I am going to try and put as much information into this post as I can to better assist those willing to look over my specific issue. I'm pretty green, in fact, today marks my 30th day into FPV and the RC world in general, so I'm still having to absorb a lot of information, especially when it comes to tuning. Below you will find the error I am experiencing; my build details (link to pictures); an array blackbox logs with relevant PID values in the titles, as well as a full flight log; and lastly a video illustrating the error and some normal flight footage with rolls, flips and everything else.


1. YAW BOUNCE-BACK & ALTITUDE GAIN
When preforming a full stick yaw spin, in either direction, my quad increases in altitude and bounces back at the end of the spin. When taking a look at the blackbox log, for a left turn let's say, the gyro moves at a ~45 degree angle upwards until it reaches the setpoint (which is dropping rapidly) and then moves diagnally downwards at the same angle, overshooting the setpoint (now at 0 degrees) again slightly which causes bounce back. This error is practically identical no matter what changes are made to the yaw PID values, which is making me wonder if it has anything to do with those values at all. I have successfully tuned the roll and flip axes (both could use a little more), but the methods used to tune those do not seem to be working in this scenario.



At first I thought increasing P-term was the way to go because of the clear lack of "oomf" to the gyro and the fact that at the start of the move, the gyro seems to follow the setpoint (meaning feedforward is doing its job, right?). With that said, given the speed of the bounce back and the increase in altitude, it also made me think it's the other way around regarding P-term. The PID_Error for the yaw axis when preforming a yaw spin is always around 570 degrees.


2. RUNAWAY TAKEOFF TRIGGERED
With my current yaw PID values (P=68, I=99, D=0, FF=119), if I then set P-term to ~60, or if I set P-Term and I-Term both to 100, my quad issues a "RUNAWAY_TAKEOFF" and disarms itself upon the execution of a full stick yaw spin. In both cases, the gyro fails to keep up with the setpoint. I have included blackbox logs for these failures as well.

Now I'm starting to wonder if this has anything to do with my rates.
Am I trying to spin the quad faster than it can actually go?


CURRENT PID VALUES
Roll   P:65  I:94  D:59  FF:119
Pitch  P:70  I:99  D:65  FF:125
Yaw    P:68  I:99  D:0   FF:119
-------------------------------------------------
CURRENT RATE VALUES
Roll   RC:1.94  Rate:0.64  EXPO:0.21
Pitch  RC:1.94  Rate:0.64  EXPO:0.21
Yaw    RC:1.65  Rate:0.64  EXPO:0.21
-------------------------------------------------
TPA=0.75
TPA Breakpoint=1750


ALL BLACKBOX LOGS & FLIGHT VIDEO
https://www.dropbox.com/sh/rcyq4l9l0uhdz...iPQSa?dl=0

QUAD BUILD INFO (710 Grams AUW)
https://pastebin.com/raw/mpk3Rhth

QUAD BUILD PROGRESS PHOTOS
https://i.imgur.com/WMH9NQg.jpg

BETAFLIGHT DUMP
https://pastebin.com/raw/Zy8qPzqK

BETAFLIGHT DIFF
https://pastebin.com/raw/SXa8QPXH

BLHELI_S CONFIG
https://imgur.com/g14qQkO

ronin will be your guy to overlook your bb-log :-)

personal i would start by cut the d value by half, look if it isnt going fully away. not as the solution, just to get the issue sorted out.

(31-Jan-2021, 12:58 PM)hugnosed_bat Wrote: ronin will be your guy to overlook your bb-log :-)

personal i would start by cut the d value by half, look if it isnt going fully away. not as the solution, just to get the issue sorted out.

Cut the D-term of Roll and Pitch in half? Could you explain how this will affect the yaw axis? Again, I am still learning, so insight into your thought process is always appreciated.

I am currently tuning a new profile using Ronin's guide... in the snow. Hopefully I can gather some additional understanding on my specific issue, like UAV Tech's videos did for most everything else. I can't wait to be cleaning the dirt out of my motors later tonight. So much fun!!!

On a side note:
When I crash into the snow, where the interior of the quad gets packed with the stuff, everything generally works out just fine. I've put two healthy layers of conformal coating on my boards and sealed various other components with liquid tape (antenna mounts, cracks in VTX & camera, etc); there's really not much I can't throw at it by now. The problem comes with the current sensor reading in my OSD. Voltages read perfectly fine, but the mAh consumption value increases at a very high rate, until the end of the flight when it's ~3000mAh, where the actual reading is more like 900mAh. Once the quad fully dries out, the readings go back to normal. This has occurred twice so far, but I cannot figure out what is causing the issue. The first time it happened, I put another layer of conformal coating on everything just in case and then used the liquid tape to seal up other exposures. Any ideas what could be causing this? The only areas I haven't sealed up are the [Air Unit -> FC] & [ESC -> FC] pin connectors.

yess both roll and pitch

as you are in ronins guide process, dont follow my advice :-)
i guess you optimize the filters, it might be an issue there?

(31-Jan-2021, 10:50 PM)hugnosed_bat Wrote: yess both roll and pitch

as you are in ronins guide process, dont follow my advice :-)
i guess you optimize the filters, it might be an issue there?

You know what dude? You might actually be right; filter settings might be the key!

Here's a full stick yaw spin I did while tuning via Ronin's guide.
Comparing it to the OP photo, I'd say the gyro is looking quite a bit better.



It's certainly not perfect, but it's headed in the right direction. Although it seems Ronin's starting P-term values were a little high for me, but will do more testing tomorrow. I flew a 1100mAh Thunder Power Adrenaline Series 100C lipo today for the first time ($45 battery what am I crazy?!) and could not believe the resulting performance boost. I receive 7 more of them (there goes the rent money) in a few days and am now very curious to see how they affect the gyro tracking and PID response. Seriously, it felt like I was flying a completely different quad, and the battery is only 10g lighter than my normal lipos.

ROLL (D20): https://i.imgur.com/WUkaqeq.jpg
ROLL (D25): https://i.imgur.com/aJ9OLEQ.jpg
PITCH (D20): https://i.imgur.com/OArHuGN.jpg
PITCH (D25): https://i.imgur.com/yyezAuY.jpg

Hey man sorry I missed this. I was going to say off the bat your D's might be too high in the 50's but I see you've brought them down to the 20-30 level.

What I've started to look for (which I really ought to update in the guide) is the I term taking over for the P term. That might be why you think my P's are too high. There will always be I movement but if I see an improvement in the I movement between 2 scenarios, I will take that into consideration. It's not always straight forward too as it might seem like it gets better then worse across scenarios. So I take it with a grain of salt. But if your P term is clearly too low, you will very clearly see I term take over. (Which is no bueno.)

In your case, have you touched feed forward? When I can't get P to tighten that gap between setpoint and gyro, I move onto feed forward to do what remaining tightening I can.

(13-Feb-2021, 07:35 PM)the.ronin Wrote: Hey man sorry I missed this.  I was going to say off the bat your D's might be too high in the 50's but I see you've brought them down to the 20-30 level.

What I've started to look for (which I really ought to update in the guide) is the I term taking over for the P term.  That might be why you think my P's are too high.  There will always be I movement but if I see an improvement in the I movement between 2 scenarios, I will take that into consideration.  It's not always straight forward too as it might seem like it gets better then worse across scenarios.  So I take it with a grain of salt.  But if your P term is clearly too low, you will very clearly see I term take over.  (Which is no bueno.)

In your case, have you touched feed forward?  When I can't get P to tighten that gap between setpoint and gyro, I move onto feed forward to do what remaining tightening I can.

When following your guide, I played around quite a bit with feed forward for all three axes (from 125 to 175 in steps of 25,) which helped get the gyro to track setpoint on the way in and out of the move a little bit. It actually made my roll really sharp, however, fast flips would still get some bounce back, and setting the FF value for pitch higher than 125 was making the motors hotter than I was comfortable with. Roll was handling higher FF values well though.

Before I make a full response, I am going to do a fresh run at tuning using your guide. I will collect and organize all of the data and describe my changes in detail so you can see my thought process. It is very possible I am simply reading the logs incorrectly and making some mistakes as a result. But regardless of that, the yaw is still weird. Adding feed forward to that axis didn't help from what I recall. I also had no luck following UAV Tech's method (with filters,) in regards to fixing the yaw issue. It was actually worse when I followed his guide, which you can see in the original post's picture.

Does I-term fight against P-term at all? Do they contradict each other? In the last picture, it sort of looks how you are describing, where I-term takes over for P-term, but when I would reduce the I-term, stability issues would surface. Anyways... I'll follow your guide again and get some good data we can look over once the snow melts a little; will probably be in a week. I will also try to find a field to do this in instead of the desert, that way I can crash without having to interrupt the tuning process due motor cleaning.

I term does not fight against P term. It's D term that serves to counteract both P and I term. However, P term is "first on scene" (as UAV Tech would say), I term will take over if it is insufficient. That's when you see the I term induced oscillations. Often, one might reduce P term when seeing these slower oscillations thinking they are P term induced.