Christmas test: Mavic Air 2 weight vs. consumption chart

[szakeetm]

I have been experimenting over a year with external batteries on my Mavic Air 2. As you all know, this is a delicate subject as extra power means extra weight, which increases power consumption and might even decrease propeller efficiency.

To do the test, I attached a bottle to my drone, and used a precision balance to measure its weight. I added 50, 100, 150... 300ml of water. The holding mechanism (originally for dropping objects) weighs an extra 80g.

Here is the setup on the ground and in the air:




I measured the drawn power by downloading the flight log. Power = battery_voltage * drawn_current

A typical log looks like this (I lift the bottle, then put it back by lowering altitude, then lift again, and take the average consumption why lifting):



Here is the chart:

• The stock drone is 570g, which is the first data point. Flight efficiency is the best
• At 120g weight, this data point is slightly outlier: the bottle was too light and it was wobbling in the drone's wind
• As weight is increased, power increases and also efficiency decreases: at 940g weight the drone consumes more than double as at 570g!
• Going above 800g caused the motors to be very hot (70C)

To test if my theory is correct, I tested with two external batteries:

• Stock drone: 30min30s hover time (DJI claims 34 mins flight time - LOG: )
• 165g, 2600mAh LiHV: 32 min hover time (LOG: )
• 250g, 4000mAh LiHV: 36 min hover time (LOG: )

Final remark: ribbon cables are not the best to connect an external battery, probably the contact resistance is too high. When soldering the external battery to the original DJI battery (know what you're doing, can be dangerous!), I got 34 minutes hover time with the 2600mAh battery, and only 32 minutes with the ribbon cable.

Conclusion:
You can increase flight time by external batteries, but not dramatically (max. +20%). You will compromise flight safety (unbalanced drone, cables touching propellers, etc.) and motors will be very hot. That said, I needed to try anyway, so why not post the results?

 

[blubyyou]

I have been experimenting over a year with external batteries on my Mavic Air 2. As you all know, this is a delicate subject as extra power means extra weight, which increases power consumption and might even decrease propeller efficiency.

To do the test, I attached a bottle to my drone, and used a precision balance to measure its weight. I added 50, 100, 150... 300ml of water. The holding mechanism (originally for dropping objects) weighs an extra 80g.

Here is the setup on the ground and in the air:

View attachment 140725
View attachment 140724

I measured the drawn power by downloading the flight log. Power = battery_voltage * drawn_current

A typical log looks like this (I lift the bottle, then put it back by lowering altitude, then lift again, and take the average consumption why lifting):

View attachment 140728

Here is the chart:

View attachment 140726

• The stock drone is 570g, which is the first data point. Flight efficiency is the best
• At 120g weight, this data point is slightly outlier: the bottle was too light and it was wobbling in the drone's wind
• As weight is increased, power increases and also efficiency decreases: at 940g weight the drone consumes more than double as at 570g!
• Going above 800g caused the motors to be very hot (70C)

To test if my theory is correct, I tested with two external batteries:

View attachment 140729View attachment 140730

• Stock drone: 30min30s hover time (DJI claims 34 mins flight time - LOG: )
• 165g, 2600mAh LiHV: 32 min hover time (LOG: )
• 250g, 4000mAh LiHV: 36 min hover time (LOG: )

Final remark: ribbon cables are not the best to connect an external battery, probably the contact resistance is too high. When soldering the external battery to the original DJI battery (know what you're doing, can be dangerous!), I got 34 minutes hover time with the 2600mAh battery, and only 32 minutes with the ribbon cable.

Conclusion:
You can increase flight time by external batteries, but not dramatically (max. +20%). You will compromise flight safety (unbalanced drone, cables touching propellers, etc.) and motors will be very hot. That said, I needed to try anyway, so why not post the results?

Thank you for that info. I will NOT be using ANY external batteries.

 

[lee82gx]

But you started with only 92% with the smaller external battery….? So there could be a bit more juice in there no?

Another thing to note when using external 3s packs is that they are now in parallel to the original 3s battery inside, and it is never a great idea to connect in this way. This is because when either of the packs drop in voltage (due to very high load), the drop is not balanced between the 2 packs. Especially internal resistance (add cell IR to ribbon cable Internal resistance and this becomes much much worse as you noticed) will be different too. In such a case current drain will be higher in one of the packs. This leads to one of the packs with higher remaining voltage charging the other cell with its potential difference (maybe small, say less than 1V but with across high IR due to heat build up in the cells and connectors) this is a high wastage and hence you don’t get that much into the motors anyhow.

Finally, when you do an external 11V parallel, the flight computer is still reading the original cell voltage as remaining capacity, which may be inaccurate but not preventable in the case of emergency no battery landing- ie you may have unused charge in the external pack. The way to check this is after landing you can apply a big load say 10 ohms and 20W into the internal pack and read the voltage, compare to the external pack with the ribbon / solder with same load and read the voltage.

This is why the best way to have a bigger pack is always to have a proper 3s2p pack, ie each individual cell is paralleled with another individual 4.35V pack at its own terminals (Cathode n Anode) to minimise additional IR. Not only that, your individual cell reading in V will be accurate and you can then fly by reading the voltage. It will not auto land so soon as well. But the danger is this is always a one way modification and you need a way to reset any permanent failure flags in the battery management chip during disconnect and add a cell, which can cause voltage imbalance and trip the chip into permanent failure flag. It’s already possible by using a chip to encode and reset.

 

[szakeetm]

But you started with only 92% with the smaller external battery….? So there could be a bit more juice in there no?

For some reason recording only started at 92%. The recorded flight time is only 30 min, but the total time was 32min, see title.

Another thing to note when using external 3s packs is that they are now in parallel to the original 3s battery inside, and it is never a great idea to connect in this way. This is because when either of the packs drop in voltage (due to very high load), the drop is not balanced between the 2 packs. Especially internal resistance (add cell IR to ribbon cable Internal resistance and this becomes much much worse as you noticed) will be different too. In such a case current drain will be higher in one of the packs. This leads to one of the packs with higher remaining voltage charging the other cell with its potential difference (maybe small, say less than 1V but with across high IR due to heat build up in the cells and connectors) this is a high wastage and hence you don’t get that much into the motors anyhow.

Do you mind if I contradict? I have an electric scooter where I paralleled two large (8Ah and 12Ah) batteries and I have voltage and ammeters on both. What happens is indeed one dropping voltage faster (due to different wire and internal resistances), but the other doesn't charge the smaller, it simply sends more current to the drone (or scooter).

Concrete example during load:
- Scooter controller at 36V
- "Closer" internal battery at 37V
- "Farther" external battery at 38V

This won't result in the external battery charging the internal, it will send more current from the external to the controller. The math is so accurate that on the long term, the ratio of the currents is the the ratio of the battery capacities.

Of course, when I stop the scooter or land the drone (no load), then indeed the external battery starts charging the internal, but this is at the end of the flight.

Finally, when you do an external 11V parallel, the flight computer is still reading the original cell voltage as remaining capacity, which may be inaccurate but not preventable in the case of emergency no battery landing- ie you may have unused charge in the external pack. The way to check this is after landing you can apply a big load say 10 ohms and 20W into the internal pack and read the voltage, compare to the external pack with the ribbon / solder with same load and read the voltage.

Yes, this is a good point. What I did was simply disconnecting the external battery, which "restored" itself to around 11V after a few minutes, whereas the internal only to 10.4V, confirming what you hint: the external battery still had some charge. (Your under load test would me more accurate)

This is why the best way to have a bigger pack is always to have a proper 3s2p pack, ie each individual cell is paralleled with another individual 4.35V pack at its own terminals (Cathode n Anode) to minimise additional IR.

In the photos indeed it's just the battery leads being paralleled. However I did tests earlier with the balance leads also paralleled, creating a full 3s2p pack (attached), the difference wasn't dramatic.

Not only that, your individual cell reading in V will be accurate and you can then fly by reading the voltage. It will not auto land so soon as well. But the danger is this is always a one way modification and you need a way to reset any permanent failure flags in the battery management chip during disconnect and add a cell, which can cause voltage imbalance and trip the chip into permanent failure flag. It’s already possible by using a chip to encode and reset.

So far the BMS survived connecting external batteries, I could even charge with the original charger, so that the internal BMS was balancing even the external cells. Of course the state of charge display was off, as it counts the mAh passed but not the voltage.

Thanks for your comments! I wish they found a way to unlock factory BMSes, so we could create true 3s2p packs and "teach" the BMS to the new capacity.

 

[lee82gx]

For some reason recording only started at 92%. The recorded flight time is only 30 min, but the total time was 32min, see title.



Do you mind if I contradict? I have an electric scooter where I paralleled two large (8Ah and 12Ah) batteries and I have voltage and ammeters on both. What happens is indeed one dropping voltage faster (due to different wire and internal resistances), but the other doesn't charge the smaller, it simply sends more current to the drone (or scooter).

Concrete example during load:
- Scooter controller at 36V
- "Closer" internal battery at 37V
- "Farther" external battery at 38V

This won't result in the external battery charging the internal, it will send more current from the external to the controller. The math is so accurate that on the long term, the ratio of the currents is the the ratio of the battery capacities.

Of course, when I stop the scooter or land the drone (no load), then indeed the external battery starts charging the internal, but this is at the end of the flight.



Yes, this is a good point. What I did was simply disconnecting the external battery, which "restored" itself to around 11V after a few minutes, whereas the internal only to 10.4V, confirming what you hint: the external battery still had some charge. (Your under load test would me more accurate)



In the photos indeed it's just the battery leads being paralleled. However I did tests earlier with the balance leads also paralleled, creating a full 3s2p pack (attached), the difference wasn't dramatic.




So far the BMS survived connecting external batteries, I could even charge with the original charger, so that the internal BMS was balancing even the external cells. Of course the state of charge display was off, as it counts the mAh passed but not the voltage.

Thanks for your comments! I wish they found a way to unlock factory BMSes, so we could create true 3s2p packs and "teach" the BMS to the new capacity.

You are right practically at extremely high loads the higher voltage battery would probably NOT charge the lower voltage battery...I suppose I was exaggerating. You cannot totally ignore Ohm's law : Higher potential difference will always flow current, even if one is 11.2 and one more is 11.1, the actual potential difference means 0.1V, and that will be exposed to whatever IR both cells have and current will flow across them, converting to heat.

Balance leads paralleled - I'm not convinced they are low enough in resistance compared to spot weld.. but thats just a hunch haha. I guess even if you do it with almost no resistance the chemical / internal resistance is still too substantially different between 2 different type of cells. It would have to be chemically similar AND geometrically similar to achieve equal resistance, something impossible to do with another hobby pack battery. BUT easy enough to do with an actual MA2 battery pack disassembled...! Everytime you have imbalance in IR you will lose charge to heat.

if your BMS is charging and balancing well it is either you were very careful during the attachment and balanced the cells prior to attachment (bravo) or very lucky (haha).

There is not much different in the electric scooter example, so I am glad you have prior experience.

To unlock the BMS :
the cheapest way is to use this freeware:

Dji battery killer last version updated · Issue #245 · o-gs/dji-firmware-tools

Hi Mr Mixeysan! Can you get link of Dji battery killer updated software to me? Thanks you!

github.com

you will need USB to SDA SCL / I2C converter which is very cheap and you can at least reset permanent failure flags.
If you will, you can directly probably directly use very very high capacity cells and just disguise it to the BMS as the original 3530mAh cells, perhaps the voltage curve still can be used.

Otherwise you need the actual or reverse engineered TI Battery management adapter and a license (either paid or otherwise) to fully unlock and rewrite everything on the BMS.

This is how hackers in my country do it, we buy original cells, link them in parallel, and use the software to rewrite the capacity.

 

[szakeetm]

Many thanks for the valuable, detailed input.

Just a small comment:

if your BMS is charging and balancing well it is either you were very careful during the attachment and balanced the cells prior to attachment (bravo) or very lucky (haha).

I just fully charged both batteries - the DJI with the stock charger and the external with a typical LiPo balance charger that does top balancing. After rest, all six cells stabilized at ~4.3V, so I deemed it safe to connect.

I agree with the rest - heat, internal resistance and thin balance wires make my solution not really ideal.
As you say, doubling capacity using original cells would be the best, I'll have a look if those cells can be bought on Aliexpress. Meanwhile I sourced myself with a spot welder so I'm up to experiment again, until I get it right or burn down the house

 

[lee82gx]

Many thanks for the valuable, detailed input.

Just a small comment:

I just fully charged both batteries - the DJI with the stock charger and the external with a typical LiPo balance charger that does top balancing. After rest, all six cells stabilized at ~4.3V, so I deemed it safe to connect.

I agree with the rest - heat, internal resistance and thin balance wires make my solution not really ideal.
As you say, doubling capacity using original cells would be the best, I'll have a look if those cells can be bought on Aliexpress. Meanwhile I sourced myself with a spot welder so I'm up to experiment again, until I get it right or burn down the house

you know what you are doing so I'm very glad to meet you!

In my country I just have to send my pack and USD80 for a rework / modification to double the capacity with full capacity rewrite. Its so convenient that I neglect to do any more handiwork.

38.99US $ |Ev2400 Bqstudio Debugger 2300 Drone Battery Repair Maintenance Smb Communication Box T16 T20 Ev2400pro Ev2400mini - Instrument Parts & Accessories - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

(this is a cloned TI debugger). I think you can request for the software from the seller. If he can provide also ask for the full unlocking authority. Let me know if you do end up buying this.

14.71US $ 15% OFF|3pcs/lot New 3.8v 3500mah Battery Cell For Dji Mavic Air 2 2s Air2s Drone Diy Intelligent Flight Battery - Digital Batteries - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

This is example of raw cell. I am surprised that some third party are now supplying 4000mAh (!) and genuinely I don't think they can last. I wouldn't dare to use 3rd party chemical on my 500g flying piece of rock over the sea or people.

Lastly, here is the example with DJI Battery killer (The author of this tool is Ukrainian, and I am praying his is OKAY).
CP2112 used to be cheap as chips and now its expensive as chips....

 

[szakeetm]

you know what you are doing so I'm very glad to meet you!

In my country I just have to send my pack and USD80 for a rework / modification to double the capacity with full capacity rewrite. Its so convenient that I neglect to do any more handiwork.

38.99US $ |Ev2400 Bqstudio Debugger 2300 Drone Battery Repair Maintenance Smb Communication Box T16 T20 Ev2400pro Ev2400mini - Instrument Parts & Accessories - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

(this is a cloned TI debugger). I think you can request for the software from the seller. If he can provide also ask for the full unlocking authority. Let me know if you do end up buying this.

14.71US $ 15% OFF|3pcs/lot New 3.8v 3500mah Battery Cell For Dji Mavic Air 2 2s Air2s Drone Diy Intelligent Flight Battery - Digital Batteries - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

This is example of raw cell. I am surprised that some third party are now supplying 4000mAh (!) and genuinely I don't think they can last. I wouldn't dare to use 3rd party chemical on my 500g flying piece of rock over the sea or people.

Lastly, here is the example with DJI Battery killer (The author of this tool is Ukrainian, and I am praying his is OKAY).
CP2112 used to be cheap as chips and now its expensive as chips....

Many thanks again.

The author of this video claims that the BMS auto-learns the higher battery capacity after while. To be tested, I'll report back.

 

[lee82gx]

Many thanks again.

The author of this video claims that the BMS auto-learns the higher battery capacity after while. To be tested, I'll report back.

i am skeptical (simply because DJI implements too many locks on its batteries for good or bad), but always open to new things...let us know!

 

[szakeetm]

Cells arrived from China, up for some balancing, soldering and testing, will teport back!

 

[szakeetm]

Cells arrived from China, up for some balancing, soldering and testing, will teport back!

Assembled a DIY pack, will make it nicer if works.
Not for the faint hearted - my small spot welder doesn't weld to the steel cathode, so I needed to solder, and there are 100 different ways to accidentally short out something while doing this. I have created a spark when a balance cable accidentally touched the wrong area, but to my surprise and luck the BMS didn't lock itself.

The final design weighs 374g (original: 200g), and I could get 77Wh out (original: 40Wh, but this had already 50 cycles). Note that I shifted the extra cells forward to match the drone's center of gravity.

The drone is at DJI for an unrelated repair, will test in a week or so.

 

[szakeetm]

Maiden flight: hovered indoors for 41:37. (Stock battery: around 30 minutes). Will fly outdoors tomorrow. It seems that there is a certain self-learning of the BMS, the indicated capacity is increasing by 100mAh after each charge cycle. I will report more details later!

 

[szakeetm]

Maiden flight: hovered indoors for 41:37. (Stock battery: around 30 minutes). Will fly outdoors tomorrow. It seems that there is a certain self-learning of the BMS, the indicated capacity is increasing by 100mAh after each charge cycle. I will report more details later!

I posted the conclusion here: Mavic Air 2 44:20 min / 19.5km battery mod