www.rcgroups.com/forums/showp...postcount=7460
http://www.rcgroups.com/forums/showt...5#post17311692

oh and thanks to your post, i now know what OP means because of the way you used it in the setnence i never knew i a was too lazy to find out-- it means original post?

edit: oops i said that wrong, fixed
sirbow2 is offline Find More Posts by sirbow2
Last edited by sirbow2; Jul 03, 2012 at 03:23 PM.
Reply With Quote
Old Jul 03, 2012, 03:06 PM
Atomic Playboy
United States, MN, Inver Grove Heights
Joined Jun 2012
218 Posts
My understanding is that if I is too *low* it causes drift, and too high causes slower oscillations. You should decrease I until drift is unacceptable, and then raise it up.

Derivative (D) is the resistance to predicted delta from a setpoint. As you increase D, you should gain stability, but it also reduces the weight of P. As for being sluggish... I don't think it matters on the MultiWii platform because you can set the "Rate" which is 0 by default. This is a linear decay of your PID's based on stick movement. You can have super stability at no or low stick, and at higher stick it will become super responsive.

I have not played with my D values yet. From the reading I've done, D should be as high as you can get it.
signal15 is offline Find More Posts by signal15
Reply With Quote
Old Jul 03, 2012, 05:55 PM
Registered User
DeSoto MO
Joined May 2003
580 Posts
For me setting I and D to zero and then changing only P until I get a fast wobbles in Forward Flight. Then I back P down .2. Once I have P set I increase the I value until the copter will hold its angle and not try to self correct when I push forward on the pitch or roll stick. Once P and I are set, I leave D alone since it does nothing but cause me problems. D stays at zero.
Nick_P is offline Find More Posts by Nick_P
Reply With Quote
Old Jul 03, 2012, 07:40 PM
Atomic Playboy
United States, MN, Inver Grove Heights
Joined Jun 2012
218 Posts
Quote:
Originally Posted by Nick_P View Post
For me setting I and D to zero and then changing only P until I get a fast wobbles in Forward Flight. Then I back P down .2. Once I have P set I increase the I value until the copter will hold its angle and not try to self correct when I push forward on the pitch or roll stick. Once P and I are set, I leave D alone since it does nothing but cause me problems. D stays at zero.
While that works, getting D tuned properly would likely greatly improve the stability of the platform. D is going to give you damping of oscillations caused by momentum. Get your copter up in the air and give it a quick full stick to one side and let go. When it stops, you're going to see some oscillation that can last anywhere from a split second to a couple of seconds depending on the platform and various factors. Now increase D and try it again. Eventually, these oscillations will start to go away the higher you put D. Keep in mind that a side effect of increasing D puts less weight on the P value, so you may have to increase P more.

The result is a super stable platform, and with the I factor set appropriately, little to no drift over time. The downside is slow response times with the high P and D values. With MultiWii, you can overcome this by setting the Rate above zero so the PID values are lowered the farther you push the stick. When you snap the stick back to the middle, your PID values are high again, giving you super stability.
signal15 is offline Find More Posts by signal15
Reply With Quote
Old Jul 03, 2012, 07:43 PM
Registered User
United States, FL, Miami
Joined Jan 2012
194 Posts
Quote:
Originally Posted by sirbow2 View Post
oh and thanks to your post, i now know what OP means because of the way you used it in the setnence i never knew i a was too lazy to find out-- it means original post?

edit: oops i said that wrong, fixed
Hah, glad I could help out too! I did mean it as original post, although it can be also used to refer to the person making the original post (the original poster).

I thank you guys for helping me out with this. I'm usually quick on the uptake with these types of things but for some reason these setups have been giving me trouble.

sirbow2, you've helped me understand the meaning of the values better than I thought I did.

I tried Nick_P's idea of setting I and D to 0 and started at 1.8P. It was stable, but too sensitive so I started raising P until it started wobbling at 2.3. I then raised the values of D, which in turn let me raise P. Eventually I was able to get to 3.4P 0.005I and 25D with just a little wobble when not moving sideways.

I could go 3.5P and 27D but then it starts feeling a tad too sluggish...needs a lot of control input to change directions. Granted this might be better for flying outside, but inside my room I need it to react quickly before I hit a wall. I'll have to test that out more once I can take it out somewhere.

In terms of drift I don't know what's normal. I don't expect it to hover as if it was glued to a pole on the ground, but with small constant corrections I can keep it in a one foot radius. If I don't touch the sticks it starts slowly going in some direction. I don't have the space to see how long it takes to correct yet. If I raise I, it will wobble again so I'd have to increase D (I mean the actual value in the box) and make it more sluggish or decrease P and make it more sensitive. I'll try to see if I can find a better balance.

Hopefully the BT module I ordered won't take too long to get here to make this adjustment process easier...right now I have to connect it to my PC each time.

All in all I'd say this has been a great improvement thanks to you guys and I really appreciate your help.
DaMaDo is offline Find More Posts by DaMaDo
Reply With Quote
Old Jul 03, 2012, 08:43 PM
DiaLFonZo - UAV/Drone
Dialfonzo's Avatar
Canada, QC
Joined Oct 2007
6,868 Posts
Small quad are much harder to fly.
I did a couple .. you need to eliminate ALL vibrations prior to try to tune anything.
Dialfonzo is offline Find More Posts by Dialfonzo
RCG Plus Member
Old Jul 03, 2012, 09:24 PM
Registered User
United States, FL, Miami
Joined Jan 2012
194 Posts
Yeah, I balanced the props. Might try using dampers on the FC.

A couple of weeks ago I was thinking of buying some kind of vibration sensor.

Or I wonder if it's possible to do with my oscilloscope and a mic placed on the frame. Not sure where I'd go from there though.

I think I would have to figure out the resonant frequencies the FC doesn't like then compare it to the frequencies generated by the frame at various RPMs.

I haven't taken those classes yet.

I'll hopefully flash my HK ESCs next week so that may help a bit.
DaMaDo is offline Find More Posts by DaMaDo
Reply With Quote
Old Jul 03, 2012, 11:38 PM
Atomic Playboy
United States, MN, Inver Grove Heights
Joined Jun 2012
218 Posts
Quote:
Originally Posted by DaMaDo View Post
A couple of weeks ago I was thinking of buying some kind of vibration sensor.
There are apps for iPhone and android that will use the accelerometer to measure vibration and graph it. Tie the copter down, use throttle hold to get repeatable throttle position, and set the phone on top somewhere or on each arm.
signal15 is offline Find More Posts by signal15
Reply With Quote
Old Jul 04, 2012, 10:42 PM
Registered User
United States, FL, Miami
Joined Jan 2012
194 Posts
Quote:
Originally Posted by signal15 View Post
There are apps for iPhone and android that will use the accelerometer to measure vibration and graph it. Tie the copter down, use throttle hold to get repeatable throttle position, and set the phone on top somewhere or on each arm.
That's a good idea, I downloaded Vibrometer by Smart Tools for android, strapped it to the top of the kindjal and hovered above my bed. Screenshot of the result attached. Looking at it while hovering, I saw mostly 5.8-6.2 while it was wobbling a little. The lowed reported mean must be due to the amount of time it moved around which was more stable (no wobble while moving).

This might be useful for relative comparisons after adjustments to see if the max and mean went down. Although I'm sure there are other variables thrown in each time I put the phone on, but relative ballpark might help.
DaMaDo is offline Find More Posts by DaMaDo
Reply With Quote
Old Jul 04, 2012, 10:49 PM
Registered User
DeSoto MO
Joined May 2003
580 Posts
I use this program to balance motors. The number represents magnitude and not frequency of the vibration wave. So while you are flying it is picking up the large magnitude movements of you hovering the quad along with motor vibes.

I suggest strapping the quad down and mounting the phone onto the quad. Run the motors up to half throttle and monitor the values. This will give you a more realistic idea of the vibes you are getting.
Nick_P is offline Find More Posts by Nick_P
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Help! Where do I start to search for a problem I don't know if exist? Denominator Multirotor Talk 3 Apr 29, 2012 08:24 PM