Weird hovering issue - requires 3A to hover then jumps to 6A after a few and back to 3A and so on

[moadi]

Hello everyone,

I’ve just finished setting up a 4" quad running orange cube, I followed the guide on how to tune it, everything seems perfect, flies smoothly, I like it so far but I just slapped a 4s1p li-ion battery on it to test how long it hovers for and I got a very weird behavior:

• flying indoor

I take off in STABILIZE/ALTHOL, hover for a little and then put it in LOITER, it positions itself nicely and stays in place with a little drifting here and there but that’s alright
So it usually takes 3A current to get it hovering, it vibrates or oscillates (not sure what to call it) a little and starts dropping so I have to apply more throttle and it goes up to 6-7A to get it back to hovering then after a few it starts climbing so I have to lower the throttle and go back to 3A to get it to hovering level.

plotted current on log viewer (while hovering in LOITER, no user input):

I’m not sure why it’s happening, is it related to tuning?
I’ve attached the logs, I’d really appreciate it if someone could take a look at it

hovering_issue_logs.bin

Thanks alot

[dkemxr (Dave)]

How does it look with Flow disabled and using GPS?

[moadi]

I don’t think it’s related to flow since if happens even in stabilize mode and I don’t have a GPS its an indoor quad.

[dkemxr (Dave)]

I would guess that it does. Is this Standard firmware you are using?

[moadi]

Yes standard firmware, I recently updated to the latest version.

If I’m using stabilize mode doesn’t that mean FLOW is not doing anything?

I want to try to disable it and then fly in stabilize to see if it makes any difference even though it used to fly with no issues with flow.

How does one temporarily disable flow is it as easy as changing one parameter may I ask?

Thanks alot for the help

[xfacta (Shawn)]

If you are using the Cube Orange power brick it is not accurate nor precise down around those current values, so dont trust it much below 10 amps.

Try these for a start:

ATC_THR_MIX_MAN,0.5
MOT_SPIN_MIN,0.12

[dkemxr (Dave)]

@moadi

If I’m using stabilize mode doesn’t that mean FLOW is not doing anything?

Perhaps you are right and it’s just tuning. Outputs are oscillating, maybe D-term and it’s bouncing off MIN

Try dropping the Rate D-term and go here: Setting Motor ranges

[moadi]

@xfacta I think in this case it was accurate, I mean the fact that it’s pulsing and not staying the same, it drops then climbs and so on

[moadi]

@dkemxr I see, I will try what you said and I’ll follow the guide in the link and see how it goes first thing tomorrow, thanks alot

[moadi]

Ok so I have set up the motor ranges and calibrated ESCs, didn’t seem to help.

I enabled the notch filter and it kinda solved the issue, it now flies smooth in alt hold and stabilize but when I put it in LOITER now it starts wobbling like crazy, holds it’s place but wobbles way too much.

loiter_wobbling_after_enabling_notch_filter.bin

EDIT:
After adjusting the D terms as @dkemxr suggested, it seems to hover almost perfect now with the notch filter enabled aswell, slight wobbling here and there but nothing serious

dropped ROLL D RATE from 0.002 to 0.001
dropped PITCH D RATE frrom 0.002 to 0.001

Although it flies nicely, it seems to draw more amps now (4-6A while hovering, it used to take 3-3.5A) I’m not sure if my power brick is giving false reading or what

Another unrelated question:
I’m running a 4s1p li-ion battery, I read that the lowest safe voltage is 2.5v per cell, does this mean when the reading I’m getting reaches 2.5v * 4 cells (after voltage sag) that I must land? It usually goes up to ~11v after I land which means it still has more juice

I will try to run an autotune now (first time) to see how it goes.
Since this is gonna be an indoor quad, do I need to run the autotune indoor or it doesn’t matter?

Thanks alot

[amilcarlucas]

Autotune always runs better and faster indoors

[Juergen-Fahlbusch]

@moadi

@moadi

Another unrelated question:
I’m running a 4s1p li-ion battery, I read that the lowest safe voltage is 2.5v per cell, does this mean when the reading I’m getting reaches 2.5v * 4 cells (after voltage sag) that I must land? It usually goes up to ~11v after I land which means it still has more juice

If any cell voltage drops below 2.5V you risk permanent damage of your cells, so yes, you have to land immediatly if your voltage goes below 4*2.5V = 10 V. Better land little earlier becouse your cells might be not 100% identical even if they are balanced before start.
That the voltage rise again after you stopped using higher current is normal. But notice the unencumbered voltage alone is not the best state of charge indicator

[dkemxr (Dave)]

@moadi

I enabled the notch filter and it kinda solved the issue,

That’s illusory because it’s not configured properly and it’s doing nothing.
Disable FFT. FFT_ENABLE,0

And set these:
INS_LOG_BAT_MASK,1
INS_LOG_BAT_OPT,4

Then make another hover flight in AltHold and use Webtools to configure the notch filter with throttle reference. Webtools
Or post a link to the log for help.

And don’t bother with Auto Tune until the output oscillation is dealt with. We will see that in the Notch filter config flight.

[moadi]

I’ve updated the params and went for a test flight of about 30 seconds in ALT HOLD (I had to apply user input to keep it in it’s place, hope that doesn’t ruin the test)

I took the log file and put it into that web tool but I had no idea what I was looking at

Any help understanding what I’m supposed to do with those values?

Thanks alot

log_for_notch_filter.bin

[dkemxr (Dave)]

Give these parameter settings a try.

Normally you would then do another hover flight to see how it’s working as the settings chosen were based on “estimated post filter” results.
Before doing an Auto Tune you may as well increase these as they are too low for a 4". Auto Tune will likely increase them, perhaps much higher.
ATC_ANG_PIT_P,9.0
ATC_ANG_RLL_P,9.0

And adjust these:
PSC_ACCZ_I,0.34
PSC_ACCZ_P,0.17

You do not have Dshot capable ESC’s?

[moadi]

My ESCs are DSHOT capable, should I switch to that? Is it supposed to make a difference?

[dkemxr (Dave)]

Absolutely you should switch to Dshot and do it now before you go any further with configuration and tuning. Steps to follow:
Flash the BDshot version of firmware for Cube Orange by going here CubeOrange BDshot and downloading the .apj file. Then use Mission Planners “load custom firmware” option to flash it.
Move the Outputs to the AUX channels on the cube and re-assign the motors to those chanells. Info here Dshot Use Dshot600.
Then configure for Bdshot. Info here Bdshot
Make a flight and see if you have ESC RPM in the log.

After doing this the notch filter configuration will change for the better.

[amilcarlucas]

Take a look at my blog post on this forum. I explain how to configure bdshot600, esc telemetry, notch and so on. All step by step without skipping any details

[moadi]

Thanks @dkemxr, I flashed the BDShot firmware on the cube and now I’m trying to configure the ESC through the blheli suite.

I didn’t have anything from the ESC wired to the cube except for the motor signals so I wired as follows:

ESC TX ↔ Cube Serial4 RX

My ESC comes from this stack: Review: HGLRC FD435-VTX Stack – HGLRC Company

Then I tried both:
SERIAL4_PROTOCOL = 16
SERIAL4_PROTOCOL = 38 (FETtec OneWire ESCs — Copter documentation)

But it doesn’t seem to respond, the cube is not querying my ESCs when I hit ‘Read Setup’ in BLHeli.

Any ideas what could I be doing wrong?

@amilcarlucas Great post, I noticed that I skipped a few steps I will definitely go back to it once I’m doing with this, much appreciated!

[dkemxr (Dave)]

You don’t need ESC telemetry at this point. The goal is to get Bdshot working which will supply ESC RPM. But set the Protocol to 16 in any case. You also don’t need to configure any settings in BLHeliSuit at this stage either but have you set these parameters?
MOT_PWM_TYPE,6
SERVO_BLH_AUTO,1
SERVO_BLH_BDMASK, (to the outputs you have connected the motors to).
SERVO_DSHOT_ESC,1