Info: When Balancing Occurs, and Pack Maintenance

[kennybobby]

Here's a theory open to peer-review and open critique:

As Otmar has stated "The open circuit voltage at no load, ignoring internal resistance, can indicate SOC of a brick, but SOC is just indicating a percentage of available Ahrs from that brick."

My theory is that effective balancing can only occur when the main contactors are open, and this is also the condition when a meaningful OCV can be measured and SOC can be determined.

The balancing circuit is controlled by a TI chip, some aspects of which are noted in the Technical Forum here.

Once the desired balancing set-point voltage is loaded into the memory registers and the enable register flag is set, then balancing will be turned on whenever the brick hits that value regardless of the main contactor position, e.g. at 93% during charging. But to be effective the contactors must be open, since balance current is only ~ 100 mA for a whole brick.

There are 6 bricks per module, 16 modules per 85kWh pack. The balance resistance is 39.5 Ohms. It is left as an exercise for the reader to calculate the total power consumed if the entire pack was balancing for 24 hrs, and compare this to the vampire load.

 

[Quantum`]

Gary it couldn't be on a cell basis as they're paralleled in each brick with no individual MOSFETs. (that would be too expensive for such tiny cells)

Kenny, keep in mind that the balancing chip on each brick is slaved to the main BCM, it's not operating locally. This was Tesla's (wise) design choice. I was the first to identify the chip's function and to point this out.

In any case, all this 90% business is just a rumor, probably started by wk057. There is absolutely nothing about not charging the pack to full in the User's Manual nor the Service Manual, nor in any advice from the SC.

And I am sure that balancing does not occur below some given SoC as cars have sat for months and only depleted proportional to their Ri. It makes sense that balancing only occurs while charging, as a) there is no real load on the pack so true readings can be had, and b) it would reduce range if done while on the road.

 

[kennybobby]

... that the balancing chip on each brick is slaved to the main BCM, it's not operating locally.

My theory is that balancing is triggered by a set-point voltage, not a SOC or %. Once loaded into the memory register from the master BMS board, aka BCM, then it is free to do it's thing once the enable flag is set.

Seeing is believing--this shows local self-activation of balancing from wk057's thread, hope he doesn't mind, probably won't see it anyway due to ignoration policy...

 

[apacheguy]

Yikes, this balancing stuff get complicated fast. So my best bet now is to ensure the main contactors have opened so there is no load on the pack? What do I have to do next? Do a song and dance and fan my S with Palm fronds?

 

[wk057]

Gary it couldn't be on a cell basis as they're paralleled in each brick with no individual MOSFETs. (that would be too expensive for such tiny cells)

Kenny, keep in mind that the balancing chip on each brick is slaved to the main BCM, it's not operating locally. This was Tesla's (wise) design choice. I was the first to identify the chip's function and to point this out.

In any case, all this 90% business is just a rumor, probably started by wk057. There is absolutely nothing about not charging the pack to full in the User's Manual nor the Service Manual, nor in any advice from the SC.

And I am sure that balancing does not occur below some given SoC as cars have sat for months and only depleted proportional to their Ri. It makes sense that balancing only occurs while charging, as a) there is no real load on the pack so true readings can be had, and b) it would reduce range if done while on the road.

I don't start rumors.

My theory is that balancing is triggered by a set-point voltage, not a SOC or %. Once loaded into the memory register from the master BMS board, aka BCM, then it is free to do it's thing once the enable flag is set.

Seeing is believing--this shows local self-activation of balancing from wk057's thread, hope he doesn't mind, probably won't see it anyway due to ignoration policy...

View attachment 111224

That FLIR shot shows that the balancers stay on once commanded by the main BMS, and stay on until commanded otherwise. In this case it actually turned out that they had run for far too long and were in fact pulling the module OUT of balance. I had speculated that they may be acting on their own and making the decisions, but turns out they were just following their last orders. Something to keep in mind if breaking down a pack and removing the BMS.

I've confirmed that the balancers only ever enable while trying to reverse engineer the protocol used to talk to the BMBs (the boards on each module with the balancers). I could throw cell groups way out of balance, and the BMS would still not enable any balance resistors until the pack was in a CV charge mode (constant voltage, taper portion, close to full), at which point immediately the expected balance resistors came online for the groups I had not partly discharged.

This could have changed at some point, but I don't know why it would. The cells are in fact top balanced.

 

[Cottonwood]

Yikes, this balancing stuff get complicated fast. So my best bet now is to ensure the main contactors have opened so there is no load on the pack? What do I have to do next? Do a song and dance and fan my S with Palm fronds?

It's really pretty simple. Reading from earlier posts by @wk057 and @Todd Burch, as well as insight from many other threads, here is my view of getting a pack balanced and calibrated. Comments are welcome.

1. To ensure balance, charge to over 93% (I like 95% or higher to make sure), then wait a few days for balancing to complete. Going over 93% SoC triggers a good balance measurement at the constant Voltage part of the charge cycle. Once the balance measurement has been done, go ahead and use that extra SoC and discharge back to 90% or less. The balance measurement is used to bleed the charge off of stronger bricks[SUP]*[/SUP] to bring the bricks into balance over the next few days. This bleed cycle can bleed off about 1% per day of charge, so giving the process a few days, allows for a worst brick correction of a few percent. The bleed/balance operation will continue at any SoC.
2. To get a good calibration of the battery pack capacity, charge once to over 90% and then discharge down to 10-20% (slightly lower is better for low end calibration). If you have time, do this after the above balance cycle; the calibration will still be pretty good if you overlap balance and calibration.
3. If at all possible, don't leave your pack for too long a time at over 90% or under 20% SoC. No need to be parinoid, but try to correct back into the mid-range in less than a day or so; correcting within hours is slightly better. High SoC for long periods at high temps should be avoided.
4. Charging to 100% does not help balancing, but don't worry about charging to 100% SoC before a long trip, or long segment. Just don't leave the pack at a high SoC for too long.

[SUP]*[/SUP]The battery pack in the 85 consists of 7,104 individual cells. These are arranged into 96 "bricks" of 74 cells in parallel (96*74=7,104). The 96 bricks are packaged into 16 modules of 6 bricks each. In each module, the 6 bricks are in series. The 16 modules are in series, putting all 96 bricks in series. Each of the 96 bricks has it's own bleed resistor set for balancing, i.e. there are 6 bleed resistor sets in each module.

 

[Todd Burch]

So, if making the equalization decision only while charging, and when only when over
93% SoC ... that would seem to be a substantially sub-optimal design decision.

I'm sure the determination of how much bleeding is necessary is made after charging has stopped or slowed down significantly.

IMO, it is more likely that the start and stop decisions are made on a "cell" basis,
in real time, and perhaps only when stopped or parked.

But there are over 7000 cells in the Model S. There aren't that many bleed resistors, and there isn't that much silicon to switch at such a granular level. There isn't a reason to have that many anyway, since the cells in parallel will (by nature of being in parallel) be at the same voltage.

 

[OddB]

1. To ensure balance, charge to over 93% (I like 95% or higher to make sure), then wait a few days. This triggers a good balance measurement at the constant Voltage part of the charge cycle. The balance measurement is then used to bleed the charge off of stronger bricks[SUP]*[/SUP] to bring them in balance. This bleed cycle can bleed off about 1% per day of charge, so giving the process a few days, allows for a worst brick correction of a few percent.

1. 
   
   Are you saying ; Charge to 95%+ and leave the car at that SOC, without driving for a couple of days ?

 

[mspohr]

Yikes, this balancing stuff get complicated fast. So my best bet now is to ensure the main contactors have opened so there is no load on the pack? What do I have to do next? Do a song and dance and fan my S with Palm fronds?

I think the point of what the tech told the OP is that you don't have to worry about it. The car will take care of balancing by itself and you don't have to do anything. Just follow Tesla's recommendation for daily charge up to 90% and leave it plugged in overnight when possible or when you are gone for extended periods. If some special procedure was necessary, it would be in the manual. Just relax and enjoy your car.

 

[LargeHamCollider]

My only comment is that I has to be tricky for the BMS to turn on the bleed resisters because of the flat charge curve. Perhaps this chemistry has more slope than the LIFEPO4 cells I have observed.

Yes, NCA has way more slope than LFP.

 

[kennybobby]

fyi there's lots of brick voltage data found in Lolachampcar's CAN bus logging thread in the driving dynamics section, e.g.

 

[Cottonwood]

Are you saying ; Charge to 95%+ and leave the car at that SOC, without driving for a couple of days ?

Sorry, no; poor wording. I updated my post to try to make it more clear.

Take the SoC to 95% then discharge to 90% or less within hours. Wait for days for balancing to complete.

 

[Quantum`]

Or, charge it to 100% and leave it plugged in. It will only balance while plugged in.

100% is not 'radioactive'. Ask your SC.

It's clear here that balancing is only taking place at the top.

 

[cynix]

Or, charge it to 100% and leave it plugged in. It will only balance while plugged in.

What do I do if I have no home charging facilities and rely on superchargers/public charging, and therefore cannot leave it plugged in for an extended period of time?

 

[Todd Burch]

Or, charge it to 100% and leave it plugged in. It will only balance while plugged in.

Whoa, whoa, whoa. This information is contradictory. What do you base it on?

 

[zwede]

I don't see any reason it would only balance while plugged in. Seems it needs to be charged to the CV phase (>~93%) at which point the BMS figures out how many watt-hours it needs to bleed off from which bricks. It will then enable the bleed resistors and leave them on for a specific time. It doesn't matter if you drive the car before balancing is complete as the BMS already knows how much it needs to bleed off. The balancing can take days.

That's what I seem to glean out of several threads on the subject, at least.

 

[CygnusX1]

Any idea where this obvious periodicity within the 96 brick voltages comes from? Different voltage drops due to different brick positions within the pack repeating in a regular order? This could explain the "balancing" at the top: When charging current is reduced, voltage drop is reduced as well... in this case, the differences in voltage would not be real but a current artefact. Just a theory...

 

[wk057]

Or, charge it to 100% and leave it plugged in. It will only balance while plugged in.

This is false. The car does NOT have to be plugged in for balancing to continue once it's started.

 

[ecarfan]

I think the point of what the tech told the OP is that you don't have to worry about it. The car will take care of balancing by itself and you don't have to do anything. Just follow Tesla's recommendation for daily charge up to 90% and leave it plugged in overnight when possible or when you are gone for extended periods. If some special procedure was necessary, it would be in the manual. Just relax and enjoy your car.

Ah, a voice of reason. Thank you.

 

[Electroman]

I don't see any reason it would only balance while plugged in. Seems it needs to be charged to the CV phase (>~93%) at which point the BMS figures out how many watt-hours it needs to bleed off from which bricks. It will then enable the bleed resistors and leave them on for a specific time. It doesn't matter if you drive the car before balancing is complete as the BMS already knows how much it needs to bleed off. The balancing can take days.

That's what I seem to glean out of several threads on the subject, at least.

That is exactly how I understood from reading all the posts in this thread. You summarized it nicely.