Blackbox Log Video Responses - RC Groups
Shop our Airplanes Products Drone Products Sales
Thread Tools
Aug 19, 2015, 03:46 PM
Registered User
Discussion

Blackbox Log Video Responses


If you are interested in donating to support my efforts, you can use this link to Paypal me: paypal.me/joshuabardwell
I also have a Patreon here: https://www.patreon.com/thedroneracingengineer?ty=h

Any contribution is welcome.

---

In order to help more people learn to analyze Blackbox data, I'm going to start a thread where I (and anybody else who wants to) posts video analyses of people's logs. Here are some examples of what I'm talking about:

Blackbox trace analysis: ingleburn (10 min 27 sec)


Talking through a yaw tuning issue with Blackbox (11 min 25 sec)


There is already a thread on Blackbox log analysis, and I don't mean to take any attention away from that, but I think that video analysis may be more helpful, and I think it may be useful to get this all in one place. If you need help getting Blackbox working, the place to ask is the official Blackbox thread, not here.

Here are the guidelines:
  • Post your log file to this thread and I will take a stab at analyzing it. You can attach the log file directly to the post, although you may need to change the extension or put it in a zip file. You can also link to a Google Drive, PasteBin, Dropbox, etc... file.
  • If you have an FPV video, it will help a lot because it will show what the copter is doing more clearly than looking at traces. There are lots of times when a Blackbox trace may look questionable, but it's obvious from the video that the copter flies just fine. If it ain't broke, don't fix it! If you want to link to a YouTube video, I think that will work because there are sites that allow YouTube videos to be downloaded. DropBox or something like that would be more convenient, though.
  • Posts asking for general suggestions are fine, but posts that are trying to solve a specific issue are more interesting and informative. If you can give some impression of what you don't like about your current tune, that'd be nice.
  • Log files containing just a hover are not very useful, because the PID controller is not actually doing very much. For best results, include a log file that has forward flight, punchouts, drops, hard turns, flips, and rolls.
  • For best results, include a full config dump from the CLI, which will allow me to see your current settings. PLEASE DO NOT PASTE THE CONFIG DUMP INTO THE POST as it will fill up the whole screen and be annoying. Attach it as a file or provide a Pastebin link or something similar. Or use the CODE tags in the editor to at least prevent the CLI dump from taking up too much space.
  • Before you post here asking for help with your Blackbox logs, please work your way through this thread. Blackbox tuning is most useful for what I call the "last 15%" of tuning. The "first 85%" of tuning can be done largely by sight and feel. Blackbox tuning is also essential for setting the cut_hz filters correctly.

Lastly, I do have a job and a family, so depending on how busy this thread gets, I may not be able to respond to every single post, but I will do what I can. I hope that this thread can become a resource for people who are trying to learn how to tune, and also help me get a wider perspective on what kind of performance people are getting out of their copter.

Anyone else who wants to contribute is welcome to contribute to the thread. I would like to suggest that other contributors also use the video-response format, but I understand that's not everybody's cup of tea, so text and picture discussion is also fine.

I am not the best tuner out there, and there is a lot that I don't know, but I seem to know more than some people, and I'm willing to put the time in to try to explain what I think, so I hope that I can contribute something. There are a lot of people who don't really know where to begin with Blackbox. Anybody who catches me saying something wrong is more than welcome to correct me.
Last edited by joshuabardwell; May 13, 2016 at 09:11 PM.
Sign up now
to remove ads between posts
Aug 19, 2015, 04:57 PM
Registered User

gyro_cut_hz with lpf 188


Hi,

Short test hovering with (it's dark) 3 different gyro_cut_hz settings with gyro_lpf = 188:
1. 0
2. 80
3. 60

3rd one is very smooth on gyro roll/pitch but yaw is still a little noisy. I could try going even lower but where is the limit?

I tried gyro_lpf = 256 and quad flies perfectly in hard fpv acro but with a lot of vibrations on video.

Setup:
BO 6" with MN2206 2000kV on HQ6045 (balancing helps), 4S. I could upload video of the vibrations later (short test flight).

Edit:
set gyro_lpf = 188
set gyro_cut_hz = 0
set pterm_cut_hz = 40
set dterm_cut_hz = 15
vibrations (1 min 25 sec)


set gyro_lpf = 188
set gyro_cut_hz = 0
set pterm_cut_hz = 40
set dterm_cut_hz = 10
vibrations (0 min 45 sec)
Last edited by icdark; Aug 19, 2015 at 05:24 PM. Reason: adding video links
Aug 19, 2015, 06:15 PM
Registered User

Slightly high P gain


Blackbox 2015-08-19 - Slightly high P gain (10 min 35 sec)


Raw flight video is available here for anybody who wants to play with it in Blackbox. Use a tool like http://keepvid.com to download it.
Aug 19, 2015, 06:33 PM
Registered User
Quote:
Originally Posted by icdark
Hi,
Short test hovering with (it's dark) 3 different gyro_cut_hz settings with gyro_lpf = 188:
1. 0
2. 80
3. 60
Your copter is, over all, very noise free. Yaw is the noisiest axis, and it is far noisier than pitch and roll, but the overall noise level is very low, even on the yaw axis. I don't think you need additional filtering, and frankly, you could even run gyro_cut_hz=0 if you wanted to, and if you found that flew better for you. Removing gyro_cut_hz will slightly improve the latency of the control loop. It's a subjective decision whether this offsets the increased noise.

Because the log file was only during hover, there really isn't much else to say. For best results, the submitted log file needs to contain some more dynamic flight, like fast forward flight, hard turns, flips, and rolls. That is, assuming that's the type of flying you're going for. If you're trying to build a super-stable camera platform, that's another story. (I'll add this recommendation to the first post for future reference.)

I see a tiny bit of P-term oscillation in your videos, on hard turns. I would suggest reducing roll P perhaps two steps, and pitch P perhaps 1 step, where a "step" is 0.1 for PIDCs except Luxfloat, and 0.01 for Luxfloat, if you are using Boris's new GUI version that gives you three digits of precision. If you are using stock Luxfloat, then reduce roll P by 0.1 and leave pitch P alone.

For the type of vibration that you're looking for, where in forward flight, you have a sort of low-level medium-frequency instability, the most likely cause is noise in the system. Since your gyros are so quiet, the leading candidate would be excess D gain amplifying the small amount of noise that is there. If that was the case, the solution would be to reduce D gain.

Blackbox Log Analysis: icdark 8/19 (6 min 58 sec)
Aug 19, 2015, 06:51 PM
Registered User
Boris B's Avatar
Quote:
Originally Posted by joshuabardwell
https://www.youtube.com/watch?v=7q9NNVMvE9c

Raw flight video is available here for anybody who wants to play with it in Blackbox. Use a tool like http://keepvid.com to download it.
The short oscillations you see are only during prop wash scenarios where you pretty much fly "backwards" and use throttle to get yourself out. It's like a car with slipping tyres.
Your quad looks pretty well tuned actually and many of pilots with agressive style suffer from this on different setups. It just may be the nature of our quads. What is more important is the time till it settles. In your case it's gone before you notice it.
Aug 19, 2015, 06:56 PM
Registered User
Boris B's Avatar
A good example of it. This new setup is pretty much very sharply tuned. It definately never oscillates, but when I fly backwards and I want to throttle my self out you will hear a very short ringing. I pretty much think that's the stress on the frame/props being root cause of that. I totally don't mind. The footage is smooth....the motors are cool and it flies like a dream.

FVT Littlebee 20A Test (1 min 28 sec)
Aug 19, 2015, 07:02 PM
Build & Crash
Thanks for the videos Joshua, very instructive. I'm really looking forward to your discussion of a 'perfectly' tuned (racing) quad. I do miss some sort of reference of what's possible...

Boris, you must have some logs
Aug 19, 2015, 07:06 PM
Registered User
Boris B's Avatar
Quote:
Originally Posted by SplinterNL
Thanks for the videos Joshua, very instructive. I'm really looking forward to your discussion of a 'perfectly' tuned (racing) quad. I do miss some sort of reference of what's possible...

Boris, you must have some logs
The flight above? Unfortunately no blackbox attached on this one. Fresh build and new setup on this one as well.
Aug 19, 2015, 07:31 PM
Registered User
Great idea Joshua, I have a new mobius landing next week so can get some video.

Will wait till that point as I feel like im 95% there with my tune already.
Aug 19, 2015, 07:36 PM
Registered User
Quote:
Originally Posted by Boris B
The short oscillations you see are only during prop wash scenarios where you pretty much fly "backwards" and use throttle to get yourself out. It's like a car with slipping tyres.
Your quad looks pretty well tuned actually and many of pilots with agressive style suffer from this on different setups. It just may be the nature of our quads. What is more important is the time till it settles. In your case it's gone before you notice it.
Thanks for the insight, Boris. And thanks for the compliment on the tune.
Aug 19, 2015, 09:53 PM
Registered User
Thanks so much Joshuabarwell for taking time doing this. I enjoy your interpretations and hopefully I'm getting a little education from them. Still trying to get my head wrapped around tuning but the videos you do are awesome. You explain things very well.
Aug 19, 2015, 10:32 PM
Registered User
Quote:
Originally Posted by joshuabardwell
https://www.youtube.com/watch?v=C7uKssxDPyo

Anyone else who wants to contribute is welcome to contribute to the thread. I would like to suggest that other contributors also use the video-response format, but I understand that's not everybody's cup of tea, so text and picture discussion is also fine.
It's worth noting that Ingleburn was using PID 2 Lux on the official Cleanglight v1.9. (He posted in the Blackbox thread his PIDs and dump so I've reattached his dump to this post for reference).

Official CF 1.9 is missing the many improvements that Boris has made which make he D term far more usable. I think this probably explains the domination of the D term despite it only being set to 20?

Ingleburn, it would be great if you could repeat the hover test again with Betaflight with filters active, using same PIDs so we can see just how much smoother D is now on your setup.
Last edited by RC Maca; Aug 20, 2015 at 12:32 AM.
Aug 20, 2015, 01:13 AM
Registered User
Original post here.

This copter is very noise free, most likely because you're using gyro_lpf=42, which is standard on stock CF 1.9. Your D gains are far too high for CF 1.9 without Boris's filters. The D terms are amplifying system noise and causing roughness in your motor outputs, which results in wasted power and additional heat buildup. Without low-pass filtering, the D term in CF cannot be turned up high enough to fully do its job without other negative effects. My suggestion is to turn D gain down to 20 on all three axes, then raise it up again as needed, possibly as high as 25 or 30, but probably not any higher than that. Assess the D traces and the motor traces as you raise the D gains to see how high you can go. P gains can probably be raised a bit.

Blackbox Log Analysis: layer 8/20 (13 min 49 sec)
Last edited by joshuabardwell; Aug 20, 2015 at 01:22 AM.
Aug 20, 2015, 02:26 AM
EU drones
Genivos's Avatar
Hello Joshua,

Great to see this thread. Myself I'm also trying to analyze the blackbox logs. If you find time to look at my log and give comment on it, I'd be very thankful.
As far as I can judge my own logs I don't see much wrong. Not looking solely only at the log but also how it flies. If there's some improvement to make, just curious to know what it could be.

Video from this log with details about the quad setup. Please don't mind the sound and also better turn down volume, I forget to remove it (again):
LOG00015 (4 min 57 sec)


I was trying lower dterm_cut_hz values for testing purposes but unfortunately for some strange reason my blackbox logger decided not to log correct anymore. Don't know why, last Saturday I had totally no issues. In this log you can see as well it missed some parts unfortunately. Only thing I could think of and what I overlooked before installing the SD card, was to use the SD card formatting tool. This I did last night after the flights and it will show next time if it will log correct or not (note that my CLI settings were exactly the same as Saturday when it was working correct, so that can't be it was my conclusion).
Aug 20, 2015, 03:45 AM
Registered User
Quote:
Originally Posted by joshuabardwell
Because the log file was only during hover, there really isn't much else to say. For best results, the submitted log file needs to contain some more dynamic flight, like fast forward flight, hard turns, flips, and rolls. That is, assuming that's the type of flying you're going for. If you're trying to build a super-stable camera platform, that's another story. (I'll add this recommendation to the first post for future reference.)
https://www.youtube.com/watch?v=wt4sH3G3zrU
Thanks joshuabardwell! I'm gonna configure pids on gyro_cut_hz 0 and 60 today. I'll record the final versions with some acro. I can make this quad quite smooth to the point the only vibrations I can see is some very small yaw movement (twitches). I really have a problem with the yaw :-)

Cheers