Cell Balancing in Leaf

[Ingineer]

Does anyone actually know if the Leaf has any method of active balancing in the car? My guess is not.

Most production large-format lithium systems do not have anything except monitoring. I know for sure the PHV (PHEV) Prius does not.

FYI: Here's what inside a Battery ECU on the PHV: (there are 3 of these)


Here's the Pack:


Mod update :
Split from here : http://www.mynissanleaf.com/viewtopic.p ... 46&start=0

[DaveEV]

Does anyone actually know if the Leaf has any method of active balancing in the car? My guess is not.

Yes, it does. When charging it monitors the voltage and tapers the charge based on the voltage of the highest one. At which point it appears to discharge those cells with the highest voltage to come more in line with the rest at which point it will charge the pack a bit more. It appears that the Leaf will do this at least twice if left plugged in long enough.

What's unknown is how fine-grained the balancing is and if it simply burns off the energy from the higher voltage cells or directs that energy to the lowest voltage cells...

(There's a thread around here somewhere that discusses this)

[Ingineer]

Yes, it does. When charging it monitors the voltage and tapers the charge based on the voltage of the highest one. At which point it appears to discharge those cells with the highest voltage to come more in line with the rest at which point it will charge the pack a bit more. It appears that the Leaf will do this at least twice if left plugged in long enough.

What's unknown is how fine-grained the balancing is and if it simply burns off the energy from the higher voltage cells or directs that energy to the lowest voltage cells...

(There's a thread around here somewhere that discusses this)

I just looked at the service manual (EVB section), and the system description shows each cell has a "bypass switch" (bleed transistor), so it actually will bleed off charge energy such that cells that are full will no longer charge, yet the other ones in the series string will continue to. I would assume the bleed current is relatively low, so the charger probably is set to a low constant current during this top equalization process.

[DaveEV]

I would assume the bleed current is relatively low, so the charger probably is set to a low constant current during this top equalization process.

That's the image someone else posted earlier.

The charge behavior we have seen goes like this (Jimmydreams has a TED on his EVSE):

Charge at 3.7 kW AC until approach full when power tapers off over about 30 minutes
Wait a couple hours
Charge at 3.7 kW AC and immediately start tapering
Wait another couple hours
Charge at 3.7 kW AC and immediately start tapering

In between the separate charge events there is no apparent power going to the car and given the text in the image "bypass switch is turned ON to perform the discharge of a cell with the high capacity" one has to assume that the higher charge cells are being discharged during those wait periods.

Here's a link to the thread where you can see this behavior:
Charging timer: start time based on END time?

[Ingineer]

This is not how top balance works. The "bypass switch" is pretty much exactly what the Japanese translation implies, it bypasses charging current so the cell in question stops charging, and the others continue. It is not used to later discharge the cell, it is only on when the charger is.

The cycling you are seeing is likely due to the charge carrier stratification phenomenon. Nearing the end of charge, the charger tapers (from CC to CV), all cells come up to full charge voltage, and then the charger shuts off. However, after a short time the extra charge carriers (Lithium Ions) migrate to the electrodes, then the cell voltage falls and charging re-starts. This process will likely continue if the EVSE is left connected but the gaps between restarts will get larger. I would guess the car will indicate full charge (or very near it) before this.

Another thing that can temporarily halt charging is Delta-T rise. When the cells are near full, the charge efficiency drops and more heat is produced. The BMS might pause charging with high temps to allow cool-down before resumption. These thermal concerns are exacerbated by the dissipation of the "bypass switches" which are converting the energy that ordinarily would enter the cell into heat.

That's the image someone else posted earlier.

The charge behavior we have seen goes like this (Jimmydreams has a TED on his EVSE):

Charge at 3.7 kW AC until approach full when power tapers off over about 30 minutes
Wait a couple hours
Charge at 3.7 kW AC and immediately start tapering
Wait another couple hours
Charge at 3.7 kW AC and immediately start tapering

In between the separate charge events there is no apparent power going to the car and given the text in the image "bypass switch is turned ON to perform the discharge of a cell with the high capacity" one has to assume that the higher charge cells are being discharged during those wait periods.

Here's a link to the thread where you can see this behavior:
Charging timer: start time based on END time?

[garygid]

Ing...,
I believe your "bypass charging current" is incorrect.
The switch would short out the cell.

Do you not believe the translation in the Service Manual?

[Ingineer]

Ing...,
I believe your "bypass charging current" is incorrect.
The switch would short out the cell.

Do you not believe the translation in the Service Manual?

It is not a "short" circuit, but a calibrated load. It simply bypasses the charging current which is probably well under an amp during top balance. It is typically a MOSFET with a series resistor, but in the Leaf I suspect a FET-type device in an ASIC not driven to full gate charge so it can dissipate the charge current. Some dead-simple (and flawed) balancers some people have constructed use a zener diode for this purpose, so it functions similar to that; when the voltage hits a certain point it switches on the load and prevents further charging in the cell. Just as in a more complex computer controlled system, it only conducts enough to prevent further charge, not discharge.

I have a similar top-balancing system implemented on my 6.5kWh pack of 864 A123 26650's that's in my Prius Plug-in conversion. It has 72 series strings of 12 parallel cells each, and on each of the 72 strings there is a tiny BMS board that contains a microcontroller. The micro can enable a 250ma load (~1 watt) when the cell hits 100% SOC to bypass the charge current. Each tiny board has voltage and temperature monitors, and communicates serially with it's downstream neighbor. The serial bus terminates in the battery ECU which controls charging, and when all 72 boards report full charge, the charger is instructed to shut down.

The SM translation gets the point across, but doesn't describe much detail.

[garygid]

Yes, this only works if the "balancing" current is low (like 1 amp, or 0.5 amp, or zero), something that a shunt can handle without generating too much heat.

However, we have not yet seen any evidence of a LOW-current phase. Rather, it seems to be more like a 45 minute period of essentially NO current.

[GroundLoop]

I have the same system in hobby balancing chargers.

When the first cell reaches full voltage, the charger drops to half-amp (regardless of capacity), adds a (resistive) load to just the one cell via the balancing taps, and charges the remaining cells at half-amp. One by one they get the same treatment.

The limit is because that's the maximum power the balancer can dump.

[Ingineer]

Yes, this only works if the "balancing" current is low (like 1 amp, or 0.5 amp, or zero), something that a shunt can handle without generating too much heat.

However, we have not yet seen any evidence of a LOW-current phase. Rather, it seems to be more like a 45 minute period of essentially NO current.

"Essentially no" != None

When in balancing, my charger is around 70w draw. Are you sure this isn't what's occurring? It might even be less than that. Let's say the Leaf uses 50ma of balancing current, that would be around 20 watts. That would be consistent with an ASIC-based balancer design.

Keep in mind a pack that's healthy with good cell parameter parity wouldn't need much balancing, so even a minuscule current would be sufficient.