Mavic - Battery life improvement & Internal Electronics

[CyberNate]

The best ones at the moment are 3000 or 4000 LiHV Multistar ones (Hobbyking), used as a single battery.

How are you mounting a single pack? Are you putting on one side? Do you use the 3m lock strips, which are much stronger than Velcro?

 

[Arkarian]

Hi guys !

I'll be a Mavic owner soon, and I'm already looking for battery improvement.
Depending on how much risks (about having a dead battery) we're willing to put on, I think there is a way to replace LiPo cells by Li-Ion cells, still using the same battery circuit board. Picture of the board there.

After some calculations (mostly based on internet researches), there is a way to improve mAh capacity by 50% by adding ~50g (battery weight estimated of 280-290g).
It's about swapping the 3S1P (which means 3 batteries in series) original LiPo, by a 3S2P (which means 2 parallel circuits of 3 batteries in serie) which is 6 cells of NCR18650PF Li-Ion.
The increase would be going from the original 3830mAh 340g, to a 5500mAh ~380-390g (depending of the circtuitry's weight).

One of the advantages is that we would be able to charge the battery using the tradidionnal charger, and no risks of surtension by branching the battery externally like some other mods are offering.
The negative side is that we'll need to get rid of the hard case, the batterry will be a little bit bulkier.

There is some data research needed to be sure that the calculations are correct and the improvement possible.
Still not having yet in possession the Mavic, do some owners are willing to share some .DAT log files of a full lenght flight to make some checks about tension, current, min/max draw... ?
And other test would be to study data log for a full flight with an added weight of ~100g on top to compare current draw and flying time evolution.

Looking forward to see if some of you are interested to try going on this way, have a nice fly !

PS: I'm not under FAA/US regulation and always flying LoS or FPV, just want to improve flight duration as safely as possible and having fun proving that we can improve consummer goods.
Be responsible and remember that local regulations apply ;-)

Just get extra batteries. I have 4. That's a lot of flight time.

 

[Crypto]

Just get extra batteries. I have 4. That's a lot of flight time.

Yes you are true about aving more batteries = more cumulated flight time, but this isn't only about having more of these.
It is about having the opportunity to improve existing and having fun with electronic and computer science, which is my motivation behind this. Even if I'll try and finish by destroying the battery, at least I'll know that I've tried instead of later regretting never trying it.

Autodidacticism and having more understanding about a new technology through (partial) reverse engineering is the motto followed to fulfill myself.

 

[Lsv]

I think you should be careful with high temperature

 

[Oingo Boingo]

You might run into a few issues. One being the battery time remaining indicator. The DJI cells have small circuits on that tallk to the processor and give flight times. Replicating this might be a bit of an issue

 

[CactusJackSlade]

Interesting thread... thanks OP

 

[Kilrah]

You'd be discharging a 2P pack of 18650s at 3C constant (per cell), up to 10C. Firstly they won't like it at all, secondly any extra energy will be dissipated as heat in them at this rate since they're not made for high discharge applications at all.
Also note that if you extrapolate the discharge curve to 3-4C and consider the Mavic's battery circuitry will cut dead at 3.1V/cell you'll have at best the same usable capacity as the original battery, being able to use only about 2Ah per cell. And they'll likely overheat.

 

[Arkarian]

Watch out for thermal runaway.

 

[Arkarian]

Yes you are true about aving more batteries = more cumulated flight time, but this isn't only about having more of these.
It is about having the opportunity to improve existing and having fun with electronic and computer science, which is my motivation behind this. Even if I'll try and finish by destroying the battery, at least I'll know that I've tried instead of later regretting never trying it.

Autodidacticism and having more understanding about a new technology through (partial) reverse engineering is the motto followed to fulfill myself.

I don't dis you for innovating but these batteries are very technical. The charger is actually part of the battery. The power pack is not the charger.

I noticed the Inspire has two batteries and still about the same flight time.

 

[CactusJackSlade]

I don't dis you for innovating but these batteries are very technical. The charger is actually part of the battery. The power pack is not the charger.

I noticed the Inspire has two batteries and still about the same flight time.

Are you talking the Inspire 1 (single battery) vs the Inspire 2 with the dual batteries?

 

[Robbyg]

I like the idea that Crypto has come up with. I am all for someone trying out a method of getting more battery life using the existing battery shell. The software in the battery pack might not be all that hard to modify to make it work for different cells. I just cannot imagine that it's a huge amount of code and if you can obtain an emulator for the micro inside it might be very doable with code changes and a few resistor changes and a system of monitoring the data stream during bench testing.
My main concern would be safety. If your going to do this the modified battery pack has to be charged and stored in a safe location at all times. Good luck and please keep us informed.

Rob

 

[Crypto]

I like the idea that Crypto has come up with. I am all for someone trying out a method of getting more battery life using the existing battery shell. The software in the battery pack might not be all that hard to modify to make it work for different cells. I just cannot imagine that it's a huge amount of code and if you can obtain an emulator for the micro inside it might be very doable with code changes and a few resistor changes and a system of monitoring the data stream during bench testing.
My main concern would be safety. If your going to do this the modified battery pack has to be charged and stored in a safe location at all times. Good luck and please keep us informed.

Rob

Unfortunately, trying to change a software parameter can be very difficult, so it'll be only the last option to consider.

Li-Ion batteries go much lower than 3.50/3.20V or so, but the DJI bird doesn't see this. It just sees that it is at 3.50-3.20V and needs to land. I have tried with li-ions (18650) on both Phantom and Inspire, with very bad results. You can try, but you will probably be disappointed.
So the draw and current do not really matter, it's the voltage left that will stop you from flying.

Maybe some tests have to be run first to know if the Mavic is stopping by using the lowest voltage on a single LiPo cell or the total amount of volts still in the battery pack. For the second option an hardware solution can be to use 4S1P instead of 3S1P to go on lower voltages for Li-Ions in comparison with LiPo (+another hardware added to trick the results of the voltage measure of singles cells).

You'd be discharging a 2P pack of 18650s at 3C constant (per cell), up to 10C. Firstly they won't like it at all, secondly any extra energy will be dissipated as heat in them at this rate since they're not made for high discharge applications at all.
Also note that if you extrapolate the discharge curve to 3-4C and consider the Mavic's battery circuitry will cut dead at 3.1V/cell you'll have at best the same usable capacity as the original battery, being able to use only about 2Ah per cell. And they'll likely overheat.

The Mavic is hovering around 8 to 9 Amps. Using a 3S1P of Li-Ions at 3400mAh will represent around 2.5C, but a 3S2P will go under 1.25C. I agree that using Li-Ions will be better if we stay in low Amps range and not use sport mode at all.
DarXidE says that he was using Li-Ion cells in average between 2 to 4C in an other project and had problems over this rate of discharge. By using 3S2P on a Mavic that's peak around 20 to 24 Amps at full throttle we'll stay close to 3C.
But this problem can be put aside and solved by using 18650PF instead of 18650B (which are accepting 4x more C).

 

[FredzMaxxUAV]

@Crypto - I have done the tests with LiIons on the Phantom, the Mavic and the Inspire. Otherwise I wouldn't write about them
They perform worse than LiPOs, as I wrote, because of the voltage difference.

 

[Arkarian]

Are you talking the Inspire 1 (single battery) vs the Inspire 2 with the dual batteries?

No compared to the Mavic the Inspire 2 only gets about 6 more mins with two batteries. It's more about redundancy than extending flight time.

 

[Crypto]

Great news !
I found (coincidentally) someone on internet which is using Sony Li-Ion 18650 3000mAh batteries on Mavic Pro !
I'll post all the info I've found about it there later in the night.
Just for thought, he is claiming obtaining 20min on a single 3S1P pack (but didn't know yet if there's any voltage hack).

 

[Crypto]

An other way I was thinking about is using an Arduino Leonardo (or Raspberry Pi) to overcome the problem of the voltage sensor in the Mavic.
Just converting Li-Ion % to LiPo % by measuring voltage out of Li-Ion batteries and increasing it to make the Mavic think it's using the same voltage range.
For example: 100% to 0% LiPo which is 4.2 to 3.6v/cell converted into 100% to 0% Li-Ion which is 4.2 to 2.8v.
This can be easily feasible by using analog input and output with custom software and basic electronics (resistor and capacitors).
It only depends on which data is using the Mavic to determine % battery left (cell voltage or total voltage).
This solution will not need any internal/software changes on the Mavic itself nor the original batteries.

____________​

Now let's talk about the informations I've found about the guy which is using 18650 batteries.
There's some picture extracted from his Youtube videos :

And now the videos links :

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

​

He is using 3S1P Sony US18650VTC6 3000mAh (Specs here)

I've contacted him to know if he would like to share his work.

 

[Natedoggz]

Let me know!

 

[aklvis]

Just looking at the 3.3v rail on the Mavic Pro battery. When the battery is upside-down and the connector is closest to you then the 3.3v rail is the left-most pin. With the negative rail as the ground I am seeing on my oscilloscope a data stream. This occurs every second, but the data is not repeated every time. I guess this must be the authentication stream from the battery to the mavic itself. I haven't got my logic analyser going yet so I'm not entirely sure what the data is. Either the Mavic interogates the battery and the battery returns a complimentary sequence to or the battery just sends out a sequence every second that conforms to some rule and then the Mavic takes it as authentic.

Once we know the data sequence we may just be able to have an arduino pump out that same data stream to fool the Mavic into thinking the larger aftermarket battery is authentic.

 

[BudWalker]

Just looking at the 3.3v rail on the Mavic Pro battery. When the battery is upside-down and the connector is closest to you then the 3.3v rail is the left-most pin. With the negative rail as the ground I am seeing on my oscilloscope a data stream. This occurs every second, but the data is not repeated every time. I guess this must be the authentication stream from the battery to the mavic itself. I haven't got my logic analyser going yet so I'm not entirely sure what the data is. Either the Mavic interogates the battery and the battery returns a complimentary sequence to or the battery just sends out a sequence every second that conforms to some rule and then the Mavic takes it as authentic.

Once we know the data sequence we may just be able to have an arduino pump out that same data stream to fool the Mavic into thinking the larger aftermarket battery is authentic.

I suspect this may be the battery info signal that ends up in the .DAT. That signal is 1 HZ. If so the record should look like

0x55, 0x26, 0x00,0xB0,0x88,0x13,.....

The length of the record should be 48

 

[ScottJD]

Yes, my first though was also to use Panasonic NCR18650B, but as you said there are rated for a max between 1.5 and 2C (which means between 10 and 15Amps).
It's why I proposed different NCR18650PF which are supporting much more current draw, up to 4C (which means 20A for a 3S2P), but with a little bit less for gravimetric energy density at 207Wh/kg instead of 240Wh/kg for 18650B.

I've made some other calculations, for cell type comparison, for a 3S2P:
18650B : 295gr + circuit board, for 6600mAh (@15A peak)
18650PF : 288gr + circuit board, for 5650mAh (@30A peak)

Analysing logs of Mavic with weight addition will make us know how much current draw and peak it's using, and will let us choose between NCR18650B and NCR18650PF.

It is very interesting that you used 18650B at 2C, did you tried to go further than 2C ?
What was the result and conclusion of using a battery bloc at a 2C average for 30min ?

The PF cells are also good cells, they are used in this Ryobi 24V pack for my weed wacker.
Ryobi OP241A 24V 57.72 Wh Lithium Battery Pack - Cordless Tool Battery Packs - Amazon.com

Looking at them I did a current sense in the middle and no surprise this pack gave me 17.84 Amps out when weed whacking a tree like weed branch.
This is also done in pairs, it's a total of 12 cells, 6 parallel in and 6 set of those in parallel 6P6S pack.
Rated total is 57Wh, I think my cells are older from 2010 and are 1200mAh, so paired together for a capacity of 2400mAh each set.
Ryobi did the cutoff correctly with high drain amperage. They cut off the batteries at 3V per cell, but under load that equaled more to 2.6/2.55V per cell. The bottom rated species for the cells. The cutoff cuts the pssotive, but the total lack battery voltage can still be sensed with the T temperature plug. If the pack falls below 2.5V per cell measured on this line then the charger won't charge it. They also only charge it to 4.1V and not 4.2V per cell.
I've noticed a lot more things are doing this to be on the safer side these days, not a bad thing considering what could happen with over charged packs.

Ryobi OP241A 24V 57.72 Wh Lithium Battery Pack - Cordless Tool Battery Packs - Amazon.com