Info: When Balancing Occurs, and Pack Maintenance

[Quantum`]

I can now confirm this to be false. My last max charge was a couple months ago, followed by partial charges to less than 90%. Last Friday the car was set to max charge, BUT unplugged at 232 Rated miles and driven right away. 232 rated is more than 90%, but less than 93% SOC. One more partial charge from 180 rated to 209 over the weekend, then it sat until today. Service screen confirmed the battery is perfectly balanced(like a new car would read). It's safe to say that the BMS takes care of the pack so you don't have to worry about it, and that there is a ton of estimation drift that goes on with rated range.

This doesn't mean that I am false. It just means that your modules are very similar to one another. There is no evidence that balancing takes place below 90% nor when unplugged. I personally know of several cars which have sat for months at between 20 and 60% SoC and only lost proportionate to the Ri.

I for one will wait to see Gary's evidence.

 

[Soolim]

@Quantum
How often do you balance, i.e. charge above 93 SoC?
How often do you run flat (0 RM) the battery?
What is your km range at 100% SoC?

 

[Ampster]

It is possible that balancing is triggered not so much by pack state rather than by the first group of parallel cells (bricks?) That hits the balancing algorithm point. Obviously this can only be triggered during charging. This voltage might be around 4.05-.10 for this chemistry. There probably is also a high voltage cutoff (as @garygid pointed out) some where around 4.20. This is how it works in two BMSs (EMUS & Orion) that I have used. It is likely that this point would be reached at the higher SOC (state of charge) As mentioned the balancing would continue even after charging stopped.

If the above is true, then it is possible that an out of balance pack might trigger balancing at a state of charge less than 90%. Following that logic it is also an explanation for why if you never charge above 90%, your pack could still be in balance or conversely not be unbalanced.

 

[Cottonwood]

It is possible that balancing is triggered not so much by pack state rather than by the first group of parallel cells (bricks?) That hits the balancing algorithm point. Obviously this can only be triggered during charging. This voltage might be around 4.05-.10 for this chemistry. There probably is also a high voltage cutoff (as @garygid pointed out) some where around 4.20. This is how it works in two BMSs (EMUS & Orion) that I have used. It is likely that this point would be reached at the higher SOC (state of charge) As mentioned the balancing would continue even after charging stopped.

If the above is true, then it is possible that an out of balance pack might trigger balancing at a state of charge less than 90%. Following that logic it is also an explanation for why if you never charge above 90%, your pack could still be in balance or conversely not be unbalanced.

Very good point.

If I were designing this algorithm, I would measure the difference in Amp-Hr that it takes for each brick of 74 paralleled cells to hit the start of the constant Voltage charge state. The difference in Amp-Hr is a very good approximation of the Amp-Hr that need to be bleed off.

The constant Voltage trigger point is probably the best Voltage point at which to do delta-Amp-Hr measurements, but lower, but upper end Voltage points could be used as well.

 

[AmpedRealtor]

"Tesla recommends"? Reference please.

If you never charge above 90% your pack will always be out of balance. Take it or leave it.

The range estimate might be off due to not knowing the full top end voltage, but that is not a balancing issue. My 90% varies from 224 to 227 miles, has nothing to do with balancing.

- - - Updated - - -

Fine. You'll live with an unbalanced pack. Sorry but I really don't care.

Tesla service told me the pack balances all the time. Good enough for me.

 

[brianman]

balancing doesn't kick in until you hit the 93% charge level.

I've found such assertions suspicious for along time. In 2012 firmware, there was "Range" (100%) and "Standard" (92-93%). This was changed to "Trip" (91-100%) and "Standard" (50-90%). If 93% was a magic rebalancing point there are two problems:
1. Older firmware - A "low" standard charge (92%) would not rebalance. This suggests that some cars that rarely or never Range charge would get inconsistent rebalancing.
2. Newer firmware - Because 90% < 93%, all cars that rarely or never Trip charge would get inconsistent rebalancing. Some cars that use Trip charge would get inconsistent rebalancing when doing so.

Some might assert that "91-93% is the rebalancing threshold in older firmware" and "89-91% is the rebalancing threshold in newer firmware". This would be a more viable theory given the two noted problems.

 

[Ampster]

Very good point.

If I were designing this algorithm, I would measure the difference in Amp-Hr that it takes for each brick of 74 paralleled cells to hit the start of the constant Voltage charge state. The difference in Amp-Hr is a very good approximation of the Amp-Hr that need to be bleed off.

The constant Voltage trigger point is probably the best Voltage point at which to do delta-Amp-Hr measurements, but lower, but upper end Voltage points could be used as well.

I don't know how that would be done on a serial pack without a separate current meter on each brick. That could create impedence that could create it's own balancing issues. Even if you could measure the difference, it would depend on what state each brick was when charging started. Most balancing algorithms I know use voltage near the top because that is the easiest to measure. My experience is confine to off the shelf BMSs and I have no specific working knowledge of the Tesla BMS.

Defining the contstant voltage charge state presents its challenges as well. While charging these non Tesla packs I can set the max pack voltage and when the charger hits this point, current begins tapering. However since the cells are in series, each cell is in a unique state of its own. A cell that is at capacity could spike it's voltage as it approaches the steep part (the knee) of its charge curve, while others are at a lower voltage. That would be when the BMS would turn on the shunt resistor(s).

This may be more thread drift that necessary to understand the specifics around the Tesla algorithm.

 

[Soolim]

I don't know how that would be done on a serial pack without a separate current meter on each brick. That could create impedence that could create it's own balancing issues. Even if you could measure the difference, it would depend on what state each brick was when charging started. Most balancing algorithms I know use voltage near the top because that is the easiest to measure. My experience is confine to off the shelf BMSs and I have no specific working knowledge of the Tesla BMS.

Defining the constant voltage charge state presents its challenges as well. While charging these non Tesla packs I can set the max pack voltage and when the charger hits this point, current begins tapering. However since the cells are in series, each cell is in a unique state of its own. A cell that is at capacity could spike it's voltage as it approaches the steep part (the knee) of its charge curve, while others are at a lower voltage. That would be when the BMS would turn on the shunt resistor(s).

This may be more thread drift that necessary to understand the specifics around the Tesla algorithm.

For 85kWh pack I thought the 96 bricks are all in series, therefore the current for all bricks at the same, but voltage for each brick could be different depending on the SoC?

 

[Ampster]

Yes, I you are correct, I didn't clearly state the issue. The SOC of each brick could be estimated from the voltage and the Amphours going into the pack.