cell balancing

cell balancing

Hi folks,

the following question was just asked in the German Tesla forum:

when does cell balancing / conditioning of the battery take place? Is that a continous process or does that only take place when the car is parked? Does the car have to be plugged in for this?

derotam | December 30, 2019

I don't have absolute knowledge of this but if we think about it a sec... It is at a minimum maintaining balance during charging. It could also maintain balance during discharge as well. The car has control at least to the module level on how power goes in and out.

gmr6415 | December 30, 2019

Generally speaking cell balancing is done at the end of a charge while still plugged in. Admittedly, I'm not as up to date on LI technology as compared to other battery technologies.

Years ago I built a battery conditioner for my 2003 Honda Civic Hybrid (NIMH battery pack) after it started showing codes for cell imbalance and cell balancing was achieved after 3 cycles of deep discharge with each discharge going farther...almost to the state of polarity reversal on the third charge, full recharge and soaking after full charge for up to 12 hours.

After perfecting it with the battery pack out of the car I installed a cable in the trunk which turned my 2003 HCH into a plug-in hybrid that would not only condition the battery, but top it off when not driving.

gmr6415 | December 30, 2019

"almost to the state of polarity reversal on the third charge"

should be: almost to the state of polarity reversal on the third discharge

derotam | December 30, 2019

@gmr6415: I don't believe that is accurate. I think it is continually balancing. Should be pretty easy to test... Only question I have from you would be what kind of time frame at the end of the charge cycle do you propose that it does the balancing?

Testing the theory can be accomplished by monitoring the cell imbalance value from the CANbus/OBDII and cutting the charging off before your proposed balancing time period, for multiple charges. Doing this with a supercharger should theoretically quicken the imbalancing vs any AC charging.

gmr6415 | December 30, 2019

@derotam, As I stated above "generally speaking" and "Admittedly, I'm not as up to date on LI technology as compared to other battery technologies."

With the battery technologies I've personal worked with, not LI, the pack must be close to full charge before the pack can balance. In the simplest terms some cells will reach full charge before others and a slow soaking at reduced amperage after what is considered collective pack full charge will slowly bring the lagging cells up to full charge or closer to full charge if the cells aren't so degraded that they can't accomplish a few percentage points within full charge.

With the NIMH pack from the 2003 HCH I individually monitored the "sticks", which if my memory is correct were the equivalent of 9 D cell batteries welded together in series, in order to find sticks of lower voltage within the pack (indicating one or more weak cells in those sticks) when the pack showed full charge collectively. A 2% variance between sticks would trigger a code on the car which is what I was trying to get around on a very old pack.

By Honda's specks that pack was fully charged at 144 volts DC, yet with my conditioner it didn't reach full soak until it got to around 174 volts DC, so after my initial trial I adjusted my conditioner to top out at 174 volts DC.

While doing a slow low amperage soak (350 mA) of the whole pack, after full pack collective voltage was reached, using not enough amperage to damage the fully charge cells, yet enough to trickle charge the sticks showing lower voltage, the weaker sticks would balance out with the higher voltage sticks over time...taking up to 12 hours of soak after the whole pack had reached what was considered collective pack voltage.

I could get into much more detail, but I'm not trying to write a book here.

I extended the life of a battery pack Honda had condemned for almost 4 more years by using my conditioner on a regular basis. In other words my charging technology made up for the shortcomings of Honda's charging technology of the day.

For economy reasons the HCH would only charge the pack to about 80% and discharge it to around 50%, i.e. the car had to burn fuel to charge the battery pack, so Honda picked a balance between battery pack life and fuel consumption. With NIMH batteries that in itself results in crystallization at the anodes. My process eliminated most of the built up crystallization caused by years of the designed charging cycle.

I don't know this to be true of Tesla or LI. I was speaking in general.

bruryan | December 30, 2019

From what I have gleaned from Jack Rickard, cell balancing is done at the top of the charge. Kind of like flooding a rice field. I have attention deficit problems, and hallucinations though. So in all likelihood, once the cells start getting a charge they self-level is my thinking. The only reason I say that is I doubt you need to charge to max to level the cells.

Patronus | December 30, 2019

This soaking at high charge states may be why Tesla (and Elon in particular) recommends charging to 90% nightly.

Frank99 | December 30, 2019

Speaking as an engineer (but not as a battery engineer), I would guess that Tesla balances the cells at the end of a charging session - which doesn't have to be a full charge - when they detect that there's a noticeable imbalance in the cells.

The Model 3 LR battery packs are composed of 96 groups of cells in series, each group has 46 (IIRC) cells in parallel. Basic physics requires that all 46 cells in a group will have the same voltage. So, the charging system is capable of measuring the voltage of each group of cells (96 different measurements).

While charging, they will be measuring the voltage across each group of cells, and they'll terminate high-speed charging when one group of cells reaches the voltage that corresponds to the charge limit you've set - 80% or whatever. They then have two choices - they could discharge individual groups that have a high voltage to match the voltage on the lowest group of batteries, or they could charge individual groups that have a low voltage to match the voltage on the highest group of batteries. I'd guess they use the latter, but both are reasonable.

Unlike the NiMH batteries that gmr6415 is used to, LiIon batteries really hate any amount of overcharging - you can't trickle charge the pack at a low rate (how I interpret his method) to charge individual weak cells, because the other cells will get overcharged. Not a problem for NiMH, but a big one for LiIon. That's why you have to charge individual groups - the cells are all in parallel in a group, so all the cells are at the same voltage at the same time; you can charge to precisely 4.15V (or whatever Tesla calls 100%) and all 46 cells in the group will be at 4.15V.

dave | December 30, 2019

@gmr6415 I wish I had known you in 2013.
My 2003 Honda Civic Hybrid was still running great, at 150K miles. But the check engine light was always on, with an error code from the battery pack due to one bad cell. The only solution from Honda: replace the entire battery for $3500.

My next car, a 2013 Toyota Prius, was better in many ways but surprisingly never reached the mileage of the Prius.

My Tesla Model 3 is better in so many ways, and getting better all the time!

jimglas | December 30, 2019

thanks for the explanation Frank99

jordanrichard | December 30, 2019

So the assumption here is that the original question on the German forum is actually asking about cell balancing or about balancing the pack for the sake of getting an accurate range number.

If it is the latter, it is the algorithms that get wonky if one keeps the SOC within a short window like 20%. Meaning starting at 80% and then only going down to 60% before charging back up to 80%. The packs needs to be "exercised" from time to time by running it down to about 10% and then back up to 100%. It's sort of like a rubber band in that if you only stretch the battery out by "x" amount, over time you forget just how far it can really be stretched. | December 30, 2019

It's my understanding for the S/X pack that the balancing is only done after you charge to or above a voltage threshold, thought to be near 93% SOC and you finish charging. For example, you charge to 95%, then when it gets to 95% it stops charging.

At that point, the balancing is done by adding a small resistive load to the groups that are higher than the lowest group within the module until each "high" group is brought down to match the lowest group. These resistors are on the BMS boards within each module and only drain at 100 mAh per group. . The balancing is only done when no more charging is needed, the car is not driving, and likely the contactors are open (i.e no major loads). Balancing can take several days to complete, but there is no issue if interrupted and you drive off. It already knows what is needed and will continue balancing when the car next goes to sleep. The cells are never "overcharged" to do load balancing.

I've not heard any details of how load balancing is done on the Model 3. Likely it is similar, but it could be different or have different SOC thresholds. I also do not know if there is any value in forcing a balancing every so often by charging above 93% SOC. Perhaps balancing once a year may be good. Since Tesla has never stated it is important, I'd not worry about it. Just use the car normally :)

gmr6415 | December 30, 2019

@dave, Mine had around 210,000 miles on it when I purchased my model 3. I had found Panasonic replacement sticks online that had 40% more capacity than the original sticks Honda was using. I think those sticks cost me $900 or so.

Honda was reconditioning batteries. They didn't sell new replacement batteries. I tried and they were all reconditioned. They would test the sticks in returned packs and only replace those sticks out of parameters, then use those reconditioned packs for replacement in customer cars at a cost of around $3000 to $3500 including labor.

In the end I'm not sure how much the additional capacity really helped. The regeneration in those cars was minimal and it took shadow charging to charge the battery beyond what regeneration could do. Shadow charging consumes fuel, but Honda balanced it pretty well, so you really didn't often feel a loss of power. After installing the higher capacity sticks I was just burning more fuel recharging the greater capacity. I'm not really sure it did anything. At least not enough to make a noticeable difference in fuel mileage. You could floor it and get that electric motor boost a lot longer though when needed.

The difference was the fact I could come home with a half charged battery, plug in overnight and start out with a full battery the next day. That was still minimal.

And thank you Frank99 for a better explanation than what I could do with my limited knowledge of LI battery packs. I have an electronic battery charger for RC cars. When set to LI it will do deep cycles of the pack. I can only assume it's using a different method of balancing the smaller LI rechargeable packs than what it does for NIMH.

Frank99 | December 30, 2019

gmr6415 -
If you have RC cars, you should notice that the battery pack has two connectors - one with big wires, and one with a number of small wires. A small wire goes in between each cell in your pack so that the charger can measure the voltage of each cell, and can either charge (with a very small current) or discharge each cell so they match.

TeslaTap -
Great backstory on S charging; I didn't know any of that. Sounds like a reasonable approach, and wouldn't surprise me if it was precisely how Tesla does it on the Model 3.

sixstring09 | February 20, 2020

Cell balancing works during the charge cycle. The management system monitors each cell during charge. If there are cells that are lower in voltage than the others, it will send more charge voltage to those cells. If there are some cells that are at higher voltage than others, or at the charged voltage, the management system will not send charge voltage to those cells. If there are cells that are over charged, it will bleed off voltage. If there are cells that are not charging properly, they are taken out of service by the management system. Users do not have access to this data. This is how it works. | February 20, 2020

@sixstring09 - Not quite as complex as you state. My understanding is cells are attached in groups, 86 cells in the case of an S100 pack. There is no ability to tweak individual cells, as cells within a group are hard-wired in parallel.

The pack has 96 groups, each is monitored for voltage and temperature. Charging completes when a single group reaches the set charging threshold. There are bleed resistors for each group that can slowly bleed the cell group voltage to match the group with the lowest voltage. Ideally, the pack is balanced when every group has the same voltage. This process used to only occur when charged to 100% and left to sit for hours, but there have been changes to the process, and it may occur at lower SOC levels now.

While I'm not aware of Tesla going into details, I suspect cells are tested and matched to go into the same group. The manufacturing process may be at such a high-quality level now that this is not done - i.e. all the cells when made are very close to each other. Just some of the internal details we don't know about.