Lets take one step back and discuss the battery problem.
Myself and others have had power issues with the Phantom 3.


So I want us to take one step back and be thorough and civil about it.
Heres a lot of knowledge about these beautiful drones on both
the DJI side and the User side.
Lets treat each other as adults and get started!


2015-09-01
I was at work near my DJI Dealer mid winter and suddenly lunch brake fell over me.
Any normal person would just get in the car, max the heater and eat something.
Well, I admit I'm an addict on drones as well as the scenery mother nature provides.
So I went up there and took them for a flight in my P2P

http://rcbutikken.no/nb-NO/kategorier/dji-phantom-multicopter.aspx in the middle of this picture:

You cant really not fly in scenery like this.

Anyway heres the map and general statistics


And here's the Flytrex data

You can easily see the sensor noise on the voltage data.
The Phantom 2 never had a problem with this in any climate.
My theory is that the P3 is more sensitive about the noise and lands for a sensor spike.

I forgot the album is here if you wanna visit your Norwegian Dealer

Should add something about the spikes..
The spikes is not from the intelligent LiPo battery as some seems to think.
In fact, that's one of the areas that have made all kinds of batteries awesome: They don't spike or make any kinds of noise.
They are caused by electrical components like regulators and motors.

Thought there'd be more discussion on this thread as it seems that you might be on to something with the sensor noise thing.
I've not had any issues, but I have a hard time coming to terms with the "don't fly unless your battery is fully charged" mantra. It's fairly common in the RC world to do multiple flights off single batteries after checking voltages post flight.

mike.wildlight Posted at 2015-12-16 09:58
Thought there'd be more discussion on this thread as it seems that you might be on to something with ...

'It's fairly common in the RC world to do multiple flights off single batteries after checking voltages post flight.'

Yes, it is common to make several flights on a single battery, but until a cause is firmly established or DJI make firmware changes, it is not advisable with a Phantom, and there is a difference.

As one used to writing code for micro-control systems, a spike is not usually considered, a good code writer averages out values over a period of time to determine a realistic value, and that should be the case here.

I am of the opinion that there is a valid reason to think that some condition causes the aircraft to shut down its motors in flight, but the question is, in response to what? Going over the cases, I think I have seen that when it happens, the aircraft is not dead in the air, at least one person said the lights were still on while it fell to the ground. So that tends to rule out the idea that the battery itself is shutting off power. It then seems to suggest that the aircraft is shutting the motors off for some reason. And it seems unlikely that it is a bunch of people initiating a CSC manoeuvre.

As an example, if I was writing code for this I would have a section of code that is the only code able to shut the motors off and when that section is called, the first thing would be to establish if the aircraft is airborne at the time, and if so, exit without shutting the motors down. I know you can use a CSC to shut the motors off in flight, but to be honest, I cannot think of a valid reason for doing so. Anyone?

Geebax Posted at 2015-12-15 17:25
'It's fairly common in the RC world to do multiple flights off single batteries after checking volt ...

I'm with you Geebax. I was shocked to find out that a CSC would actually work while flying. Seems like it would be an easy code fix. I'll be interested to see if anyone can come up with a good reason.

Barney

It would be good to have a set of questions to ask people who report their drone falling out of the sky. We already know to ask them about the battery state before take-off, but I would like to know if the lights were still on as it came down or after the crash, and what they did immediately prior to the engines shutting off.

I suspect this is not a battery fault, and the power is still on to the lights, and I also suspect, from the reports so far, that it most frequently happens when taking off, which is when the most power is drawn from the battery.

Geebax Posted at 2015-12-15 18:25
'It's fairly common in the RC world to do multiple flights off single batteries after checking volt ...

That all sounds pretty well thought out! I read a really, really long thread the other day by a couple of different people who've been doing some pretty serious testing - specifically over this very battery issue.

I think you and I are in the same boat. Something is happening (or *not* happening, as the case may be) that's giving me pause - no pun intended. As for why anyone would execute a CSC while flying:

(1) you accidentally landed in a tree, the battery did not disconnect and the motors are still trying to run
(2) You've had a moderately hard landing and the Phantom flipped over - not really flying now, of course.

#2 sort of happened to me the other day. Grazed a palm tree and fell into the grass. The Phnatom landed more or less totally upside-down - no broken or bent props, but the motors were still "humming" without the props turning, so I used the CSC to immediately shut off the engines. In the roughly five months since I've had my Phantom, this is the first time I've ever done that.

As for having to execute a CSC while you're actually flying? Unless you're about to smash into some other airborne craft, I don't know!

Art - N4PJ
Leesburg, FL

Art, I think the two examples you have given are very good, but it should be possible for the aircraft to work out what has happened in those two situations and shut off the motors. The 'upside down' scenario should be possible to detect using the internal sensors, and it should also be possible to tell of one or more motors are unable to rotate by examining the motor currents of each motor compared to each other. It appears the main board is able to read the motor currents. In that case, the aircraft should accept a CSC.

I don't think the case of colliding with another craft is reason to accept a CSC, an avoidance manoeuvre would probably get you out of that quicker, because the CSC takes a fairly long time to act.

The thing is,  the lights are not a good indicator of battery capacity.  LED lights use a very small amount of current,  even very bright ones.  Compared to the motors,  the lights are almost nothing.  I can easily envision a situation where the lights could be on and not have the power for the motors and other electrics.  The only way to get an accurate idea of the battery condition is by putting a load on the battery.  Just sitting on the ground at idle is not really a load.  I don't know if the software responds quickly enough to display this,  but I bet if you could monitor the voltage with motors off,  then observe as you initiate the csc,  you would see a large drop initially until the motors stabilized at idle.  Then on takeoff,  another drop but not as dramatic as the csc drop.  Electric motors typically use 4 times as much power to start as they do to run at a steady speed.  I suspect the nature of the li-po batteriess combined with a lack of any meaningful load test on first powering on the quad is going to be the culprit for the apparent 100% charge,  fall out of the sky issue.  If only a static voltage is being measured by the Dji-Go app,  well..... many types batteries exhibit this " voltage float "  false reading.  In fact,  all batteries I am familiar with will do so.

I don't think It's shutting down the motors, at least not in all cases.
I got a Landing message, same as my good friend Pyrre who was able to regain control and keep flying for eight more minutes.

Does anyone have Flytrex or similar "Black Box" data from a P3 for comparison?

Kneepuck Posted at 2015-12-16 15:59
The thing is,  the lights are not a good indicator of battery capacity.  LED lights use a very small ...

The fact that the lights are still on indicates that power was not shut off to the aircraft from the battery, it tends to suggest the motors only were shut down.

Geebax Posted at 2015-12-15 23:45
The fact that the lights are still on indicates that power was not shut off to the aircraft from t ...

Yes that was the point I was trying to make.  To illustrate,  I am sure many people here are familiar with the NiCad memory effect.  I had a cordless drill that had a small led light mounted by the collet.  After the batteries began to suffer from memory effect,  the drill would work only for a little while at full speed.  Then it would sloooowwwwllllyyy wind down until it was not turning at all.  However,  even though the motors would no longer turn,  the light was still on.  If I left the drill alone for a few minutes and then squeezed the trigger,  the drill would go fine,   for a few seconds.  Then it would wind down again.
This is a consequence of memory effect.  A normal,  fully charged NiCad will give up most all of its power and die suddenly when healthy.  No lingering slow downs.  Trying to fully discharge the battery fully by holding down the trigger for a long time would not work.  The battery will see the load and refuse to give up the current to make the motor run,  but the light will still be on.  In fact,  that was the method I used to use to recondition NiCads.  I would attach an LED across the terminals and let it sit for 3 or 4 days until the light went out.  Then a full charge and about 50% of the time the battery would be restored to almost like new.
I realize the chemistry is different with the LiPo batteries,  but there are still parallels that apply.
It seems to me that the P3 is sensing the voltage condition and deciding that it has not enough power to run the motors.  It surely has a circuit that is a threshold for motor activity.  If the voltage is too low,  then a device will try to make up the difference by pulling more current,  causing excess heat.  Using Ohms law illustrates this clearly.  A 100 watt light bulb,  for example, will light fine with 10 volts.  However,  it will require 10 amps at that voltage.  Supply it with 100 volts and it will need only 1 amp to light the bulb.  Both situations provide the same brightness, for a while.  But most 100 watt light bulbs will burn out quickly with 10 amps of current supplied to them. Components are rated for certain amounts of current and so,  a cut off is employed.  This is an oversimplification of the situation, but it will suffice for this instance.  It is certainly possible to control the behavior via software,  and I am sure that such a method is employed here.

My apologies if you already knew this about the NiCad memory effect.

Kneepuck Posted at 2015-12-16 05:59
The thing is,  the lights are not a good indicator of battery capacity.  LED lights use a very small ...

Heres a data set showing the voltage drops youre talking about:

A paraglider dude asked if I could snap some pics of him in the air.

I turned on the P2 and took off on his first attempt.
You can see two thigs on the voltage graf: The voltage drops and the spikes gets bigger.
Then I hovered while he prepared for one more go.
Now the voltage comes a little bit up again due to lower load from the motors.
Finally when he gets on his way a new voltage drop and more noise.

As for the memory effect : Never fully discharge any LiPo battery!
I reckon you allready knew this, but I'm still saying it so new folks dosent try somthing..

I've allways thought that 20% was the lower limit for wats healty on a LiPo but that might have improved over the years.

Thanks guys, very interesting. So the question remains, when the 'fell out of the sky' event occurs, have the motors shut off, or are they trying to lift the aircraft? A number of the reports I have read seem to indicate the motors shut off completely, so if that is the case, one would want to know why? Balthazar's graph shows some interesting correlation between battery voltage and speed, but it also seems to show that the motors were still running, even at reduced voltage.

Geebax Posted at 2015-12-16 22:34
Thanks guys, very interesting. So the question remains, when the 'fell out of the sky' event occurs, ...

Well, I've had a lot of Phantoms.
None of them have ever shut off in mid air without a user commanding it to.
And that actually  happened to one of my P2's on its third day.
Some of the artists on our rock festival wanted to fly a little.
One of them managed to stop it like five meters above gravel...
Shit happens, but from then on it was clear that the old P1 Basher days were gone.
Now with the Zenmuse and the cost the Phantoms have left their childhood

I still don't understand why they have the silly csc command anymore now we have multiple buttons on the controllers. Get rid of the silly stick movements and assign a combo of buttons or buttons and sticks so that the combination cannot be pressed by accident.

[RCG]Balthazar Posted at 2015-12-16 09:49
Heres a data set showing the voltage drops youre talking about:

A paraglider dude asked if I could ...

Interesting to see the relationship between the voltage and the battery temp.  As the voltage drops,  the motors pull more current thereby increasing the heat. You notice it is exactly the opposite of the first graph you posted in that respect.  Also the first graph is the opposite of what you would expect.  The heat starts high and gets cooler,  the voltage the opposite.   That kind of supports the theory of a floating voltage condition.  It takes current to make heat.  Judging by the temp line,  the heat was rapidly decreasing.  I have checked all my flight logs on the HealthyDrones site and in every case,  the temp steadily rose during the flight.  In fact,  you can not even charge the battery immediately after a flight.  You have to wait for the battery to cool down some or the charger will not start.  I don't see the spikes on the speed line as being anything more than noise,  possibly due to something like bad grounds or poor connections.  They seem to indicate poor data from the sensors, but I can not see how they would affect the motor shutting off.  Unless a poorly designed feedback loop or something similar.  Imho,  the issue is mostly firmware.
Dji states that the LiPo battery should be drawn down to 8 % periodically to maintain battery condition.  I think I read that in the battery manual.

It will be interesting to see what Dji determines is the problem.  If they tell us,  that is.

Geebax Posted at 2015-12-16 13:34
Thanks guys, very interesting. So the question remains, when the 'fell out of the sky' event occurs, ...

Running and running with enough power to lift the craft,  very different.  No way to know for sure I suppose.  And I sure don't feel like trying to recreate the problem.   Maybe it's alien tech.  Could be....

I suspect that if the probl4em is found and solved we will not be told. So far they have given no explanation for anything that is fixed. Giving explanations amounts to admitting there was a problem, and that means admitting liability.

Geebax Posted at 2015-12-16 18:52
I suspect that if the probl4em is found and solved we will not be told. So far they have given no ex ...

I hope you are wrong about that,  but sadly,  I suspect you are exactly right.

[RCG]Balthazar Posted at 2015-12-17 01:49
Heres a data set showing the voltage drops youre talking about:

A paraglider dude asked if I could ...

Hey Kneepuck, please forgive my ignorance as I am a newbie. Why should a LiPo battery never be fully discharged?

sifu128 Posted at 2015-12-16 19:47
Hey Kneepuck, please forgive my ignorance as I am a newbie. Why should a LiPo battery never be ful ...

Because of the chemistry involved,  over discharging the LiPo battery will damage the battery,  shortening battery life.  Permanently.

Kneepuck Posted at 2015-12-17 12:14
Because of the chemistry involved,  over discharging the LiPo battery will damage the battery,  sh ...

Understood and thank you!

Wonder if DJI's 8% really means 28%? or if LiPo's have improved this much.
Another battery tip for the experienced users:
If you do deplete a battery below the point where a normal computer charger is able to charge it
you can use NimH settings to carefully get it back up a little to save the battery.
Note: This will probably not work for the P2 and P3's intelligent batteries and you cannot trust the battery any more.
New users: This is also dangerous as doing it wrong might end up with a nasty LiPo fire!

I imagine LiPos are similar in this respect to Nicads, but the reason you do not discharge a multi-cell battery is because one cell will reach zero volts before the others and it then driven into negative charge, which is damaging to the cell.

Geebax Posted at 2015-12-17 09:07
I imagine LiPos are similar in this respect to Nicads, but the reason you do not discharge a multi-c ...

Makes sense, modern more accurate production making cells more equal so you can go lower perhaps, can anyone confirm please?

Boring Awesomeness.
Watching a ton o youtubes I just came across this video about LiPo's.
New pilots: Please watch it!
Experienced pilots: Well, I had to pause it, but I will watch the rest now.
Getting a little refresh on the basics cant hurt anyone

Well, it seems like the flight log files DJIFlightRecord_2015-11-22_[11-38-14].zip (18.75 KB, Down times: 29)

2015-12-19 Upload

Click on the file name to download attach can easily be uploaded to Flytrex.
I'll make a tut on that in another topic later and edit in The Link.Edit: Here's The Flight Data Tutorial
Here's 2 data sets from 2 different flights, 2 different P3P's from the same pallet at the same dealer the same day.

First off my missing P3P's final flight:


Look at the altitude graph!
Seems like the sensors think the bird was flying off to the skies!
This is not reflected in the Go app and I dont know what to make of it.
The Voltage Line confirms Geebax's thoughts about only motors shutting down as it gets back up a little on the end.
Temp doesent seem to be reported? Maybe Flytrex cant read the P3 temp data for some reason.
I'm getting a little vibe on compability sensors-craft-log-app here. The data needs chewing a little.

Then Pyrre's data.
He managed to recover, fly back then fly for another 8 minutes after the incident. Lets have a look at his data set:

Here we can see voltage drop roughly corresponding with altidude increase.
Followed by voltage gain somewhat corresponding with his drone dropping some 30m before his RTH command.
His speed graph is a mess not corresponding with the volt.
Temp doesent seem to be reported, or Flytrex cant read it.
Wish I knew more about this sensor data stuff so please help crack this for all of us!

[RCG]Balthazar Posted at 2015-12-19 01:00
Makes sense, modern more accurate production making cells more equal so you can go lower perhaps,  ...

In the old NiCad days, certain manufacturers, like Sanyo, made sets of closely matched cells available on a custom order. This was done to overcome the negative charge problem, but these days I think they rely of the balancing charge to gets the cells charged correctly. But it was always standard operating procedure to not fully discharge a pack if you wanted the pack to have a long lifetime.

Wow.  I can't believe how inaccurate the use of terminology is in that video.  His outcome is good and he can get along with what he believes is correct, but it is technically quite inaccurate.  It is good to have all the information in one place, but it would be nice to teach people the correct terms and definition to use.

AG0N-Gary Posted at 2015-12-22 16:57
Wow.  I can't believe how inaccurate the use of terminology is in that video.  His outcome is good a ...

Too complicated wordings will only confuse those most in need for the lessons in the video I think.
But do you know of a better or more accurate video please post it.

Sounds like you know your way around these things, can you confirm the lower limits of LiPo?
Is it 8%, 20% and is DJI's 8% really the 20% wer'e used to plus 8% margin on top?

[RCG]Balthazar Posted at 2015-12-23 04:25
Too complicated wordings will only confuse those most in need for the lessons in the video I think ...

That, I can't do.  I come from a 56 years in electronics, but LIPOs are new to me and I'm learning as we all are.  My only charging knowledge has come from Battery University.  I'll see if I can find the link, it is mentioned quite awhile ago in this forum and I think I reposted it at one time or another.  It is general in nature, and doesn't speak specifically to self managing batteries like DJI with built-in balancing, but does give you quite a basis for understanding some of this.  Look around that site, as there is tons of info there.

I'm currently discharging to about 10% when I feel it needs to 'rebuild' itself.  This happened last week when I pulled the batteries out and found one of them had less than 100% life left per the build-in check.  I charged it and it was still flashing the last LED, so I ran it down to 10%, cooled it, and charged it again.  It has been at 100% rating since then, but I will be careful with that one in particular.

Don't run them dead, and don't recharge at more than 1C rate for best life.  Store in the recommended range, I use 2 LEDs for the storage level.

As for any corrections to the video, it was a confusion between voltage and current, and the over-complicated explanation of how to determine rating and charge current.  His result is the same, but for someone who might want to use the knowledge gained there for something else, it will lead to an incorrect understanding of electricity.

Merry Christmas to all!

I've got decades of RC on my butt but the world moves forward specially when your'e not looking
The numbers puzzle me and I have to say I don't even think I know what % really means or how it's calculated.Guess it's time to relearn something.

And yeah! Merry Cristmas!

This is a copy of a post I made elsewhere:

I have a theory that what is happening is that taking off with a less than fully charged battery causes a high current drain on the battery because the motors are working hard to gain altitude. In doing so, the already partially depleted battery may suffer a voltage spike severe enough to take the battery down to a very low level, low enough that it dips below the voltage required to keep the control board running. When this happens, the on-board processor stops or goes through a reset procedure, and in doing so, the motors are shut down. This would also explain what djiphantom3xx reported, that when it hit the ground it switched itself on again. I would suggest it had not actually switched off, but went through a reset function.

The new 1.6 firmware feature of limiting the motor speed in cold weather would also tend to support this theory, as it would tend to prevent the sudden dip in voltage. Now I don't want to be cynical, but if I am right, then this new modification in 1.6 may be intended to address the 'fell from the sky' issue without leaving DJI open to claims.

Sounds like we have pretty similar ideas then