It has probably been asked before, but I couldn't find any good information. Anyways, I got some issues with what I think is ESC desync when doing a fast right roll. According to the blackbox log, the right rear motor stalls as the motor signal goes to maximum and the quad tumbles out of the air. This happens everytime I do a fast roll to the right.

my setup:

- Xracer V3.1, BF 3.1.7, 8k/8k, Dshot600
- Racerstar shot 4 in 1 ESC with BLHeliS 16.65
- Aokfly fire phoenix 2205 2500kv motors

Is there any settings I could change to prevent this from happening?

High Demagnetisation Compensation could help.

Increase minimum throttle. Since you're using Dshot, i think that value is called digital_idle. On oneshot or multishot it would be min_throttle.

Thanks guys, I'll try increasing the min throttle first to see if it fix anything.

(15-Jun-2017, 08:04 PM)oyvinla Wrote: Thanks guys, I'll try increasing the min throttle first to see if it fix anything.

As FF said, it's idle speed because you're running DShot.  In fact, once DShot is enabled, you won't even see min/max throttle etc. in Betaflight anymore, just  idle adjustment (but it's in the same location ) From the Betaflight Wiki:

To check what value digital idle should be set to, connect to configurator, go to the motor tab, take your props off. Click the little check box half way down the page that says you have taken your props off, you know what you are doing, you take full responsibility for your actions and you consider fingers pointless appendages anyway.
Then plug in the LiPo to the quad (USE a Current Limiter Always). Click on the master slider and use the up arrow on the keyboard to increase the master slider one point at a time until you reach a point where all 4 motors have just started to spin, I don't mean twitch, stutter or start stop, I mean the lowest value where they are spinning, it is probably about 1010, don't worry if it isn't, but I am going to use that number as an example.
You can now take the master slider back to zero and unplug the LiPo.
Take the number (for example purposes 1010) and add thirty points to it (so in the example we are now at 1040).
Not we need work out what percentage of the throttle range this is, I don't know your math skills so sorry if this is a little hand to mouth.
Subtract 1000 from it, then divide by 10
So in the example:
1040-1000=40
40/10=4
Go to the cli tab in configurator
Click in the box at the bottom and type:
set digital_idle_percent = 4
(Replace 4 with whatever digital idle you arrived at in the calculation) Hit return, you will see the command processed in the back bit above, Then click in the box again and type
save
Hit return and done

Could also be a grounding issue. Joshua Bardwell has a good very about it.

maybe you have checked... also make sure motor screws are not touching windings

Desync is normally caused by either electrical noise, or too high RPM, or both.

When motors are running at higher RPM, the ESC is more likely to miss the "zero crossing", noise will also make that more likely to happen. If the ESC can't detect the approaching  "zero crossing" accurately, then it will not know when to fire the next phase (create magnetic field to push motor). If this is a persistent issue then the ESC will just shut down (ESC desync).

And when your motor spins at high RPM, it requests a lot of torque and that creates a lot of current through the coils of the motor, eventually causing higher level of electrical noise in the ESC

An easy solution is to add some extra capacitor to the ESC power.
Another solution is to change ESC settings (search for keyword desync): https://oscarliang.com/best-blheli-32-settings/

(16-Jun-2017, 10:10 AM)Oscar Wrote: maybe you have checked... also make sure motor screws are not touching windings

Yep, double checked it. It's the nimbus quad and as you know I had that issue when I first built it

I did some testing yesterday but could not get rid of this issue.

I tried: 
1. increasing the min throttle up to 1100, no change
2. changing from dshot600 to multishot 16khz, no change
3. from multishot 16khz to 8khz (same as PID loop), no change
4. set demag compensation to high, no change

I noticed something interesting at that is that all the motors seem to lock in at a certain throttle value and stay there even though the quad is tumbeling around in the air when it happens. could this be a faulty FC? this FC have had 14.85V connected on the current sensor input as the current sensor output on the previous racerstar 4 in 1 escs were faulty.

If you look at the screenshot from the BB-log, you clearly see that all motors are fixed at a certain value. If this was a desync issue I would think that the FC would try to correct for the missing thrust on the rear right motor with the 3 other motors and have the signal go all over the place.

Can you post the log? It's really hard to make any comment on a screen shot.

Sure, here is the dropbox link: https://www.dropbox.com/s/yjdcyxljbf6edp...1.bfl?dl=0

---

One thing that is also quite interesting is how much P oscillations I have even though my P is quite low at 40

Here is also an interesting place to look in the log. One motor is maxing out and everything starts to oscillate:

Hmmm...

I don't think there is anything else going on here other than motor one not spinning up when you start to end the roll to the right. As you still have the throttle at zero, the motors can't increase more than air mode allows them to do, so they stay pegged at the maximum thrust that can be commanded at zero throttle.

My guess is that you have either a faulty motor, or a faulty ESC. Try swapping motor one with motor four and see if the problem moves with the motor or stays on motor one.

Looking at the first flight in the log, you also have some fairly strong vibration on the roll axis that is getting through the filters and shows as strong oscillations at anything over half throttle.

I'm also wondering what props you are using.

These 2500kv motors with 4S are pulling huge amounts of current, if your current sensor is to be believed. It may be more than the 4-in-1 20A ESC can cope with.

The current sensor is not calibrated in this log, they draw less than 100A at full throttle. the ESC is 30A and props are Dal Cyclone 5045. I'll probably try changing the motors around or put on some different motors. I also noticed that the CCW motors are more "coggy" when you turn them by hand than the CW, so it could be different batches with different magnets. (With coggy I mean that there is sharper and more resistance when turning them by hand).