iNav tilting at full throttle on latest firmware

[evranch]

Updated to the latest firmware last week to get the new notch filters etc. Copied all my settings over and my quad flies great, with 2 issues:

- Full throttle punchouts roll the quad to the right
- The props will no longer idle when armed, maybe one or two will flick a bit. I need to give a throttle command sufficient to take off before all 4 rotors will spin up. With folding props I want them all spun up before I try to take off. Solved by switching back to DSHOT

I suspect that applying full throttle is giving all the motors 100% throttle, which means it can't stabilize. It worked great with the older firmware, though. Is there a new setting that makes punchouts apply 100% or something? Or a change in ESC calibration?

I tried increasing the motor idle % but still, the props aren't spinning at idle. I'm at 15%, same as I used to have it set to, and don't really want to go much higher. I don't need it trying to fly itself at idle...

 

[ElectriSean]

I'm not up on the latest iNav, but if they've added a mode called anti-gravity, you can enable it to get rid of the the full throttle bobbling. Betaflight has had it for a long time, it just increases the I gain at high throttle.

Are you running DSHOT? I know that was recently added to iNav, so there are probably bugs.

 

[LitterBug]

Did you rerun your autotune after installing the latest firmware? Did you do an update, or a fresh install and only copy the parameters you know you need?

I find when upgrading firmware on any of my quads/planes/ESCs, that many times it is best to start fresh with the default settings and go through the DIFF and find settings that I know can be copied over like channel mapping, radio setup, and simple config. After that I go back through the tuning process as if I were starting from scratch. Many times, especially when jumping a release or two, hanging on to old settings just doesn't work well. I've been burned before, and helped others out where they tried to just copy over all settings.

Cheers!
LitterBug

 

[evranch]

I don't see anything called anti-gravity, just Airmode which I believe is supposed to be left on all the time.
I did change from DSHOT to MULTISHOT come to think of it as that was the recommended protocol from the configurator - maybe I will change back to DSHOT300 as it was working well.

I did a fresh install and set it as a 7" freestyle quad, then copied my parameters over - serial ports, OSD, RC channels and setpoints, accelerometer calibration and PID setpoints.

Maybe I should reset the PIDs to the 7" freestyle defaults and go from there. It flies really tight as it is though - just the issue at full throttle.

 

[LitterBug]

If you changed from DSHOT to MultiShot, you will need to recalibrate your ESCs. DSHOT does not require calibration, where Multishot does. Idle % and other parameters are inconsistent between those protocols too. I find DSHOT with it's higher resolution normally takes higher idle than PWM, Oneshot, or Multishot.

Cheers!
LitterBug

 

[evranch]

OK, so I'm back on DSHOT and the idle issue is solved, all motors spinning nicely at 7%. Though I might raise to 10% as there is a little jitter, I want my folding blades snapped out firmly. Is any calibration required on DSHOT then? Could some of the ESCs be receiving incorrect commands?

I'm using a quad ESC in a stack, not 4 separate ESCs.

It still rolls to the right on punchout. Interestingly, if I raise the throttle slower it does not - only if I slap it from hover to 100%.
Could this have anything to do with Throttle PID Attenuation? I have not enabled it.

If I back off the throttle it doesn't level out unless I roll it back, but I think this is the expected behaviour of ACRO, to hold attitude unless commanded.

I finally put 70% expo on my throttle because I never noticed that the default is 0% and wow, is it ever easier to hover.

Wind is up to 25MPH and gusty so it's not prime quad tuning time that's for sure.

 

[ElectriSean]

When running DSHOT you do not need to calibrate the ESC's. Enabling throttle PID attenuation would probably make things worse, that is to get rid of higher throttle oscillations. You probably want a little more roll I gain.

 

[LitterBug]

I would redo the sensor calibrations and give it another shot.

DSHOT does not require calibration.

It has been a really long time since running iNav on a quad and only have it on a couple planes right now. Not sure I have many other ideas at the moment.

Cheers!
LB

 

[evranch]

I'll give it a try. I seems fairly high already though. I'm currently at:

	P	I	D
Roll	30	50	25
Pitch	30	50	25
Yaw	60	50	0

Interestingly if I click "Reset PID controller" all the values change a lot:

	P	I	D
Roll	40	30	23
Pitch	40	30	23
Yaw	85	45	0

I can't find any documentation as to what they are actually being reset TO. The defaults for the 7" freestyle quad?

 

[LitterBug]

I'll give it a try. I seems fairly high already though. I'm currently at:

	P	I	D
Roll	30	50	25
Pitch	30	50	25
Yaw	60	50	0

Interestingly if I click "Reset PID controller" all the values change a lot:

	P	I	D
Roll	40	30	23
Pitch	40	30	23
Yaw	85	45	0

I can't find any documentation as to what they are actually being reset TO. The defaults for the 7" freestyle quad?

If I remember correctly, they get reset to firmware defaults. You would need to re-apply the 7" settings over top of that. (based on possibly foggy memory of setting up a plane on iNav a couple months ago)

 

[ElectriSean]

I'll give it a try. I seems fairly high already though. I'm currently at:

	P	I	D
Roll	30	50	25
Pitch	30	50	25
Yaw	60	50	0

50 isn't that high, and thankfully excessive I won't risk your motors or anything, it'll just make it feel stiff. You just need to find the sweet spot, or adjust your flying Or bug Pawel to implement Anti-gravity

 

[evranch]

OK, so resetting the PID controller does put it to firmware defaults. I tried changing the configuration to the defaults for a 10" quad, which increased the P gain and had a similar I and has the notch filters tuned for a quad about this size and weight. Flies nicely, more stable in the wind but still rolls at WOT.

I noticed that in ANGLE mode it tilts a little on a punchout but doesn't run away like in ACRO.

Battery is charged again and it's time to try bumping up that I term. I'm tempted just to switch to Betaflight, because the main reason I run INav is for RTH if I fly behind a hill, and Betaflight now supports a GPS rescue mode.

However I do occasionally use position hold when herding sheep so I'll see if I can't get this tuned out. I'm wondering if it was just not visible before with the heavier batteries, and now that I'm running lighter packs it has less inertia or something.

 

[ElectriSean]

Makes sense thatit would behave better in angle mode, it will always try to go back to level. Betaflight recue mode is decent enough for long range failsafe situations, but I think you need to be 100-150m away from home for it to work.

 

[evranch]

It probably would be good enough. All it really needs to do is maintain a controlled climb and move towards me somewhat.

My main RTH usage case (and I've done it a couple times already) is getting caught up in working the sheep and losing my video feed behind a hill. Before I got the GPS working I would slap ANGLE mode and punch out until I climbed back into video range. Pretty nerve wracking, though I always got it back.

Now with RTH I just hit the switch and it's set to climb to 100m and return, usually this is enough to regain control.

I set up INav for inflight adjustments of the I term and put the PIDs on my OSD. But now the wind is up over 30MPH so it's going to be the evening or tomorrow!

 

[evranch]

Wind came down enough that I could fly. OK, so it looks like the I term doesn't have much effect on this rolling issue. I used the inflight adjustment to vary it from 45 to 70. At 70 the quad was starting to toilet bowl a bit in the wind. At 45 it was feeling kind of... loose in the wind.

The main factor is how hard I throw the throttle. If I throw it 70->100%, no effect. 50->100%, slight pull to the right. 20->100%, dramatic roll to the right, 0->100% almost an immediate tilt to 90 degrees requiring active recovery.

Varying the throttle without hitting 100% does not cause the effect. i.e. 20->80% just results in a stable climb. This means the simplest fix is just to limit my throttle travel... but that doesn't answer what is going wrong.

Maybe I should flash Betaflight to see if it does the same thing.

 

[LitterBug]

With props off. Do your motors all spin the same speed at 100% in the motors tab? If you spin them by hand, do they all feel the same?

 

[evranch]

With props off. Do your motors all spin the same speed at 100% in the motors tab? If you spin them by hand, do they all feel the same?

Heh, they sound fine together and feel good, but the right rear motor sounds awful when run alone. No ammeter on the quad but the voltage sags more when this one is running without a prop. I remember this one I had to disassemble and re-solder the leads to it after they were torn off in a testing accident, and I was wondering how long the bearings would take to pack it in.

I have a set of spare motors, so I'll drop one on. But first, testing with Betaflight to see if the problem was in software.

I learned a lot about quad tuning today from Betaflight without having any built-in profiles like iNav does. It turns out that the default settings in Betaflight are good for race quads, but can barely keep an 8" in the air. I also upgraded my ESCs to bidirectional DSHOT and set up Betaflight's RPM based filters, which are really cool. Literally cool, my motors are stone cold and the quad hovers rock solid.

Ultimately my PIDs are all much higher than they were under iNav. I gains are well over 100. I was doing some mini-punches indoors and it flies straight up with no rolling. The quad feels super tight. Also I now have feed-forward gain which should really add some stick response. Can't wait to rip it outside tomorrow!

 

[evranch]

So, bad motor notwithstanding, the issue is completely gone under Betaflight with the higher PID settings and Anti-gravity enabled.

Also, I think power consumption is down with the 96kHz ESC firmware and better filtering? I flew aggressively for 9 minutes and only discharged my 3s 18650 pack to 3.8v before I cut it too close to my old tractor in the wind and sheared a prop. Charging right now to see how many mAh I burned.

The resulting tumble lost me 3 props, the VTX antenna, bent the soft aluminum motor mounts, and ejected that 18650 pack about 20 yards where it slammed into the shop wall... and was perfectly fine! Cylindrical cells for the win!

The quad is back in shape and ready to fly already, but I need to tune the yaw PID as that's the only input that results in bad behaviour. Yawing at a decent rate causes the quad to wobble.

 

[LitterBug]

Awesome to see you have made progress! I have mixed thoughts about iNav. It's OK when it works right, but my recent loss of control (complete lockup) on an otherwise proven stable wing build has left a bad taste in my mouth. I know serveral others with much experience flying that have used it that shy away from it too. I still have it on one RMRC Nano Skyhunter, but am not sure it will go back into the Strix Nano Goblin that I had to rebuild when the stuff hit the fan. (fence, ground, and tree actually)

Cheers!
LitterBug

 

[evranch]

iNav definitely worked better out of the box with its pre-set profiles for different quad sizes. However, Betaflight just has that many more features focused on performance.

There were a couple really odd behaviours in the inital setup of Betaflight that never happened with iNav, the worst being some sort of windup in ANGLE where a resonance developed and then the quad decided to start climbing even after chopping throttle to zero. I had to disarm and drop it out of the air a couple times. The solution was dropping "strength" down from 50 to 30, which must be a general level for the ANGLE PID? Unfortunately I need ANGLE working because that's the only way I can fly LOS to land it... embarrassing...

I can't believe I only burned 1100mAh in 9 minutes of flight, at fairly high throttle settings too, but that's what the charger put back in the pack. That means I should be able to fly well over 15 minutes with the 3000mAh pack, considering amperage rises as voltage drops.

I'd like to get out flying again but that accursed wind really ruins this time of year! There are only a couple short windows every day that are any good for flying.

I didn't change my yaw gains but I did enable "Absolute control" which is supposed to control attitude changes resulting from yaw. Since the yaw itself is working well, but the attitude bobbles on hard yaw, maybe this is the solution.