In-depth review of Mavic Pro extended capacity battery (Hobbitec 6830mAh) - Performance and raised questions

[crada]

Hi
I'm new here, and the forum has been of great help in my current project, so I wanted to share my tests on the after-market battery available from Chinese suppliers like this one on Aliexpress. Soon I'll post about my project and explain why I DO NEED some extra flight time in a relatively old drone.

The Hobbitec battery is marketed as 6830 mAh (78 Wh), and I'm happily surprised it is genuinely what it should be.

I first made an indoor hover test, waiting until the battery showed 0% before landing. The flight lasted 35 minutes 14 seconds, which is 11 min 44 sec more than what I get with a stock battery with 33 recharge cycles in it. That means a 49.9% increase in flight time.

I then retrieved the DAT files from the GO 4 app and extracted the data using CsvView. To analyze the used power, I extracted the fields containing voltage (BatteryInfo:ad_v: D), battery current (BatteryInfo:I: D as ave_I: D seems to have some kind of smoothing) and the current used by each motor (Motor:current:RFront, LFront, LBack, and RBack).

Using this data, the energy used by the motors was a 203% of the energy used in test with stock battery (from 34.9Wh to 70.7Wh), which shows that the battery indeed had twice the "juice", and the fact that the flight time increased only 50% is perfectly reasonable given the increase in weight. All battery mods I've seen have similar results.

Interestingly, when I look at the current reported by the battery, it is way too low. My original interpretation was that internally, this battery was a clone of the stock battery with a second set of cells connected in parallel, the same way many battery mods work, for example, using this power adaptor. This raised the concern that maybe the charging of the second set of cells was not balanced, which could lead to a shorter life of those cells. However, after opening the battery, it turned out that it consisted of a set of three pairs of Li-po cells. Pairs are connected in parallel and the three pairs are in series. So it is somehow intentionally reporting less current, perhaps for some sort of compatibility issue or to make battery life estimations work properly. Anyway, it seems like the actual current is exactly twice the current reported by the battery, which makes perfect sense when looking at the figures I just mentioned of the current used by the motors (which are measured by the Mavic's ESC I believe, NOT by circuitry inside the battery).

In this case, the total energy delivered by the battery was 78.2 Wh, which is a tiny little more than the specs. Very nice!
Weight-wise, according to my kitchen scale, the stock battery weighs 237 g (240g according to specs), and the extended battery weighs 405 g, so again a tiny bit better than the specs that say 406.8 g. Then, the extra battery weight is only 168 g, an increase of 70.9% over the stock battery weight.

My stock battery with 33 recharge cycles on an identical hover test delivered 38.7 Wh, significantly less than the specs of 43.6 Wh. So, the extended Hobbitec battery delivered 202% of the energy of my used battery and 179% of the energy of an up-to-specs stock battery (with 43.6 Wh).

The extra weight made the efficiency (in terms of minutes of hover per Wh) drop from 0.607 min/Wh to 0.450 min/Wh.
Using these figures, the theoric hover time of a stock battery with a full 43.6 Wh would be 26 min 28 sec (flight time according to specs is 27 minutes with no wind at a consistent 25 km/h). Therefore, the additional flight time over a full-capacity stock battery would be 8 min 45 sec, which means a 33% increase.
In terms of energy, the additional energy over a theoric full capacity stock battery (43.6 Wh) is 34.6 Wh, which means a 79.3% increase, a very similar figure to the increase in weight.

I also compared the reported battery percentages with the actual battery left (which I was able to calculate after the flight because I landed at 0%), and it is as accurate as the stock battery, which only shows deviations smaller than 2%.

Summarizing:
The battery is good and has the advertised capacity and weight.
It indeed delivers significant extra flight time both in hover tests and real flight tests.

Now, the fact that it reports half the delivered current raises some concerns. I've made more flights after the initial hover test, and a couple of important observations:
1.- I don't get the error message of battery over current that I sometimes got with the stock battery. This means the battery no longer limits the power the drone uses.
2.- I've observed total motor currents as high as 31 Amps (at full throttle upwards), and ESC temperature as high as 80°C (176 F), values I never saw with the stock battery.

Therefore, my concerns are that I could burn the ESC or the motors by using full throttle with this heavier battery that can deliver so much power. However, given the ESC measures current and temperature, DJI engineers must have been smart enough to make it self-limit its power to avoid overheating. Right?
I have not found the Mavic ESC specs regarding maximum current and temperature. For ESCs in general, I've seen people recommend below 70°C and mention 80°C as the limit, but some ESCs are happy up to 125°C. For the maximum current, I only found some info in this forum, where I read that it is set by firmware to 36 Amps (link to the post), so that should be the safe limit... I guess.

So here are my questions for this community:

1.- Does the Mavic Pro has software failsafe controls to prevent ESC overheating and limit motors currents?
2.- Have anyone seen warning messages related to ESC overheating/overcurrent?
3.- As you have a lot of experience with crash forensics: are high currents and/or high ESC temperatures a common factor in drone malfunction?

Thanks!

 

[lolo780]

Hi
I'm new here, and the forum has been of great help in my current project, so I wanted to share my tests on the after-market battery available from Chinese suppliers like this one on Aliexpress. Soon I'll post about my project and explain why I DO NEED some extra flight time in a relatively old drone.

The Hobbitec battery is marketed as 6830 mAh (78 Wh), and I'm happily surprised it is genuinely what it should be.

I first made an indoor hover test, waiting until the battery showed 0% before landing. The flight lasted 35 minutes 14 seconds, which is 11 min 44 sec more than what I get with a stock battery with 33 recharge cycles in it. That means a 49.9% increase in flight time.

I then retrieved the DAT files from the GO 4 app and extracted the data using CsvView. To analyze the used power, I extracted the fields containing voltage (BatteryInfo:ad_v: D), battery current (BatteryInfo:I: D as ave_I: D seems to have some kind of smoothing) and the current used by each motor (Motor:current:RFront, LFront, LBack, and RBack).

Using this data, the energy used by the motors was a 203% of the energy used in test with stock battery (from 34.9Wh to 70.7Wh), which shows that the battery indeed had twice the "juice", and the fact that the flight time increased only 50% is perfectly reasonable given the increase in weight. All battery mods I've seen have similar results.

Interestingly, when I look at the current reported by the battery, it is way too low. My original interpretation was that internally, this battery was a clone of the stock battery with a second set of cells connected in parallel, the same way many battery mods work, for example, using this power adaptor. This raised the concern that maybe the charging of the second set of cells was not balanced, which could lead to a shorter life of those cells. However, after opening the battery, it turned out that it consisted of a set of three pairs of Li-po cells. Pairs are connected in parallel and the three pairs are in series. So it is somehow intentionally reporting less current, perhaps for some sort of compatibility issue or to make battery life estimations work properly. Anyway, it seems like the actual current is exactly twice the current reported by the battery, which makes perfect sense when looking at the figures I just mentioned of the current used by the motors (which are measured by the Mavic's ESC I believe, NOT by circuitry inside the battery).

In this case, the total energy delivered by the battery was 78.2 Wh, which is a tiny little more than the specs. Very nice!
Weight-wise, according to my kitchen scale, the stock battery weighs 237 g (240g according to specs), and the extended battery weighs 405 g, so again a tiny bit better than the specs that say 406.8 g. Then, the extra battery weight is only 168 g, an increase of 70.9% over the stock battery weight.

My stock battery with 33 recharge cycles on an identical hover test delivered 38.7 Wh, significantly less than the specs of 43.6 Wh. So, the extended Hobbitec battery delivered 202% of the energy of my used battery and 179% of the energy of an up-to-specs stock battery (with 43.6 Wh).

The extra weight made the efficiency (in terms of minutes of hover per Wh) drop from 0.607 min/Wh to 0.450 min/Wh.
Using these figures, the theoric hover time of a stock battery with a full 43.6 Wh would be 26 min 28 sec (flight time according to specs is 27 minutes with no wind at a consistent 25 km/h). Therefore, the additional flight time over a full-capacity stock battery would be 8 min 45 sec, which means a 33% increase.
In terms of energy, the additional energy over a theoric full capacity stock battery (43.6 Wh) is 34.6 Wh, which means a 79.3% increase, a very similar figure to the increase in weight.

I also compared the reported battery percentages with the actual battery left (which I was able to calculate after the flight because I landed at 0%), and it is as accurate as the stock battery, which only shows deviations smaller than 2%.

Summarizing:
The battery is good and has the advertised capacity and weight.
It indeed delivers significant extra flight time both in hover tests and real flight tests.

Now, the fact that it reports half the delivered current raises some concerns. I've made more flights after the initial hover test, and a couple of important observations:
1.- I don't get the error message of battery over current that I sometimes got with the stock battery. This means the battery no longer limits the power the drone uses.
2.- I've observed total motor currents as high as 31 Amps (at full throttle upwards), and ESC temperature as high as 80°C (176 F), values I never saw with the stock battery.

Therefore, my concerns are that I could burn the ESC or the motors by using full throttle with this heavier battery that can deliver so much power. However, given the ESC measures current and temperature, DJI engineers must have been smart enough to make it self-limit its power to avoid overheating. Right?
I have not found the Mavic ESC specs regarding maximum current and temperature. For ESCs in general, I've seen people recommend below 70°C and mention 80°C as the limit, but some ESCs are happy up to 125°C. For the maximum current, I only found some info in this forum, where I read that it is set by firmware to 36 Amps (link to the post), so that should be the safe limit... I guess.

So here are my questions for this community:

1.- Does the Mavic Pro has software failsafe controls to prevent ESC overheating and limit motors currents?
2.- Have anyone seen warning messages related to ESC overheating/overcurrent?
3.- As you have a lot of experience with crash forensics: are high currents and/or high ESC temperatures a common factor in drone malfunction?

Thanks!

The motors won't burn out but you can melt the plastic under the cap screw heads and the motor will pull off the arm.

MOSFETs can blow on the ESC board and I'd recommend heat sinking them. For the motors, space them up off the arms to promote airflow, and use insulating washers.

The next problem is the motor wires, you'll start to melt the insulation, mostly on the rears. Upgraded wires can be carefully squeezed through the factory hinges. This is a Mavic 2 I did, but the idea is the same.

 

[crada]

The motors won't burn out but you can melt the plastic under the cap screw heads and the motor will pull off the arm.

MOSFETs can blow on the ESC board and I'd recommend heat sinking them. For the motors, space them up off the arms to promote airflow, and use insulating washers.

The next problem is the motor wires, you'll start to melt the insulation, mostly on the rears. Upgraded wires can be carefully squeezed through the factory hinges. This is a Mavic 2 I did, but the idea is the same.


View attachment 162439

This is very interesting. Thanks a lot.
Thankfully, for the project, I'll use this battery. I'll be flying in fairly low temperatures; that should help.
Do you have any idea of how hot that ESC is getting? And how many amps are you drawing?
Do you get warning or error messages about high temperatures/amperage?
Cheers

 

[Slawaw]

This is very interesting. Thanks a lot.
Thankfully, for the project, I'll use this battery. I'll be flying in fairly low temperatures; that should help.
Do you have any idea of how hot that ESC is getting? And how many amps are you drawing?
Do you get warning or error messages about high temperatures/amperage?
Cheers

Don't create problems for yourself where there aren't any. It is impossible to overheat the Mavic Pro. My "old man" has 1600 km of flight and everything is on double batteries. The plastic does not melt, the engines are original. But there is an important point! The drone should move, not hover. Then the cooling is more efficient. I have never seen any reports of overheating.

 

[lolo780]

 

[crada]

Don't create problems for yourself where there aren't any. It is impossible to overheat the Mavic Pro. My "old man" has 1600 km of flight and everything is on double batteries. The plastic does not melt, the engines are original. But there is an important point! The drone should move, not hover. Then the cooling is more efficient. I have never seen any reports of overheating.

Thanks for sharing your experience. This is very reassuring. I'll take the advice to avoid hovering.
Thanks.

 

[crada]

View attachment 164713View attachment 164714

Did you get any overheating warnings when that happened?

 

[lolo780]

Did you get any overheating warnings when that happened?

None. The Mavic 3 uses thermal pads to transfer heat to the bottom plate. I used tiny individual heat sinks on the M1.
Motor base heat solved by elevating the motor off the arm.

 

[Conservative Nihilist]

The motors won't burn out but you can melt the plastic under the cap screw heads and the motor will pull off the arm.

MOSFETs can blow on the ESC board and I'd recommend heat sinking them. For the motors, space them up off the arms to promote airflow, and use insulating washers.

The next problem is the motor wires, you'll start to melt the insulation, mostly on the rears. Upgraded wires can be carefully squeezed through the factory hinges. This is a Mavic 2 I did, but the idea is the same.


View attachment 162439

Holy cow lolo 780 you are a mad scientist and a freaking electronics genius rolled into one. I would never dare attempt the kind of extreme surgery that you have successfully carried out on your Mavic 1 Pro, despite my BSEE earned in Cali. I tip my hat with humility to you sir. You are a boss of bosses. A don of dons, I am feverishly taking notes here.

 

[Conservative Nihilist]

Hi
I'm new here, and the forum has been of great help in my current project, so I wanted to share my tests on the after-market battery available from Chinese suppliers like this one on Aliexpress. Soon I'll post about my project and explain why I DO NEED some extra flight time in a relatively old drone.

The Hobbitec battery is marketed as 6830 mAh (78 Wh), and I'm happily surprised it is genuinely what it should be.

I first made an indoor hover test, waiting until the battery showed 0% before landing. The flight lasted 35 minutes 14 seconds, which is 11 min 44 sec more than what I get with a stock battery with 33 recharge cycles in it. That means a 49.9% increase in flight time.

I then retrieved the DAT files from the GO 4 app and extracted the data using CsvView. To analyze the used power, I extracted the fields containing voltage (BatteryInfo:ad_v: D), battery current (BatteryInfo:I: D as ave_I: D seems to have some kind of smoothing) and the current used by each motor (Motor:current:RFront, LFront, LBack, and RBack).

Using this data, the energy used by the motors was a 203% of the energy used in test with stock battery (from 34.9Wh to 70.7Wh), which shows that the battery indeed had twice the "juice", and the fact that the flight time increased only 50% is perfectly reasonable given the increase in weight. All battery mods I've seen have similar results.

Interestingly, when I look at the current reported by the battery, it is way too low. My original interpretation was that internally, this battery was a clone of the stock battery with a second set of cells connected in parallel, the same way many battery mods work, for example, using this power adaptor. This raised the concern that maybe the charging of the second set of cells was not balanced, which could lead to a shorter life of those cells. However, after opening the battery, it turned out that it consisted of a set of three pairs of Li-po cells. Pairs are connected in parallel and the three pairs are in series. So it is somehow intentionally reporting less current, perhaps for some sort of compatibility issue or to make battery life estimations work properly. Anyway, it seems like the actual current is exactly twice the current reported by the battery, which makes perfect sense when looking at the figures I just mentioned of the current used by the motors (which are measured by the Mavic's ESC I believe, NOT by circuitry inside the battery).

In this case, the total energy delivered by the battery was 78.2 Wh, which is a tiny little more than the specs. Very nice!
Weight-wise, according to my kitchen scale, the stock battery weighs 237 g (240g according to specs), and the extended battery weighs 405 g, so again a tiny bit better than the specs that say 406.8 g. Then, the extra battery weight is only 168 g, an increase of 70.9% over the stock battery weight.

My stock battery with 33 recharge cycles on an identical hover test delivered 38.7 Wh, significantly less than the specs of 43.6 Wh. So, the extended Hobbitec battery delivered 202% of the energy of my used battery and 179% of the energy of an up-to-specs stock battery (with 43.6 Wh).

The extra weight made the efficiency (in terms of minutes of hover per Wh) drop from 0.607 min/Wh to 0.450 min/Wh.
Using these figures, the theoric hover time of a stock battery with a full 43.6 Wh would be 26 min 28 sec (flight time according to specs is 27 minutes with no wind at a consistent 25 km/h). Therefore, the additional flight time over a full-capacity stock battery would be 8 min 45 sec, which means a 33% increase.
In terms of energy, the additional energy over a theoric full capacity stock battery (43.6 Wh) is 34.6 Wh, which means a 79.3% increase, a very similar figure to the increase in weight.

I also compared the reported battery percentages with the actual battery left (which I was able to calculate after the flight because I landed at 0%), and it is as accurate as the stock battery, which only shows deviations smaller than 2%.

Summarizing:
The battery is good and has the advertised capacity and weight.
It indeed delivers significant extra flight time both in hover tests and real flight tests.

Now, the fact that it reports half the delivered current raises some concerns. I've made more flights after the initial hover test, and a couple of important observations:
1.- I don't get the error message of battery over current that I sometimes got with the stock battery. This means the battery no longer limits the power the drone uses.
2.- I've observed total motor currents as high as 31 Amps (at full throttle upwards), and ESC temperature as high as 80°C (176 F), values I never saw with the stock battery.

Therefore, my concerns are that I could burn the ESC or the motors by using full throttle with this heavier battery that can deliver so much power. However, given the ESC measures current and temperature, DJI engineers must have been smart enough to make it self-limit its power to avoid overheating. Right?
I have not found the Mavic ESC specs regarding maximum current and temperature. For ESCs in general, I've seen people recommend below 70°C and mention 80°C as the limit, but some ESCs are happy up to 125°C. For the maximum current, I only found some info in this forum, where I read that it is set by firmware to 36 Amps (link to the post), so that should be the safe limit... I guess.

So here are my questions for this community:

1.- Does the Mavic Pro has software failsafe controls to prevent ESC overheating and limit motors currents?
2.- Have anyone seen warning messages related to ESC overheating/overcurrent?
3.- As you have a lot of experience with crash forensics: are high currents and/or high ESC temperatures a common factor in drone malfunction?

Thanks!

This after-market battery you've kindly alerted the drone world about is a game-changing breakthrough that breathes new life into the venerable Mavic 1 Pro. I could not believe my eyes when I noted the weight as being 220 grams, which is actually 20 grams LOWER than the weight of the stock Mavic 1 Pro battery.

Gone is the need to jury-rig piggy-backed batteries and worry about low battery warnings even while there is plenty of battery power left. This battery is PRECISELY the invention I have been waiting for over a period of YEARS. Wow! Now my Litchi waypoint mission duration will extend well beyond the current record of 7 miles round trip to well over TEN miles. Holy Toledo! What a time to be alive!

One final question though. Have you experienced any overheating issues as yet with this battery during forward flight with no hover time? If your reply is that heating is not a concern, I intend to order a pair of these batteries right away. Thanks in advance for any clarification about this heating question that can be offered.

 

[lolo780]

This after-market battery you've kindly alerted the drone world about is a game-changing breakthrough that breathes new life into the venerable Mavic 1 Pro. I could not believe my eyes when I noted the weight as being 220 grams, which is actually 20 grams LOWER than the weight of the stock Mavic 1 Pro battery.

Gone is the need to jury-rig piggy-backed batteries and worry about low battery warnings even while there is plenty of battery power left. This battery is PRECISELY the invention I have been waiting for over a period of YEARS. Wow! Now my Litchi waypoint mission duration will extend well beyond the current record of 7 miles round trip to well over TEN miles. Holy Toledo! What a time to be alive!

One final question though. Have you experienced any overheating issues as yet with this battery during forward flight with no hover time? If your reply is that heating is not a concern, I intend to order a pair of these batteries right away. Thanks in advance for

The extended battery weight is 405g as per the OP. The Mavic 1 will fly nicely around 900g including vertical flights at 10m/s. I've never tried the battery in question however.

any clarification about this heating question that can be offered.

 

[Conservative Nihilist]

The extended battery weight is 405g as per the OP. The Mavic 1 will fly nicely around 900g including vertical flights at 10m/s. I've never tried the battery in question however.

This correction about battery weight is appreciated sir, as is the assurance that overheating won't be a significant concern with the extra capacity battery. Over and out.

 

[crada]

One final question though. Have you experienced any overheating issues as yet with this battery during forward flight with no hover time?

No, I have not got any overheating problems. My flights have been flying vertically up at 10 m/s. So I would say that flying forward, your Mavic Pro will stay cooler than mine. Mine went pretty hot, but that has never turned into a problem.

 

[Conservative Nihilist]

No, I have not got any overheating problems. My flights have been flying vertically up at 10 m/s. So I would say that flying forward, your Mavic Pro will stay cooler than mine. Mine went pretty hot, but that has never turned into a problem.

I've bought three of these batteries from Ali Express, such is my level of excitement about the extended range they promise, so this assurance that overheating is not a concern is icing on the cake. I thank you kindly sir for taking a moment to respond to my inquiry on this subject.

 

[Slawaw]

There was such information about high-capacity batteries. From a real user. Weight 406 grams, flies 37 minutes to 10%. The battery itself does not heat up, the battery contacts are very hot after the flight.

 

[Conservative Nihilist]

Very hot battery contacts are cause for concern. I'll be ready soon to run my own flight tests, once that slow boat from China docks here. Slava Ukraine

 

[Slawaw]

Slava Ukraine

Героям слава.

 

[Conservative Nihilist]

Героям слава.

Після довгої ночі боротьби настає прекрасний світанок миру. Будь сильним, брате. Будь мужнім.

 

[Conservative Nihilist]

Ok so I just took delivery of three super batteries to use with my Mavic 1 Pro and Mavic Pro Platinum, and after a test within RC signal range around my house that lasted for 27 minutes, I decided to go for broke and send the Mavic 1 Pro away far beyond RC signal range for a round trip Litchi waypoint mission covering 11 miles total there and back, which was estimated by Litchi to take 30 minutes round trip.

The drone went well beyond its previous distance record but was not quite at its turnaround point before I was prompted to either permit or cancel RTH. Not wanting to push my luck, I allowed RTH to proceed and the drone arrived home with 45% of the battery charge remaining. My question is whether this drone could actually have completed the entire 11-mile round-trip if I had gone ahead and canceled RTH.

I am curious to hear from anyone who has purchased this extra-capacity battery, and who has also pushed the drone to the absolute distance limit possible with this larger-than-stock battery. I have read online that in the case of DIY extra-capacity batteries which were typically jury-rigged by Mavic 1 Pro owners before this stock-looking battery hit the market, the low battery warning would kick in when the stock battery ran low, EVEN IF the piggy-backed battery still had a lot of power left in it. I am curious to learn if the low batter warning I saw was premature for the actual amount of charge left, or was, in fact, accurate and not to be trifled with by overriding RTH. Any thoughts that other users of this giant battery can offer would be enormously appreciated.

Performance-wise, this oversized battery worked beautifully and barely got warm after the sort of long waypoint trip, that would have caused the stock battery that gets very warm. In short, anyone still on the fence about buying this high-capacity Mavic 1 Pro battery can rest assured that this is a worthwhile purchase that will dramatically increase the range of the Mavic 1. I give this battery an A grade in value for money.

 

[crada]

Ok so I just took delivery of three super batteries to use with my Mavic 1 Pro and Mavic Pro Platinum, and after a test within RC signal range around my house that lasted for 27 minutes, I decided to go for broke and send the Mavic 1 Pro away far beyond RC signal range for a round trip Litchi waypoint mission covering 11 miles total there and back, which was estimated by Litchi to take 30 minutes round trip.

The drone went well beyond its previous distance record but was not quite at its turnaround point before I was prompted to either permit or cancel RTH. Not wanting to push my luck, I allowed RTH to proceed and the drone arrived home with 45% of the battery charge remaining. My question is whether this drone could actually have completed the entire 11-mile round-trip if I had gone ahead and canceled RTH.

I am curious to hear from anyone who has purchased this extra-capacity battery, and who has also pushed the drone to the absolute distance limit possible with this larger-than-stock battery. I have read online that in the case of DIY extra-capacity batteries which were typically jury-rigged by Mavic 1 Pro owners before this stock-looking battery hit the market, the low battery warning would kick in when the stock battery ran low, EVEN IF the piggy-backed battery still had a lot of power left in it. I am curious to learn if the low batter warning I saw was premature for the actual amount of charge left, or was, in fact, accurate and not to be trifled with by overriding RTH. Any thoughts that other users of this giant battery can offer would be enormously appreciated.

Performance-wise, this oversized battery worked beautifully and barely got warm after the sort of long waypoint trip, that would have caused the stock battery that gets very warm. In short, anyone still on the fence about buying this high-capacity Mavic 1 Pro battery can rest assured that this is a worthwhile purchase that will dramatically increase the range of the Mavic 1. I give this battery an A grade in value for money.

I'm glad the battery worked for you as well.
I had a similar experience. But I was always repeating the same vertical flight. Therefore, I could check how much battery I had left on landing and progressively push the flight a bit longer.
But the short answer is yes; I had to cancel the RTH to take full advantage of the battery capacity.
The battery percentage calculation of the drone is correct; when it says 20%, it is indeed 20%. However, that 20% is 20% of a battery twice the size. And the drone doesn't know that.
For example, if the drone needs to travel 5 km to get back home, it internally calculates that it needs, let's say, 50% of the battery. So when it reaches that level, it initiates RTH. However, that 50% is double the energy that the drone thinks (due to the extra battery capacity); therefore, it only uses 25% for the actual RTH, and you end up landing with 25%.
I hope that makes sense. But summarizing, the drone calculates the RTH battery level assuming it has a normal battery, so you will always arrive back with a lot of juice left in the battery, especially if your RTH is initiated very far from the starting point. So it is safe to cancel RTH and keep flying, but you have to be very careful to do not to stretch it too much. Go in small incremental steps until you get back with a safe but not wasteful battery level (10%-20% ).