Thread Tools
Jul 22, 2015, 10:52 AM
Team AlienWarpSquad
Quote:
Originally Posted by Tom T
Guys, can I get you to load up my log and have a look at it. It flies pretty good, after some help tuning my PIDS over the weekend, but notice my traces are extremely jagged. All props are balanced and only showing about .02 on all four of them in the CF motor test tab. This was just a little flight here this morning in angle mode, but they certainly dont look nearly as smooth as some I've watched on here. Just using PID1 for now in my learning process.

Thanks
The Log shows fairly quite Gryo readings so no motor/prop/frames vibrations.
The 'P' & 'D' noise is typical without Boris' new Filters and is what these new filters are intended to clean up.
Were the logs with smoother P&D using the New filters? If so then that would be why they look smoother.
Sign up now
to remove ads between posts
Jul 22, 2015, 12:10 PM
Learning to crash
Tom T's Avatar
No waltr, Im not using boris's filters or anything. Just CF 1.9 .65 configurator. I was referring to other peoples bb traces that looked so much smoother than mine. Not sure if all of them Ive looked at were using the filters or not. Not exactly sure how to go about getting it all setup in Boris' hex and all. Tried his naze.hex the other day, and just after a liftoff it went into crazy hard wobbles and crashed. Thats when I had the issue of trying to reload the bootloader and everything from the other day. Got that all solved now and back in the air with my backup I had made.
Thank you for replying though.
Last edited by Tom T; Jul 22, 2015 at 12:36 PM.
Jul 22, 2015, 01:15 PM
Team AlienWarpSquad
Quote:
Originally Posted by Tom T
No waltr, Im not using boris's filters or anything. Just CF 1.9 .65 configurator. I was referring to other peoples bb traces that looked so much smoother than mine. Not sure if all of them Ive looked at were using the filters or not. Not exactly sure how to go about getting it all setup in Boris' hex and all. Tried his naze.hex the other day, and just after a liftoff it went into crazy hard wobbles and crashed. Thats when I had the issue of trying to reload the bootloader and everything from the other day. Got that all solved now and back in the air with my backup I had made.
Thank you for replying though.
Yep, my first flight after enabling Boris' filters went the same way, wobbled so bad it was totally unflyable. It took some time to tune but lower 'P's and higher 'D's got it flying much better than before. Lots of good discussion in the PID tuning from start to finish thread.
Jul 23, 2015, 08:50 PM
Registered User
Hello all, hoping someone can help me tune out some oscillations that occur when recovering from aggressive maneuvers.

I'm using Boris B's latest cleanflight patch with filters and luxfloat V2.

set dterm_cut_hz = 7
set pterm_cut_hz = 35
set gyro_cut_hz = 40
set gyro_lpf = 0


The best example of my issue starts at about 2:20.700, I do a cartwheel and by 2:23.5 as I'm attempting to recover I get crazy oscillations. What can I do to smooth these out?
Jul 23, 2015, 08:54 PM
Registered User
Quote:
Originally Posted by Phlarp
set gyro_lpf = 0
You want 256, which will disable gyro_lpf. I'm not sure what setting it to zero does. Zero is not a valid entry for gyro_lpf, so the code is picking some next-best valid value--possibly the default of 42. Before you do any more tuning, you have to get this right. It's possible that the gyro is being double-filtered, which is creating excess latency and making it harder for the PID controller to stabilize the copter.

Your overall values are somewhat low. Is this a particularly vibration-prone copter?
Jul 23, 2015, 09:13 PM
Registered User
Quote:
Originally Posted by joshuabardwell
You want 256, which will disable gyro_lpf. I'm not sure what setting it to zero does. Zero is not a valid entry for gyro_lpf, so the code is picking some next-best valid value--possibly the default of 42. Before you do any more tuning, you have to get this right. It's possible that the gyro is being double-filtered, which is creating excess latency and making it harder for the PID controller to stabilize the copter.

Your overall values are somewhat low. Is this a particularly vibration-prone copter?
I have applied gyro_lpf 256 but am still seeing the same things in the log, here is an updated log with similar behavior at 00:19.5

The airframe itself seems to be relatively vibration free, but I guess my perception must out of whack as I had to keep turning the dterm way down that low to get the motor traces to smooth out a bit.
Last edited by Phlarp; Jul 23, 2015 at 09:19 PM.
Jul 23, 2015, 09:19 PM
Registered User
Quote:
Originally Posted by Phlarp
I have applied gyro_lpf 256 but am still seeing the same things in the log, here is an updated log with similar behavior at 00:19.5
What are your PID values?
Jul 23, 2015, 09:21 PM
Registered User
Quote:
Originally Posted by joshuabardwell
What are your PID values?
see attached
Jul 23, 2015, 09:24 PM
Team AlienWarpSquad
Quote:
Originally Posted by Phlarp
I have applied gyro_lpf 256 but am still seeing the same things in the log, here is an updated log with similar behavior at 00:19.5

The airframe itself seems to be relatively vibration free, but I guess my perception must out of whack as I had to keep turning the dterm way down that low to get the motor traces to smooth out a bit.
I remember Boris saying gyro_lpf = 0 is the same as = 256.

dterm, pterm and gyro_cut all seem too low. try raising them.
then try increasing the PID 'D's
Jul 23, 2015, 09:27 PM
Registered User
I can't figure out why there is high-frequency noise in your D term, and even your gyro a little bit, with your cut_hz filters so low. Your roll D term especially has constant high-frequency noise.

Your pitch and roll D terms are a little higher in magnitude than your P terms. What happens if you reduce D gains to, say, 25 or maybe 30?

I can't figure out why your copter flips out with oscillations when you've done basically nothing at all with the sticks. Is this a super-overpowered rig? What's your hover throttle percent?
Jul 23, 2015, 09:36 PM
Registered User
Quote:
Originally Posted by joshuabardwell
I can't figure out why there is high-frequency noise in your D term, and even your gyro a little bit, with your cut_hz filters so low. Your roll D term especially has constant high-frequency noise.

Your pitch and roll D terms are a little higher in magnitude than your P terms. What happens if you reduce D gains to, say, 25 or maybe 30?

I can't figure out why your copter flips out with oscillations when you've done basically nothing at all with the sticks. Is this a super-overpowered rig? What's your hover throttle percent?
2208 1800kv airbot motors, 7030 CF props, KISS 18a and a 1300mah tattu 75C, so yes, relatively powerful. Hovers around 15 or 20% depending on how I fiddle with the RC settings

Overall I would say it flies and handles pretty well, with the exception of the wonky recoveries that make me almost crap myself.

I'm using a RMRC Dodo, so looptime is at ~600 and blackbox logging is set to 1/2

attached is a log of D gains higher at 48 and 50 and then lower with both at 25


Apologies for the short hovering logs, I am confined to the living room until morning.
Jul 23, 2015, 09:58 PM
Registered User
Here is a better log with more well defined stick inputs and reactions.

D gains at 25 again
Jul 23, 2015, 10:01 PM
Registered User
What happens if you raise looptime to 1000? There has been some suggestion that the additional filtering makes synchronization between the gyro loop and the output of the MCU more important.

How low can you push P, with D around 30, before the rig feels sloshy?
Jul 23, 2015, 10:18 PM
Registered User
Quote:
Originally Posted by joshuabardwell
What happens if you raise looptime to 1000? There has been some suggestion that the additional filtering makes synchronization between the gyro loop and the output of the MCU more important.

How low can you push P, with D around 30, before the rig feels sloshy?
looptime = 1000
set p_pitchf = 0.900
set i_pitchf = 0.520
set d_pitchf = 0.030
set p_rollf = 0.500
set i_rollf = 0.420
set d_rollf = 0.030
set p_yawf = 2.700
set i_yawf = 0.140
set d_yawf = 0.021


The jitters seem worse but overall it seems to oscillate less on abrupt stick inputs.

At 00:23.57 it did a really strange jump in midair without and stick inputs and I landed right away. I've noticed similar things in regular flight previously, any idea what that is?
Jul 23, 2015, 10:32 PM
Registered User
Quote:
Originally Posted by Phlarp
At 00:23.57 it did a really strange jump in midair without and stick inputs and I landed right away. I've noticed similar things in regular flight previously, any idea what that is?
Spontaneous twiches and jumps are usually excessive vibration.

That the problem still happens at "sloshy" P values means that it's definitely not excessive P gain.

I really am at a loss here... sorry.

The only other thought I have is that your cut_hz filters are really low. Did you start with them that low, or did you start at the suggested values of gyro=100, pterm=50, dterm=14 and work down? I really doubt that the soft filters are causing this problem, but I'm kind of grasping at straws here.


Quick Reply
Message:

Thread Tools