Dropgate - Inspires and Phantoms dropping out of the sky

Petr Hejl

Staff Member
Moderator
I did an experiment to see if I can get underneath the battery voltage drop issue with the Inspire 1, and actually got four batteries to drop. My article about it caused a bit of a stir on Facebook, the intention is not to trash DJI, but rather help fellow droners to avoid costly mistakes and dangerous situations. The article is here, what do you guys think?: http://www.dronevibes.com/2016/02/1...nspires-and-phantoms-dropping-out-of-the-sky/
 

maxx4wd

Member
I did an experiment to see if I can get underneath the battery voltage drop issue with the Inspire 1, and actually got four batteries to drop. My article about it caused a bit of a stir on Facebook, the intention is not to trash DJI, but rather help fellow droners to avoid costly mistakes and dangerous situations. The article is here, what do you guys think?: http://www.dronevibes.com/2016/02/1...nspires-and-phantoms-dropping-out-of-the-sky/

Nice article...I had this issue with an older P2 battery...luckily I was able to manually override and fly it into my hands before it just quit...also turned out it was losing a cell but prior to that this "smart" drain really does hamper these batteries...but of course if you charge them up right before use (or top them off) this should also not be much of an issue either. I have noticed that some batteries I've had of the smart variety often show much higher voltage then is actually present.
 

John Locke

Member
I did an experiment to see if I can get underneath the battery voltage drop issue with the Inspire 1, and actually got four batteries to drop. My article about it caused a bit of a stir on Facebook, the intention is not to trash DJI, but rather help fellow droners to avoid costly mistakes and dangerous situations. The article is here, what do you guys think?: http://www.dronevibes.com/2016/02/1...nspires-and-phantoms-dropping-out-of-the-sky/

I had conversations with DJI employees at CES about this "drop-gate" battery issue last month. I had two separate conversations, one at each of their two booths. I got the same story from both guys who were deemed the most knowledgeable about Phantom firmware in each booth. They said the problem is only in the 1.5 phantom motherboard firmware, which was quickly replaced 1 month after release with 1.6, which fixed this flaw in the MOTHERBOARD firmware, so they claim. Apparently the 1.5 motherboard firmware changed from 1.4 in a way that introduced a sensor flaw that didn't accurately read the battery voltage correctly. They claimed the firmware interpreted the voltage to be lower than reality, which they claim is the root of our DROP-GATE flaw. As an example, even though the battery voltage correctly shows 3.7V/cell in the Go App after flying a few minutes, the firmware would see 3.0V, which triggers auto-land, or worse, power shutdown if below 3.0V on any cell (these numbers are examples, not facts, but they are in the ballpark). These sudden shutdowns obviously made craft fall from the sky, which is a shocking experience for those unfortunate pilots that have experienced this. Although DJI has been honoring these incidents under warranty, it's still a major concern for all Phantom/Inspire owners.

Also in the 1.6 firmware release, DJI included new battery firmware which included a power governor that limits the current that can be sucked from the battery, hence minimizing dramatic voltage drop swings when you're on the last half of the battery. This governor seems to kick in at 3.6V, or about 30-40% charge in the DJI GO meter bar (according to my testing). Vertical ascent speeds at full throttle are normally 11mph, however when the new governor turns on this is reduced to about 7-8mph. Horizontal acceleration is also hindered in this case, but speed is still above 33mph, it just gets to that speed slower. This new battery firmware is an additional measure to mitigate dramatic voltage fluctuations, preventing premature auto landing, or worse, power shut down.

I don't understand why DJI hasn't FULLY FIXED this issue. I can understand why they want to shut down power, in case someone accidentally leaves the power turned on after the Phantom/Inspire has landed. I can easily see someone stowing away the craft in a case and forgetting it's on. So having a power shut down feature is good, WHEN THE CRAFT ISN'T FLYING. So why can't DJI create a requirement in the firmware that requires the craft to be grounded before power shut down occurs? They could do this just like pulling FULL DOWN on the left stick to shut off the motors, the craft must sense that it's on the ground before the motors will shut off. This sensing takes about 3 seconds to confirm and shut down, as experienced when you hand catch a Phantom3.
 
Last edited:

Petr Hejl

Staff Member
Moderator
Really good info, @John Locke, seems like some 1.6 people have these issues too (more on Inspire than Phantom). I am on 1.6, and my four battery test was done with this. I agree with you that the aircraft/battery should not be allowed to shut down mid-air, no matter what the capacity. I'd rather have a puffed Lipo that a pile of plastic I used to call my drone. I'm pretty sure that the aircraft can tell when it's airborne. I think they should also modify the battery firmware so it alerts you that it is in self-discharge cycle, and will not let you take off with the battery like that. That would seem like an easy fix to this, and keep at least few people from crashing their machines.
 

John Locke

Member
It would make more sense to me if they simply turn on all of the LEDs during discharge. Then we'll know. All LEDs should be on when discharging, 3 on when it's half done, and 2 on when it's close to done. Seems pretty simple to me.

It's also been tested and confirmed, the if you push the battery on button, this resets the clock that ticks down the 10 days before discharging (or whatever you have programmed it for). This is a flaw. Pushing that button should not reset the timer. DJI claimed they fixed it in 1.6, but this is false, it's been tested and it continues to reset the timer with a push of the button. Really dumb.

I recently compared history of two flights with the same Phantom, using two different batteries, one old with 24 charges and one new with 5 charges. What strange is the percentage bar reading differ quite a bit between the two flight when comparing total voltage at the same percent levels as the battery is consumed. Check these battery numbers out. This data was retrieved from the onboard flight recorder, .txt file. The 2-22 flight is the new battery (please don't hammer me for taking it below 50%). Why would voltages vary so much with at the same percent level? Seems odd.
upload_2016-2-24_18-4-32.png
 

Attachments

  • upload_2016-2-24_18-4-32.png
    upload_2016-2-24_18-4-32.png
    14.4 KB · Views: 2,586
Last edited:

Petr Hejl

Staff Member
Moderator
@John Locke , I agree that pressing the button to check the battery should not reset the timer. for some reason the picture does not open, can you, please, post full size, I'd really like to see it.
 




Petr Hejl

Staff Member
Moderator
@droneadmin When I log out, I see the images full size. I can only see thumbnails when logged in, and when I click them. I get the message that I don't have permission to view the media.
 


Gene

New Member
It would make more sense to me if they simply turn on all of the LEDs during discharge. Then we'll know. All LEDs should be on when discharging, 3 on when it's half done, and 2 on when it's close to done. Seems pretty simple to me.

It's also been tested and confirmed, the if you push the battery on button, this resets the clock that ticks down the 10 days before discharging (or whatever you have programmed it for). This is a flaw. Pushing that button should not reset the timer. DJI claimed they fixed it in 1.6, but this is false, it's been tested and it continues to reset the timer with a push of the button. Really dumb.

I recently compared history of two flights with the same Phantom, using two different batteries, one old with 24 charges and one new with 5 charges. What strange is the percentage bar reading differ quite a bit between the two flight when comparing total voltage at the same percent levels as the battery is consumed. Check these battery numbers out. This data was retrieved from the onboard flight recorder, .txt file. The 2-22 flight is the new battery (please don't hammer me for taking it below 50%). Why would voltages vary so much with at the same percent level? Seems odd.
View attachment 35

Not an expert here, but look at the temperatures. Wouldn't the higher starting/ending temps affect voltage?
 

John Locke

Member
Not an expert here, but look at the temperatures. Wouldn't the higher starting/ending temps affect voltage?
Yes, temperature does affect the efficiency of the battery, however above 75F is ideal. It shouldn't be any better at 99F. And lets say it did produce better power at 99F, why wouldn't the higher voltage last longer? The DJI power gauge seems to be more like a timer, not an actual measurement of current and voltage together (wattage output), in a way to prevent damage to the battery and maximize flight time. I makes no sense to me.

What's really interesting is the P4 numbers in comparison to the P3 numbers. It's apparent DJI has been more conservative with their battery gauge algorithm. With P4 the individual cell voltages are around 3.6V at 20%. The P3 at 20% you'd see 3.4V. I have a hunch with the issues DJI had with 1.5 firmware in P3 (craft falling from the sky), that the battery protection margins have been widened in P4. In cold weather this become an issue, so with a bigger margin of safety in the P4 battery, DJI has less risk. I would bet money that's why the P4 doesn't get the 28 minute flight times as advertised. I bet proto P4 craft were actually getting 28min flights, and they got all excited, so marketing ran with it, in error. Then engineering decided to widen the margin of safety to minimize risk (remembering P3's with 1.5 f/w), which brought the flight time down to ~21min. I truly believe with a f/w change the P4 could get 25min flight times (regular flying, not just a hover) without harming the battery at all. Damage doesn't occur in Li-Po batteries until you get down to about 3.1 or 3.0V.
 

Petr Hejl

Staff Member
Moderator
@John Locke agreed, I think that DJI raised that threshold to keep people who are "pushing it" out of trouble. Their battery gauge seems to calculate the remaining time from voltage, not the mAh used, so yeah, right about that one. I wonder if that new 10 pin connector helps keep things balanced a bit better, seems like each cell now has it's own connection to the aircraft (just a theory).
 

John Locke

Member
I don't think each cell has it's own connection. That would mean the smarts to monitor the voltages for equalization would be external, like a traditional dumb battery. But I know it's still in the battery. I believe all of those contacts in the "reversible connector" are duplications, solely to conduct enough current. Notice the P4 battery doesn't utilize the big high current blade contacts like P3. So the only way to withstand the high current needed is to use multiple small contacts because those little pins aren't big enough to conduct enough current without overheating. I assume that design choice was made to facilitate a reversible connector design for some reason. Although it's reversible with the charger, it's not reversible when installed into the craft, so I'm not sure why they felt it necessary to make the charger connector reversible, instead of keyed. I do like the reversible feature, it's just one of those little things that don't matter much to me. DJI put a lot of engineering effort into accomplishing the feature, IMO, seems pointless..... but I do like it.:rolleyes:

I'm wondering it the new Go app 2.8.0 addresses these battery meter inconsistencies. The change notes aren't clear about optimizing "battery consumption".
 

clararscribner

New Member
A voice alarm device (also called a sound alarm) is a kind of alarm signal used in dangerous places to send warning signals to people through sound and various light. The explosion-proof acousto-optic alarm is suitable for the explosive gas environment which is installed in the T6 temperature group of class II C, and can also be used in the 1 zone and 2 zone explosion-proof places with explosion-proof requirements in the petroleum, chemical industry and other industries, and can also be used in the open and outdoor.
 

Top