Almost crashed: battery failure suspected.

[Nerius]

Hi All,

It was sunny day, not a strong wind: all normal conditions. It was first flight on that day, had a 12 satelites, home point set, everything seems to be OK. In the beginning was flying at low altitude and till about 50 meters away: as usually tested everything, then started to fly normally. Wanted to make video of old mound in the middle of field. Made few circles, then started to ascend, descended and started to ascend again. At altitude more than 100 meters, started to fly away from mound: wanted to descend and make a circle at lower altitude. Started slowly descend and suddenly bird was started to drop like rock (battery was at about 60%): pushed a Up stick fully up and bird stabilized, but started to fly away from me. Switched to atti mode, then back to gps and it stabilizes. When I saw it is stable, released up stick and started to descend again, after some time: same situation and same solution, only altitude was lower. Battery suddenly dropped down by about 10%. Was trying to bring it to me as close, as possible and the for a third time: same situation. Bring it a little higher and then finally bring back without any more accidents. Decided to test with a two other batteries: second and third flight was without any problems. When at home, connected phantom to computer, dji phantom app shows battery is OK.

So, I'm suspecting a battery failure and thinking about to retire it, what di you guys think ?

This is the video of accident:

and this battery info:

 

[MapMaker53]

My initial impression was it looked like a number of bird strikes, but you would think the jarred camera view would have caught a glimpse of the bird doing the attacking. Interestingly enough though, at 1:40 into the video you see what looks like a hawk flying toward it from below.

I don't see how a bad battery could make your Phantom suddenly fly off away from you. I'm thinking an intermittent compass problem or loose compass wire connection that suddenly mess up how the Phantom is interpreting X, Y, and Z orientation. ???

 

[Nerius]

My initial impression was it looked like a number of bird strikes, but you would think the jarred camera view would have caught a glimpse of the bird doing the attacking. Interestingly enough though, at 1:40 into the video you see what looks like a hawk flying toward it from below.

Yes, I see a bird at 1:40, but I was not so far from phantom and clearly saw all 3 drops: there was no birds flying close. Anyway, thank you for suggestion.

 

[Desert Dave P]

I think you answered your own question, 43 flights on a battery seems like a good life, other batteries work fine, i'd put my money on a failing battery. if your feeling brave you could try it again the same battery a few times and see if you can get the same result.

Dave

 

[MapMaker53]

A continued problem with the same battery still doesn't mean the problem is with the battery. Makes more sense (and probably safer) to fly with a different battery and see if it happens again. If it does, then the problem isn't the battery. If it doesn't, then I would suspect the old battery. Maybe a momentary loss of power caused it to lose all satellites for a few seconds and kicked it into ATTI mode despite the switch position and the wind started taking it. With virtually no friction, it doesn't take much wind to have it start drifting away.

 

[Mako79]

Battery failure. Don't trust the DJI batteries, They are NOT TRUE 3S cell batteries.

 

[N017RW]

Battery failure. Don't trust the DJI batteries, They are NOT TRUE 3S cell batteries.

Not sure what 'true' means.

xS-xP LiPo packs at very common especially to achieve high capacity in limited space applications.

 

[Mako79]

Not sure what 'true' means.

xS-xP LiPo packs at very common especially to achieve high capacity in limited space applications.

The P2 has 3x2 batteries configuration. That means 6 batteries, in a 2 parallel x 3 series. This is bad for balancing.
I'll use the Dog sled analogy.
You have 6 SMALL dogs in a 2+2+2 config which can drag the same as 3 LARGE dogs in 1+1+1 config (true 3S).
With the 6 small dog config, when you stop and feed them (charging), sometimes one dog can eat more then the other but you don't notice this as it can still carry the other dogs weight (masking the failure). Its only when you push the dogs, you notice the failure. This is the sudden drop in %.

The P2 phantom assist cannot report each individual battery as it is in a parallel. This means in the 3x2 config, it masks the problem and cannot be detected until in mid flight under load. And with 6 batteries, any of which can go wrong, there is higher chance of failure.

I bet if you pull apart a Limefuel, you will see 3 batteries and NOT 6 halves like the DJI's.

On the other hand, the P3 has True 4S cells = Total of 4 batteries. Each battery is reported truly and errors cant be masked. All the batteries are 'fed' evenly and this is why some users are getting 100+ flights.

 

[N017RW]

The Phantom's battery reporting system notwithstanding your fears are not borne out by manufacturers.

There are legitimate reasons for the type of S-P construction DJI used. As I mentioned one is size v capacity.

Calling one true may indicate you were not aware of the construction of the cell. Nowhere did DJI state it was 3S v a 3S-2P.

The multiicell analogy has a flip-side.
An airplane that has 4 engines has a greater chance one will fail but you have more redundancy.

In a series only pack if one cell develops a problem there's no redundant current path, in a series-parallel there is.

Resistance in parallel is reduced below the lowest individual value. Two cells in parallel will have a lower series equivalent resistance. ( 1/Rt =1/R1 + 1/R2...)

LiPos are matched at manufacture and their Ri is fairly flat across their lifetime. Having said that the Ri does vary a bit more at different SOC.

All engineering exercises contain pros/cons, trade-offs, etc

If you want to really look into the issues with DJI batts. look into the LiPo charging algorithm.
Monitoring pack voltage during the charging it's clear they are not using the CC/CV method.

 

[Nerius]

Thank you all for suggestions. Made a few flights yesterday with another batteries: no problems, so decided to recycle this one. Too much risk. Anyway, so-called "Smart" battery isn't that smart: big probability, that it cannot detect some very dangerous problems and Iife-cycle could be much better also. I always store batteries at about 50% charge, after flight recharge them to 50% immediately when returning home (but not quicker than 1 hour after flight) and charging to 100% only in the morning of flight day. 48 cycles is way to low. 110Euros for battery, it's about 2.3 euros per 15min. flight .

 

[N017RW]

Agreede the battery hardware has it's problems but it does make LiPo use safer for the 'everyman' who wants a flying camera.

 

[yorlik]

I cannot see that as a battery issue at all either.

A battery disconnect for a very short time probably resets cpus and gps. My gps takes a good 3-4 minutes to gather sats, so a batt issue causing it to reset would mean a crash WELL before it could recover.

2nd scene showed your compass screwing up pretty good; back and forth until it took the nose dive to one side.... what does your flight log show compass doing at this time?

If you sit and watch compass values in assistant for 5 minutes, is it bouncing around just sitting on the table?

What does your flight log show? any jumps/drops in battery voltage? Does it show going out of GP into ATTI on that last scene when it began to run away? Want to share it here for us to examine?

I think it is a shame you retire that good battery. I suspect the issue will resurface sometime later on another battery since it really does not seem like a battery issue.

And assistant calculated both LIFE and % charge right on, so no math issues there.

Although I too disapprove of the parallel cells, your assistant cell voltages look way too close to likely have a bad cell. When dji engineered these P2 batts, the parallel cells were the only economical way to get the 5200mah in that size; today, we can buy single cells that fit same space as the parallel ones and output over 2000mah for 6000+ easily and economically. That is why some of the new non dji batts are showing up with the higher ratings now - for a product no longer in production.

 

[Nerius]

I cannot see that as a battery issue at all either.

A battery disconnect for a very short time probably resets cpus and gps. My gps takes a good 3-4 minutes to gather sats, so a batt issue causing it to reset would mean a crash WELL before it could recover.

2nd scene showed your compass screwing up pretty good; back and forth until it took the nose dive to one side.... what does your flight log show compass doing at this time?

If you sit and watch compass values in assistant for 5 minutes, is it bouncing around just sitting on the table?

What does your flight log show? any jumps/drops in battery voltage? Does it show going out of GP into ATTI on that last scene when it began to run away? Want to share it here for us to examine?

I think it is a shame you retire that good battery. I suspect the issue will resurface sometime later on another battery since it really does not seem like a battery issue.

And assistant calculated both LIFE and % charge right on, so no math issues there.

Although I too disapprove of the parallel cells, your assistant cell voltages look way too close to likely have a bad cell. When dji engineered these P2 batts, the parallel cells were the only economical way to get the 5200mah in that size; today, we can buy single cells that fit same space as the parallel ones and output over 2000mah for 6000+ easily and economically. That is why some of the new non dji batts are showing up with the higher ratings now - for a product no longer in production.

Still believe it's hight probability of battery fault: flew with remaining two batteries 6 times after accident- everything is OK. Log could be useful, but i think DJI Vision app did not provide it: only thing I noted, battery dropped suddenly during bird flight on failure when watching on iPhone. Also, voltage without load is OK, but it not means that it will be OK during heavy load (bird ascended to more than 100 meters (300 feet) before failure): one electrical engineer told me, that to proper test battery we should always test it with load, without load voltage could be OK, but it not means it will be OK with load, especially for longer time.

Anyway, better safe, than sorry: thank you for suggestions, battery recycled

 

[RoyVa]

Nerius , smart move to retire that battery. Have had simular issues with one of mine. It would show up good in the assistant but in flight at 65% it would drop rapidly to 12% which would initiate a return to home low battery. Then a few seconds later at 8% it would go into the critical battery level and go into the autoland mode. Things were happening so fast one would think it was doing crazy things, trying to go up to altitude then comming right down. Took me a couple of flights on that battery to figure it out with a couple of auto landings and a couple of new props from dropping down in the woods. Thank goodness for the "find my Phantom" setting in the app. Drove me crazy for a couple of days sorting it all ot. Thought it was a problem with the bird when all the time it was the battery doing like yours did.