Erratic behaviour at RTL after failsafe

I have quad that is rock solid in all flight modes. I was testing some old(er) battery packs the other day when
battery failsafe kicked in after a few seconds at 10.4volts. The quad climbed up to preset RTL - that is admittedly a little too high but I have a lot of tall trees around home - and then it started coming in.
That’s when havoc broke out!
It was tilting in all directions as if it was fighting strongs gusts of wind and at the end it started drifting towards some trees while it was loosing height. It could have been totalled if it wasn’t caught on a bifurcated branch 8 feet over a creek.
I am attaching here that flight’s .bin file:

The damage was repaired and a half an hour I go I went back out to see if I can replicate the problem. I flew around for a couple of minutes until failsafe voltage was reached, RTL kicks in and the quad picks up height then flies toward the landing spot then starts descenting like it is suppose to do, THEN suddenly same behaviour!

For a few seconds it dances and I am contemplating if I should/could take over control something that did not work that good the other time cause I was fighting the oscillations.
Suddenly it stabilized itself again and like nothing happened comes down and lands by my feet!
Here is this latest .bin file:

What is it happening??? I had problems with the piezo speaker before but this has been resolved by moving it to another location and isolating the vibrations with foam tape. Could it be a low voltage situation that breaks havoc in the FC?

Thank you in advance!


What you see is typical behaviour when you run out of power.
The FC is trying to stabilise above all else but it just can’t get enough power to do it properly.
Your quad is underpowered/overweight to begin with, at around 1700 at hover.
There is very little head room for error or loss of power.

As your FS kicks in you are using what precious little power you have left in climbing, and that about does it for the battery.

Thank You Mike for the quick responce!


  1. Which software did you use to produce those nice graphs?

  2. You say it is a power issue but I remember clearly that during this incident the voltage never dropped below 10.1volts. It know that because I was looking at the screen of a Nexus tablet mounted on the Tx.

  3. What do you mean by “around 1700” at hover?


Ok I studied the graphs and understood what you were talking about.

The hover throttle should optimally bounce around 1000 - in a range from 0 to 2000 - and the voltage is constantly pretty low under load…

That last one can be fixed with larger capacity battery that I do have. What about the hover throttle though.
I have 9450 props on right now - should I try larger diameter props? The motors are MN2213 950kV running on 3S.


After reading this thread:

I think I should be shopping some 4S battery packs…

I use APM Planner for graph analysis.
Ideally your PWM output at hover should be around the 1500 mark.
Higher than that the FC does not have a lot of headroom for stabilisation, consider 1800 as max power.
950kv on 3S should be spinning 10" props at a minimum, but I am not sure your battery will handle that.
9" props on a good 4S battery might be the answer.