LFP v’s NCA… and the winner is…

[fhteagle]

 

[AAKEE]

Usually “calibrating” is confirming the starting point for the Coulomb-counting method of determining SOC. If you don’t get into the extreme low or high <20%/>80% the voltage is essentially the same so you don’t really know if you are 50% or 65% SOC as an example as your starting assumption may be out of whack.

The real SOC is not known until after the charging. (except for a full charge where the maximul votlage was held until the current dropped until the pre defined value = 100% SOC. (of course some cells will be lower, and the voltage of these we do not see until we stop charging.

During the end phase of the charge where tesla states "calibrating" there is a flow of current into the battery and the highest cells are held at 4.20V and the descendants are slowly advanging ahead. This will reduce the imbalance in the battery.

Cell balancing is different and is bleeding off the high voltage cells (relatively) to align with the lowest voltage cell. Charge for a bit, then rinse and repeat. This is a lot of the reason charging >80% takes much longer as they have to make sure not to overshoot, rebalance, charge, rebalance…

Active balancing means charging/keeping the cell voltage at 4.20V but draining the high cells, like any better home charger for hobby toys.

Tesla has only passive balancing, that uses resistors to draing the high cells. Mostly performed after the charging session when the car is at sleep. So basically the cell balancing is done after the charging, but still in the last phase of a charge, keeping the cells at 4.20V, the imbalance reduces as there is a higher flow of current to the cells with lower voltage.

 

[AAKEE]

It was a good video.


I just missed one part.

There was not a very clear description about that the car reads and updates the "real" SOC when it is not in use.
The coloumb counting is when driving/charging.

Also, they never discussed the issue that coloumb counting needs a well defined and measured battery capacity (that is correct), otherwise the result of the coloumb counting will be wrong. The part where they discussed the SOC change after a sleep following a drive, they never discussed this issue. Its for sure one of the most frequent reasons to see the SOC go up or down after the drive. Coloumb conting means using the intitial SOC converted to energy in the pack(SOC*estimated capacity) and reduce this with the measured energy/current which gives the new energy on board. Then convert it to SOC again by dividing energy on board with estimated capacity.
If the estimated capacity is not correct, the calculated SOC number will be off - which corrects after a drive when the BMS get the OCV reading.

 

[gtg465x]

If you’re curious, try finding posts here and on Reddit about Teslas with NCA or NCM batteries dying and needing replacement. One person already posted an example in this thread, and you can usually find multiple example posts per week, even from new Teslas that are less than a year old. After that, try to find posts about Teslas with LFP batteries dying. It’s almost impossible. Pretty sure a massive pro for LFP batteries, at least the ones Tesla uses, is reliability. Part of the reason could be not so much the chemistry itself, but the more modern and much more simple pack design that evrepair mentioned.

 

[Dr. J]

If you’re curious, try finding posts here and on Reddit about Teslas with NCA or NCM batteries dying and needing replacement. One person already posted an example in this thread, and you can usually find multiple example posts per week, even from new Teslas that are less than a year old. After that, try to find posts about Teslas with LFP batteries dying. It’s almost impossible. Pretty sure a massive pro for LFP batteries, at least the ones Tesla uses, is reliability. Part of the reason could be not so much the chemistry itself, but the more modern and much more simple pack design that evrepair mentioned.

Or, conceivably, there could be fewer LFP batteries, or they could be newer and not yet to the point of failures. To answer this question, more data is needed. We need a failure rate, adjusted for age, I would say.