Diagnosing and resuscitating a set of DJI drone batteries

A couple of months ago, spurred into action by U.S. House of Representatives passage of legislation banning ongoing sales of DJI drones (and the still-unclear status of corresponding legislation in the Senate), along with aggressively-priced promotions on latest-generation products (likely related to the aforementioned legislative action), I decided to splurge on an upgrade to the Mavic Air that I mentioned back in October 2021 that I’d recently bought.

Truth be told, the Mavic Air was still holding its own feature set-wise, more than six years after its January 2018 introduction. It supports, for example, both front and rear collision avoidance and accompanying auto-navigation to dodge objects in its flight path (APAS, the Advanced Pilot Assistance System), along with a downward-directed camera to aid in takeoff and landing.

And its 3-axis gimbal-augmented front camera shoots video at up to 4K resolution at a 30-fps frame rate with a 100-Mbps bitrate.

That all said, its diminutive image sensor leads to notable image quality shortcomings in low ambient light settings, for example. Newer drones also extend collision avoidance capabilities to the sides, along with augmenting conventional cameras with LiDAR. And other recent government regulatory action, details of which I’ll save for a dedicated writeup another day, has compelled me to purchase additional hardware in order continue legally flying the Mavic Air in a variety of locations, along with needing to officially register it with the FAA per its >249 g weight.

Still, I plan for the Mavic Air to remain in my aviation stable for some time to come, alongside its newer sibling, details of which I’ll also save for a dedicated writeup (or few) another day (or few). After all, with greater weight frequently also comes greater wind resistance. And as regular readers already know, I’m loath to ditch perfectly good hardware. As such, after pressing “purchase” on my new toy, I decided to pull the Mavic Air out of storage, dust it off and fire it up again (I’d admittedly barely used it since acquiring it more than three years back). At the time I bought it, I’d also purchased three factory-refurbished spare batteries (a key addition for long operating sessions given its ~15 minute per-battery flight time) from a reputable eBay dealer.

The original battery installed in the drone still held a modicum of charge, I was happily surprised to see when I fired it up, and it also recharged back to “full” just fine. Here’s what it looks like:

Here’s a closeup of those markings:

See: still alive!

And here’s where it fits in the drone body cavity:

The three additional batteries, on the other hand (knowledgeable readers are already shaking their heads in shared sorrow at this point)…dead as a doornail (or, if you prefer, a parrot), all three of ‘em, with no LED illumination whatsoever in response to front button presses. And, for what it’s worth, I’m not alone in my woe.

My subsequent research (PDF) informed me that per Battery Storage list entry #3, “the battery will enter hibernation mode if depleted and stored for a long period…leave the battery unattended for 5 minutes, and then…recharge the battery to bring it out of hibernation.” Unfortunately, that didn’t work for me, even after an hour’s worth of charger tethering. Which led to Battery Storage list entry #2: “DO NOT store the battery for an extended period after fully discharging it. Doing so may over-discharge the battery and cause irreparable battery cell damage.” Readers of my recent post on SLA and lithium-based batteries may be feeling more than a bit of déjà vu at this point.

What I initially figured had happened was something analogous to a situation I’d diagnosed more than six years ago, involving a wireless charging-capable battery case for my wife’s iPhone that wasn’t able to recharge itself over Qi because it its integrated battery cells had drained beyond the point where they could even power the charging circuitry itself. Reality, it turns out, was even more complicated than that due to the DJI battery pack’s multi-cell structure and the consequent need for an integrated battery management system (BMS) analogous to, albeit somewhat simpler than, the one discussed in a recent EDN writeup by another author.

How did I learn about this quirk? Reluctant to just toss the batteries and replace them with (pricey) new(er) ones, I came across someone online who was offering a resurrection service. My eventual savior, who I’ve agreed to only refer to by his first name Erik (and I’m referencing anonymously otherwise because, in his own words, “I don’t want to bring any attention to what I do because I don’t want to risk DJI unleashing their might upon me”), explained it this way:

The cells have likely been discharged below the threshold where the BMS sets a fault and keeps you from charging them. I don’t know how low they’ve drained until I open them up but, more often than not, they end up being just fine after a nice balanced charge and a few charge/discharge cycles before the BMS is reprogrammed. There’s no design fault in these batteries. The BMS is pretty standard and the cells are actually very high quality.

Upfront, I paid $34.99 per battery (minus a modest quantity discount; also note that he offers a full per-battery fee refund for any batteries that he can’t fix), plus $9 for quantity-independent return shipping, along with $22.80 to ship the batteries to him in the first place. They arrived on a Friday, and he’d successfully resuscitated, tested and shipped them back to me by the next Monday. Here’s how he described the background and the process in his own words (with only light-touch editing by yours truly…note that he in-advance agreed that I could republish them):

They all have been successfully repaired, bench tested and test-flown.

Getting a bit of the battery’s history and the problem doesn’t hurt, but it’s unnecessary. The BMS stores all the information I need. It’s like pulling codes from your car with an OBD2 computer, and like repairing a car, it’s up to the technician to interpret them.

When a battery arrives, I physically inspect it and make sure I didn’t get a battery that works fine. It’s happened before.

After that, I heat the back cover using a heating mat to loosen up the glue that secures it. I use 40°C for 20 or 30 minutes. The cover is also secured with a half dozen tabs that need to be worked on. An X-Acto knife and a flat plastic spatula do the job. I try to use plastic tools in order to not puncture the battery contents or short-circuit anything.

Once it’s all opened, the cells need to be tested. I made an adapter so I can connect all three cells to a LiPo balance/charger, bypassing the circuit board. I use it to cycle the cells twice by charging them at 0.25C the first time and 1C the second. Discharges are done at 1C the first time and 4C the second time. I use 4C the second time because it’s how much current the drone “pulls” while at flight. [Editor note: he’s referring here to the so-called C-rate]

If the cells check out ok, I proceed to working on the BMS. I look to see if the cells balance correctly within 0.2 – 0.3v and if there isn’t much voltage sag. These cells seem to be of good quality and are therefore very resilient, but again remember that your and others’ cells haven’t been cared for and are not brand new, so I do not expect them to work as if they were.

To connect, gain access, and communicate with the BMS, I use the EV2400 from Texas Instruments. Communication is done via SMBus. That said, any SMBus-capable interface adapter could conceivably be used.

Access to the proprietary BMS is protected by passwords. You either already have them or you need to “break in” in order to gain access. I can’t tell you my passwords, but I know of programs out there that will “generate” codes and let you in.

I use BQStudio, also by Texas Instruments, to tweak parameters and code as needed to “erase” the permanent faults (PF). Doing so enables the associated MOSFETs to open or close, thereby once again successfully enabling charges and discharges.

After reprogramming it’s time to charge the battery, now using the DJI charger, to see if it’s successful. If it charges as it should, good!

Now, I put the battery in the drone and check for firmware updates via the relevant DJI application.

Finally, it’s time for the first flight. I like to fly low and slow for the first couple of minutes. After that, I fly normally and, if it all checks out, I fly enough until the battery reaches 50 – 60% charge left. I like to ship these batteries at about 50% charge which is about 3.85 V per cell. That’s the manufacturer’s recommended storage voltage.

It’s also recommended that these batteries be recharged every 4 or 6 months when not in use so that the voltage doesn’t fall below the BMS’s threshold, which would cause a PF flag to once again be set by the BMS. This goes for all “smart batteries”, i.e., batteries with built-in BMSs.

Safety is a must. Lithium batteries are flammable, so great care must be taken and common sense must be used when attempting to repair, charge, or use these batteries, and they must never be charged unattended. I’ve had two incidents happen, and it’s a little scary. I didn’t get hurt, but I had to buy a few new instruments for my lab.

I’ll augment Erik’s wise words with two additional quotes from the earlier referenced DJI documentation:

Discharge the battery to 40%-65% if it will NOT be used for 10 days or more. This can greatly extend the battery life. The battery automatically discharges to below 65% when it is idle for more than 10 days to prevent it from swelling. It takes approximately one day to discharge the battery to 65%. It is normal that you may feel moderate heat emitting from the battery during the discharge process.

and

Fully charge and discharge the battery at least once every 3 months to maintain battery health.

Likely unsurprisingly to many of you from what you’ve already read, Erik also noted in a subsequent message:

The components on the [battery] board are off-the-shelf. Both the system chip and microcontroller are from Texas Instruments.

And Erik even shared some photos with me, to subsequently reshare with all of you, of the various disassembly, rejuvenation and reassembly steps:

As mentioned upfront, I’m not the only person who’s been struck by “dead DJI battery” lightning. And, as it turns out, Erik’s not the only one who’s figured out how to revive ‘em. Check out, for example, this chronologically ordered series of videos from an outfit in New Zealand who specializes in such services (among other drone-related things):

In one of the videos, the technician even uses TI’s EV2400; the others, to Erik’s comments that “any SMBus-capable interface adapter could conceivably be used,” showcase an inexpensive USB-to-SMBus bridge board based on Silicon Labs’ CP2112 IC. And as far as software goes, the most common approach leverages a freeware utility called “DJI Battery Killer”. The tool’s lineage is unknown, although it seems to have originally come from Ukraine, and it’s seemingly disappeared from sites that originally hosted it (although in at least one case, I happen to know that it’s still downloadable via the Internet Archive Wayback Machine intermediary), so consider yourself duly warned, and no, I’m not going to provide any direct links. I’ll only note that the utility’s documentation (which, yes, I now have a copy of) specifically mentions that it supports several battery management controllers: the BQ30Z55, BQ40Z50 and BQ40Z307 (BQ9003).

In closing, I’ll repeat for emphasis one phrase from Erik’s earlier in-depth commentary:

That last-sentence reality check is both validating and deeply disturbing. To my right as I type these words, in addition to various drones’ batteries, is my expensive set of multi-cell D-Tap video batteries that I mentioned at the end of last year and again a month later, along with a variety of cameras and proprietary batteries for them. In front of me, of course, is my integrated multi-cell battery-based laptop computer. And to my left are multiple wireless headphone sets. More generally, scattered around the house (including in long-term storage) are any number of devices that run on Li-ion, LiPo and other battery chemistries, with most of those cells fully integrated and difficult-to-impossible to replace.

All of them, it seems based on this and past experiences, are essentially ticking time bombs, just sitting there waiting to die if they’re not regularly topped off. While I’m not so cynical as to knee-jerk label them all as examples of “obsolescence by design”, I’m also not a Pollyanna inclined to completely discount this possibility in all cases. Regardless of whether it’s an upfront intention or just an unfortunate side effect of today’s dominant power source technology, it’s a sooner-or-later source of outrage for consumers, not to mention a root reason for premature, excessive landfill donations. I welcome others’ perspectives, including insights into up-and-coming battery technologies and implementation techniques that don’t exhibit this Achilles heel, in the comments.

—Brian Dipert is the Editor-in-Chief of the Edge AI and Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.

Related Content

<!–
googletag.cmd.push(function() { googletag.display(‘div-gpt-ad-native’); });
–>

The post Diagnosing and resuscitating a set of DJI drone batteries appeared first on EDN.

Source link