An experience with XIAOMI Mi robot (Roborock) vacuum cleaner (model n. SDJQR02RR).

DISCLAIMER:

The information provided here is intended only to share a specific my own experience. Due to the serious risks to people and property involved in this type of activity, this information is in no way to be considered as a suggestion and/or an invitation to do the same.
This information can be incorrect, misleading and even dangerous.
Everything described here can lead to unexpected results.
I explicitly discourage anybody from repeating similar experiments.

Symptoms:

My cleaner stops working mid cleaning cycle. It goes in a OFF state. The only way to resuscitate it is connecting it to the charging station.
Once it comes back to life I can start a new cleaning cycle, but the cleaner stops again after few tens seconds.

Investigation:

I have managed to replace the 185 resettable fuse on the cleaner's mother board as per “Internet” suggestions: no change.

In order to further diagnose the problem I removed the battery pack from the unit and, after having carefully read the Safety Data Sheet for the product BRR-2P4S-5200S, I’ve removed the bottom (or top, depending from where you look at it) cover getting access to the internal BMS circuit board (the cover is glued, but with a little patience I managed to remove it with very little damage to the plastics).

It looked like a 4S2P (2 Parallel, 4 Serial cell elements) configuration BMS: the controller chip is an S-8254A and the cells are LG’s LGABB41865.

The LGABB41865 main specifications are as follow:

The typical schematic for a basic 4S2P BMS based on S-8254A is the following: 

 

Please note as the BMS continuously verify overcharging, overdischarging, overcurrent conditions for all the 4 battery serials individually. It is sufficient for just one of those parameters to go out of limits in any of the 4 monitored battery serials for the BMS to fall into protection mode, blocking all functions (isolating the output wires from the cells). If you want to dig further into the BMS operations details have a look at its datasheet. 
Additional circuitry is present to control other safety parameters like overtemperature, ecc.

 

 The actual pack layout is as follow: 

Here are the steps i followed to test the battery pack (I had to carefully respect the polarity of the connections: red=positive, black=negative):

1. Charged the pack. I used the cleaner itself: this procedure is safe and certainly correct. An alternative procedure -not tested, less safe and may be non really correct- could have been to connect the pack to a bench power supply set at 16.8V and limited current at 1300mA.

2. After an hour or so, I disconnected the power supply from the pack. 

3. Then I’ve connected an automotive bulb (12V/21W) to the red and black wires. The lamp immediately turned ON. As it produced a great amount of heat so I had to find an appropriate location where to place it. 

Under normal conditions the lamp should stay ON for hours. Actually the lamp turned OFF after a while, let say after a time lapse similar to the one the cleaner stopped working while cleaning: I was now sure to have found the fault: the battery pack.

Then I wanted to drill down further to identify witch battery serial was the real culprit:

1. I tough that to reset the BMS out of the protection mode, I had to have simulate a "charging session”. So I sat the power supply to 16.8 Volts and 200 mA current limit; touching the red and black wires for just no more than a second, even without disconnecting the bulb, moved the BMS out of protection mode. Some times, to get the lamp turning ON, I had to disconnect it form the pack and reconnect.

2. Then, with the lamp turned ON, I’ve measured the voltage across the all four serials. Millivolts count here! They measured all the same, around 3600 mV +-10 mV.

3. I then kept measuring all of them four in turn to monitor them discharging. As soon as one of them reached the specified cutoff-voltage of 2.7 Volts, the BMS went into over discharging protection mode and the lamp turned OFF. 

This means that one of the four series (4S in my case) is more worn than the others and it is driving the BMS into overdischarging protection mode too fast.

I thought to replace the two cells in the 4S serial, but I gave up because this way you will introduce a further imbalance into the pack and, sooner or later, another serial will become the weakest and the problem will reappear exactly as it was.

Conclusion:

I bought a new battery pack for my cleaner and retained the good cells of the bad pack for future experiments. 

Curious to see the cells discharging under load? Here they are: