Let's start posting dji assistant BATTERY data, HERE!

[yorlik]

Mr. Rich, I saw big thread in rcControls forum bout batts and how to 'reset' them. so thought I would try with your defective one (in assitant @ 79 cycles says bad cell in pop up).

did the reset (power up with the equalizer plug off, then plug in) course no difference.

So I thought, what the heck, lets try another scheme.... Since the balance circuit is 37ma less than the charge rate, and the charge rate is 4000+ma, it is simply not adequate if cells get out of synch. so I decided to charge your batt at 200ma: so the 37ma balance is WAY more significant. I took your bad batt from 1500mah to its capacity (2700mah) at 200ma charge rate. all 3 cell sets charged and balanced best they could. at end of many hours, 2 sets were within 2mv each, 3rd was within 10mv. STILL HAD BAD CELL POP UP!@#$#@!

So either this pop up is result of a one shot bit set in EPROM or 10mv is too far apart....

I will next artificially up the 10mv low cell to same as the other 2 and see if the pop up goes away; I expect not. I really think their BMS chip burns a bit to post the pop up and it never goes away.

BTW, your batt again confirmed that CHARGE NOW is simply [100*capacityNow/capacityFull], and %life is simply [100*(300-cycles)/300] - for those folks who were wondering how those 2 % numbers were calculated. I have verified these over many batteries.

 

[Gordy Kehler]

Here is the data from my 5 batteries after a full discharge. Haven't done any flying with them yet. Prior to the full discharge, my bird would experience a power shortage at about the 60% mark and experience Vortex turbulence in its descent. Would require 150' or more to prevent the bird from touching the ground and only having limited vertical power. I am hoping the full discharge will have reset the batteries to give me consistant power throughout the flight. Usually stop flying in the 20 - 25% battery range.

 

[miket234]

Got 2 new batteries. Both arrived with pretty much zero charge. Here is the data from a brand new freshly charged up battery!! Think I expected the voltages to be closer together.......

 

[yorlik]

Got 2 new batteries. Both arrived with pretty much zero charge. Here is the data from a brand new freshly charged up battery!! Think I expected the voltages to be closer together.......

Ouch. From what I have learned of lipo battery life, if your new ones arrived with "pretty much zero charge" you probably should have paid about 10% of the listed new price for them. I would have my doubts they will survive to 30 cycles... And you are right, it appears you already have at least one bad cell being off 10mv already!

Not sure what "pretty much zero charge" really means; hope you put each on assistant and took snap shots of how they arrived. if less than dji min recommendation of 8% I certainly would be sending them back for full refund or replacement.

 

[miket234]

Was not enough juice in either battery to fire up the Phantom to see what the assistant said!!

 

[yorlik]

Here is the data from my 5 batteries after a full discharge

Hey Gordy, I am thinking your pdf MAY have really useful data for me! But as a pdf no page has enough data to be of any value by itself. I need on a single line (ROW) cell voltages, battery capacities, & dji assitant reported % charge.

I wonder if maybe this data is really a spreadsheet with all that data on each line? and you printed it to 10 individual pdf pages that are all unrelated to each other?

If so, could you send me your spreadsheet instead? Or maybe tell me how to arrange with cut and paste of the printed out 10 pdf pages?

Thanks!

 

[Gordy Kehler]

Managed to discharge all my batteries to 0 to 4%. After full recharge worst 2 batteries used and have full control and range fm 99% to 30% area using OXOXO props. No drop off at 50 or 60% and no crash. Yippee

 

[yorlik]

hey gordy, i PMed you with my email address. can i have that xls file?

 

[yorlik]

First, thanks to everyone who sent me data to compile along with my own. Gordon's last batch sent this weekend allowed me to finalize my study on the Phantom 5200Mah battery.

Bottom line I find from this is: Fly below 40% battery left on DJI gauge (as shown on battery lites, in assistant, or as shown on phone app while flying, and you have a very much higher chance of crashing or auto landing without enough warning to bring it back.

Here is my final chart. Anyone wanting a copy of the xls spreadsheet can have it.

The orange line is % charge as reported on a given battery, regardless of its FULL capacity; the yellow is more normalized since it is % charge compared to 5200Mah FULL.

% charge goes down pretty linearly from 100% to about 40%, having used 3000mah (60%) to this point.

From 40% reported charge down to 20% charge, the reported % drops twice as fast as the first 60% usage.

Using 14 minutes as a typical full battery flight (makes sense from reported flight times as well as the math: 5.2ah battery / avg 22amp draw, * 60 = 14.2 minutes capacity), the first 60% will be used in 14*.6= 8.4 minutes.

So at 8.4 minutes into your flight, you should expect to be at about 40% capacity left...

But then drain from 40 to 20% happens twice as fast, so this 20%, instead of giving another 14*.2 = 2.8 minutes, only gives 1.4 minutes more flight time.

Then take it below 20% and see how fast it drops: almost a straight line down to that magical 8% value where the battery will auto shut off!

So you can fly pretty safely for 8-9 minutes, then better be watching closely because the next 2-3 minutes will either shut you down or drop you out of the sky at 3-4 minutes more.

I need to go back to work, so will not clean up the last minutes comments above; it obviously needs some tweaking. But it should be sufficient to get across the idea most of us deep down realize but hate to admit: fly longer than 8-10 minutes and you get into a gray area.

The good news is DJI does not let the average battery voltage per cell go too low until below about 20% (3.7v), so flying above 20% left should save our batteries from damage from discharging to too low a voltage per cell. Not sure if this is REALLY true since flying pulling 25amps really knocks the actual cell voltage down another .2-.3v.

 

[yorlik]

I got back into testing due to conversations with folks over on the inspire forum here.... they too have the same fly down to something from 60% and less and sometimes the battery drops to 7-8% in a few seconds...

So I revived my testing here on Rich's old battery (thanks Rich). I replaced the bad cell 1 pair with a decent pair, added a couple #12 wires to the batt output connector (battery cover is off), and charged up fully and plugged into my P2v+ and turned on.

I can now hook my smart charger to the 2 batt leads and the dji smart batt thinks the total load is the P2. I set my smart charger to 7.5amp discharge, that plus the drone's .9 amp discharge give a constant 8.4amp discharge for testing on the workbench. Nothing gets hot, but I have a fan blowing on the p2 anyway

I CAPTURED THE WHOLE DATA FOR THE DREADED INSTANT DROP IN CHARGE AS REPORTED BY DJI!!!!!!!!!!!!! ON VIDEO.

I am struggling with the data to make sense of it and plot it in a way that shows what the issue is....

I can see in about 5 sec the % charge drop from near 1000mah to 300some mah! Normal drop at this discharge rate elsewhere in the cycle is - in round numbers - about 1% and 10mah! So I've captured the false info!! But I cannot figure out what it really means yet... Here is a summary chart showing what I captured:

I would be happy to share whole spreadsheet with anyone who wants it....


Basically, the blue line is the battery voltage, the orange is the % charge left per dji. Excellent dji smart battery coding to linearize the actual voltage so well. The X axis is the mah (on the bottom), and so dji does a great job linearizing the mah vs actual battery voltage in a way that makes perfect sense.

The issue is shown in the spreadsheet in read where the reported mah charge left drops by 100mah (instead of say 1mah as it 'should') over 5 seconds, and the yellow curve is time. Of course I am not in warp speed, so I am not bending time like that so it shows something wrong, but I cannot get my spreadsheet chart in right mode yet to see it. Where the yellow timeline is horizontal, it is saying time is standing still...

Obviously the straightline needs to be time, and the rest needs to bend with this faulty firmware code....

It is miller time here so I quit for now... maybe the anwser is to change the X axis from mah to time? that would linearize time from start to finish, and then we would see the descrepency in the other curves?

Bottom line is the voltage is very very constant in its drop - even thru the firmware glitch. So the glitch is the mah reported dropping by 100's digit instead of 1's digit 5 times here. I suspect a math error that creeps up down under 60% occassionally to folks that rather than decrementing 1's mah digits, it decrements 100's mah digit instead by mistake.... It appears this is exactly what the firmware fault is, but until I or someone else helps get this chart axis detail corrected, it won't be so obvious.

I would report this to dji, but history shows they will 1) ignore it, 2) say it is an obsolete product so who cares - even tho the same glitch appears to be happening in ALL the products - inspire and P3 too, and 3) they will just delete my post to them as they have done in the past on their skd site.

Anyone up for looking at this? I need a spreadsheet wizard!

spreadsheet: https://app.box.com/s/nyfryzpyxowfjid1cvm31peoq8j3o9uk

 

[yorlik]

I redid the chart with TIME as the x axis.... now it shows what the smart battery glitch clearly:

This is the dreaded "my battery quickly dropped from good x% to 4-8% instantaneously. x is reported by folks on p2, p3, and inspire forums as anywhere from 20-60%

It is quite obvious from this chart, showing the battery output voltage constant during the glitch, and with confirmed constant 8.4amp discharge rate, that it is a glitch.

Even after the glitch stabilizes at 5-ish%, the discharge mah and % charge go right back to normal! Same exact discharge ramp as before! PURE GLITCH!

Nothing in the individual cell voltages even give cause for alarm or reason for this glitch. Here is the actual cell1,2,3 voltages during the glitch:

The only other explanation that might fit, besides a random glitch, is if their firmware has built in routine that FORCES the %charge left to be 8% (that magical number again) @ 3.6v per cell.... IF they have such a subroutine in all their batteries, it would totally account for this mystery. Obviously bad programming, but I can see them coding that as their 'run battery down to 8% every 20 cycles to resink mah to % charge number.... It is easy to code something like this ini one room, while guys in other room are flying it, and never see the hidden glitch surface - because they are using brand new good batteries!

But even THIS bad subroutine for 8% is testable; we have the data. We need to plot the same thing AGAIN with a battery that does NOT glitch and compare the results.... I have video of this same battery going all the way down to 8% without a single glitch (it was at 1amp discharge rate)... I will plot it next and compare the results. That should confirm if this is a coded 8% subroutine issue or not...

Everyone, feel free to add ideas!

 

[yorlik]

The only other explanation that might fit, besides a random glitch, is if their firmware has built in routine that FORCES the %charge left to be 8% (that magical number again) @ 3.6v per cell.... IF they have such a subroutine in all their batteries, it would totally account for this mystery. Obviously bad programming, but I can see them coding that as their 'run battery down to 8% every 20 cycles to resink mah to % charge number.... It is easy to code something like this ini one room, while guys in other room are flying it, and never see the hidden glitch surface - because they are using brand new good batteries!

Hmmm.... Comparing this instant drop 'glitch' to a few dozen other compiled data lists of my and other's battery data as reported by assitant after flights, I think a good theory is that dji has a subroutine that kicks in around 10.8v level that tries to correct wrong mah/%-charge levels and kick them down to 8% artificially, instantly, as the magic 8% correction level...

Compilation of many many normal battery readings show 8% should result around 11.0 to 11.1v and normal battery readings show 10.8v should result in 0mah and 0% left....

So here I come with a battery dropping down to 10.8v and the subroutine kicks in, instantly dropping my mah & % charge down to do the magic 8% level correction.... My theory is this is a fine subroutine, but it should never be reached during flight... perhaps another subroutine that says "IF batt volt<11.0v, then Suggest LAND SOON, battery unsynced" or some such.... don't let it go until this drastic subroutine runs before reporting an issue.

Back to the data board....

dji - want to make comment? Give me $50,000.00 for finding this for you? Disagree?

 

[RichWest]

Yorlik - Glad to read that you're still on the trail for more discoveries and more importantly answers. I'm doing similar discoveries with another platform, not batteries however. Thanks for sharing the information and the education. I'm not walking down the battery trail, but the processing you've done is a helpful example of the efforts required to analyze and then hopefully solve a problem. Others will/have benefit from this exercise.

 

[yorlik]

thanks! please share your cause - I would like to see!