(21-Mar-2021, 09:30 AM)MelBourbon Wrote: So, I did a analysis of my short blackbox and filtering in spectral analyzer looks pretty good, similar to the one from L0stB1t. Unfortunately my flight was with another PID profile which I wanted to check (found on youtube) - see below and attached. 

It's from this video https://youtu.be/s7-6eIUSrHI?t=264 right? These PIDs are for high-KV F60 motors and low-pitch propellers, so I don't think the PD Balance is quite right for the low-KV Toka motors with higher-pitch propellers. The Gain is fine at 1.4, won't make a huge difference. I was surprised to see that this guy uses zero FeedForward though, you normally need quite some FeedForward to get immediate stick response. That being said, his flight is absolutely awesome and butter smooth!

(21-Mar-2021, 09:30 AM)MelBourbon Wrote: But what I noticed is that I have wavy lines of my PID D on roll and pitch in cases where not much stick movement was done (see below). What does this mean and has this any effect. I did not noticed something. The gyro data fits very well with the set point.

I see you've enabled the Expo button in the Blackbox Explorer, which exaggerates small movements. The D waves you see are actually really small in reality so probably not something to worry about. I personally don't use the Expo and Smoothing buttons in Blackbox Explorer a lot, they don't really reflect reality. Maybe they are more useful to finetune the I Term, which is subtle and slow.

(22-Mar-2021, 09:50 PM)L0stB1t Wrote: It's from this video https://youtu.be/s7-6eIUSrHI?t=264 right? These PIDs are for high-KV F60 motors and low-pitch propellers, so I don't think the PD Balance is quite right for the low-KV Toka motors with higher-pitch propellers. The Gain is fine at 1.4, won't make a huge difference. I was surprised to see that this guy uses zero FeedForward though, you normally need quite some FeedForward to get immediate stick response. That being said, his flight is absolutely awesome and butter smooth!

Yes, this I meant. I wanted to try as I also prefer a more "smooth" stickfeel. But I noticed that I only used the slider to bring this down and he changed FF directly to 0. So will try again. But also hope I can test flight your profile as I crashed before I was able to change the profile last time. But everything is ready again.

Thanks for clarifying the question with the wavy signal. And the filter settings look perfect.

@L0stB1t what is your "motor timing" set to in BLHeli? 
If it's set to default "Auto", could you perhaps set it to (commonly recommended) 16deg for a test flight?
Or maybe you've been playing around this setting for these motors already and got some thoughts?

Yup I set timing to auto. For the PWM frequency I tried 48 kHz but that leads to more wobbles at low throttle. Last flight I tried BLHeli 32.8 with a variable PWM between 24 and 48 kHz, seems to work ok, but not sure it's any better than just 24 kHz.

I'll play with the timing next time I get a chance to fly, but I'm renovating my house and have zero free time now

Rampup Power may also be an interesting parameter, KababFPV was playing with that in his latest video. A bit scary though...

Got the 6S Roma this week, was fun reading through this thread. Thank you L0stB1t for putting in so many details. I look forward to diving into it soon.

I have one question, though, that maybe someone can provide some thoughts on. I took the quad out for a first flight today. I'm getting these clicking/knocking sounds that i've not had on my other builds. At idle on the ground it will knock occasionally. When cruising it will be more often and the quad will noticeably vibrate from it. On a punchout, when throttling down things definitely don't sound right, though, i'm not sure how to describe it. In all, it's not flying very smooth. I tried the PIDs that came with the quad and BetaFlight defaults. I'm running RDQ 6S 1050 with the Gemfan Hurricane 51477. Maybe I should try the lower pitch 51433 as per your suggestion L0stB1t.

If anyone has tips on how to troubleshoot, or what the issue might be I would be very grateful!

(04-Apr-2021, 05:05 PM)showcaser Wrote: Got the 6S Roma this week, was fun reading through this thread. Thank you L0stB1t for putting in so many details. I look forward to diving into it soon.

I have one question, though, that maybe someone can provide some thoughts on. I took the quad out for a first flight today. I'm getting these clicking/knocking sounds that i've not had on my other builds. At idle on the ground it will knock occasionally. When cruising it will be more often and the quad will noticeably vibrate from it. On a punchout, when throttling down things definitely don't sound right, though, i'm not sure how to describe it. In all, it's not flying very smooth. I tried the PIDs that came with the quad and BetaFlight defaults. I'm running RDQ 6S 1050 with the Gemfan Hurricane 51477. Maybe I should try the lower pitch 51433 as per your suggestion L0stB1t.

If anyone has tips on how to troubleshoot, or what the issue might be I would be very grateful!

Maybe one of your motors has a bad bearing.  Try spinning each motor individually using the Motors tab of Betaflight Configurator, and see if the sound is coming from just one of the motors.

(04-Apr-2021, 05:05 PM)showcaser Wrote: Got the 6S Roma this week, was fun reading through this thread.

Welcome to the club!  

(04-Apr-2021, 05:05 PM)showcaser Wrote: I'm getting these clicking/knocking sounds that i've not had on my other builds. At idle on the ground it will knock occasionally. When cruising it will be more often and the quad will noticeably vibrate from it. On a punchout, when throttling down things definitely don't sound right, though, i'm not sure how to describe it. In all, it's not flying very smooth. I tried the PIDs that came with the quad and BetaFlight defaults. I'm running RDQ 6S 1050 with the Gemfan Hurricane 51477. Maybe I should try the lower pitch 51433 as per your suggestion L0stB1t.

That doesn't sound good  

I don't think the battery or props have anything to do with it. First check the basics: Are all motor screws secure? Are the stack screws and nuts ok? Is the battery lead or strap not touching any props under any condition? Is the connector between the FC and ESC properly seated? Can't hurt to re-flash the ESC's either.

If those things are fine then most likely there's a problem with the motors or ESC. Like SJChannel suggested, test your motors individually (without props!) in the configurator. If it only happens on one motor, swap it with another one to figure out if it's the motor itself or the ESC. Check everything with and without Bidirectional Dshot, and when it's enabled, check the error % like mentioned here: https://github.com/betaflight/betaflight...rification

It may also be useful to do a test flight with Blackbox debug DSHOT_RPM_TELEMETRY enabled. That way you can check for sudden drops in RPM on the motors. Drop the blackbox file here if you need help with this.

Good luck!

Thank you SJChannel and L0stB1t! Appreciate the tips.

I ran the motors through configurator and they sound fine without the props on. Put the props on and arm, and there is very distinct clicking/knocking at random intervals. I'll run through your suggestions for Bidirectional Dshot and will try to get some Blackbox info the next time i'm out.

Thanks again for the help!

Went for a short flight today to capture some logs. I'm no stranger at working with this kind of data (in a separate context) but without a reference I'm not sure what I should be looking for. If anyone has a moment to look at the logs they can be found here:

https://drive.google.com/file/d/1OS6iLeO...sp=sharing

I also tried with bidirectional DShot on and off and didn't notice a difference.

Thanks again!

I've set up Blackbox Explorer with Motor, Gyro and Debug graphs, and I've enabled Expo to clarify the gyro movement. Unfortunately the Debug graph doesn't show anything, but I see that debug DSHOT_RPM_TELEMETRY is enabled so I guess you simply had Bidirectional DShot disabled during this flight. Can you send a log where Bidirectional DShot is enabled?

   

From what I see in the motor and gyro traces, there are these "blips" in the Gyro pitch and yaw traces. It's very clear that the PID loop reacts to the pitch blips by increasing/decreasing the front and back motor commands simultaneously to try and stabilize everything. For the yaw blips it's harder to see the PID reaction.

Now the big question is if (1) the gyro chip is broken and sending bad signals, or (2) there's a problem with the ESC which causes sudden RPM and thrust loss, which is detected by the gyro. We can probably rule out one or the other if you get a log with RPM in the debug traces.

Because there doesn't seem to be any problem with the roll axis, my best guess at this point is that there's something wrong with the gyro chip

Thank you L0stB1t, my bad for not having bidirectional DShot enabled. May take a couple days but i'll get you the log. Appreciate the help.

Here's a short flight with bi-directional DShot turned on. Checked the file and the debug channel is there. As always, thank you!

https://drive.google.com/file/d/1844DYy4...sp=sharing

Looking at the log in Blackbox Explorer definitely seeing blips on the Gyro yaw axis. Even in the first ten seconds when the quad is on the ground the gyro yaw has a few occurrences of these blips. From the debug, the motor response is coming after the blip which, to your previous question, suggests the motors are responding to the gyro and not the other way round.

(15-Apr-2021, 11:04 PM)showcaser Wrote: From the debug, the motor response is coming after the blip which, to your previous question, suggests the motors are responding to the gyro and not the other way round.

Yes correct. Here I've put the graphs in the most logical order top to bottom: first the gyro sees a blip in the (pitch or yaw) movement, the PID loop changes the motor output at the same time and a little later (about 10 ms) we see a change in the RPM. (note that I zoomed in a bit to 200%)

   

If you would set the debug mode to GYRO_SCALED you'll get the gyro signals before filtering, and I bet there will be much sharper and higher spikes. A few milliseconds later there will be the filtered Gyro result and the motor output. I kinda wish it would be possible to get both GYRO_SCALED and DSHOT_RPM_TELEMETRY information so we could see the unfiltered gyro, filtered gyro, motor output and motor RPM cascade in the same screen...

With GYRO_SCALED we may also figure out if the blips occur at a specific frequency that can be filtered out efficiently. That would just be a mitigation for a real hardware problem though, so get yourself a new FC through warranty in any case

Hi L0stB1t. I was finally able to fly again and now tested your settings. I think your PID's are perfect. Thanks for that.

For the first flight I had different filter setting than yours and noticed some remaining noise for D-Term (yellow highlighted in the screenshot). So I changed them to yours like you describe in this post). And I think it worked. The only thing which confuses me is the different noise before filtering (green highlighted). Do you know what this is?

I alos noticed again some not so high frequent oszillations (green surrounded in the second screen). What does this mean?

Here is my log file if you have time to check. https://server.roeslr.com/s/5B4zdJdtjGGejWP

   

   

Hi MelBourbon

Happy you found my tuning guidelines useful. Regarding your screenshot, the part you highlighted in yellow does seem much better in the second flight. Did you put the second notch filter around 440 Hz like I did? Because it looks like a notch around 500 Hz (and cutoff at 450 Hz) would be even better in your case.

The parts you highlighted in green are for the pre-filtered D-term, which seems quite noisy. I never really looked at mine but your post-filtering D-term noise looks reasonable. Now, you must understand that for your A-flight you never raised the throttle higher than 60%, so you're not getting the full picture there. For a tuning flight, I always do a combination of some smooth flight, full punches, fast flips, fast rolls and propwash turns (full info: https://youtu.be/UIbcPX5iacA?t=1). Your B-flight is actually kinda funny because it looks like your throttle was never at 90%, as if it jumped from 80% to 100%

I loaded your full log in PIDtoobox twice (both the A and B flight) and put the 4 most interesting graphs next to each other. Let's concentrate on the roll axis (which is the noisiest):

   

The Gyro filter is doing a great job except for that very loud horizontal line at the bottom around 70 Hz. This gets even worse for the D-term noise which also corresponds to the wavy yellow D-term lines you see in Blackbox Explorer. I'm wondering if your motors are not getting too hot? 70 Hz is a lot lower than the line I had at 220 Hz, and it's unfortunate because all frequencies below 100 Hz are useful information for the flight controller. Frequencies between 25 and 100 Hz are in the propwash range, so filtering it out would hurt the propwash handling. I suggest you check your hardware first; motor screws, arm screws and stack screws in particular. Also run your motors individually (without props!) to listen if one of them has a bad bearing or something like that. I hope you can find and fix the issue. If not, do try to set the notch filter at 70 Hz (cutoff 60 Hz) and see how it goes. In either case I'd be very interested to see the difference in a log file.