White Paper - Dissipative vs. nondissipative balancing (a.k.a.: Passive vs. Active balancing)

Which is better for a given application?

Note: this paper is extracted from section of 3.2.3.3 of the book "Battery Management Systems for Large Lithium-Ion Battery Packs ". The book has a more complete and up to date discussion of this topic.

Introduction

One of the functions of a BMS is balance the cells in a battery. The capacity of a unbalanced battery may be quite low. After balancing, its capacity is equal to the capacity of its cell with the lowest capacity.

A balanced battery is one in which, at some State Of Charge, all the cells are exactly at the same SOC.

The amount of current required for balancing a Li-Ion battery tends to be less than one may guess.

Balancing is different from Redistribution, a technology that transfers large amounts of energy between cells, in order to allow use of all the energy in the battery.

Balancing can be:

• Dissipative: energy is removed from the most charged cell and is wasted in heat
• Nondissipative: energy is transferred between cells and therefore it is not wasted

Traditionally, dissipative balancing is known as "passive balancing", and nondissipative balancing is known as "active balancing". Unfortunately, the meaning of "active" and "passive" balancing has been confusing and interpreted in other ways. The terms "dissipative" and "nondissipative" are not open to misinterpretation. Therefore, we would like to take the lead in attempting to change the usage of these terms in the industry.

The disadvantages of dissipative balancing are obvious:

• Wasting energy is not environmentally correct
• At high balancing currents, the generated heat may affect the cells

At a first glance, nondissipative balancing is better because it doesn't waste energy. In reality, nondissipative balancing does have some disadvantages:

• More components that dissipative balancing: higher cost, lower reliability, more occupied volume
• Power wasted in stand-by current may result in greater losses than for the equivalent dissipative balancing

Comparison between nondissipative and dissipative balancing

Here are some examples of applications, comparing dissipative and nondissipative balancing.

It is assumed:

Common to both	Dissipative	Nondissipative
4 V cell voltage when charged 12 months self discharge time for worst cell, 18 months for best cell For a 100 AH cell, that works out to 12 mA for the worst cell, and 8 mA for the best cell, with a delta of 4 mA That means that the BMS has to transfer an average of 4 mA / cell If able to balance 24 / 7. that is truly 4 mA; if only 10 % of the time is available for balancing, that works out to 40 mA average Imbalance: half the cells (meaning that 1/2 the cells have low leakage and require bleeding to match the high leakers) Balance maintenance only (no gross balancing)	$ 1 / cell 100 mA balancing current (= 0.4 W) 0 mW standby power 0% efficiency when wasting energy	$ 10 / cell 3 A balancing current (= 12 W) 50 mW standby power* 70% efficiency when transferring energy

*) Note that the assumption is that the nondissipative balancer uses a bit of power whenever turned on, whether or not it is actually transferring energy. This is the case with many but not all nondissipative balancers.

This table compares nondissipative and dissipative balancing in some typical applications.

Application	Dissipative	Nondissipative
UPS: 100 Ah (4 mA delta leakage) 15 cells in series Always plugged in the wall	On/off duty cycle during balancing: 4 % Power of waste heat: 0.1 W Cost: 15 $	On/off duty cycle during balancing: 0.1 % Power of waste heat: 0.8 W Cost: 150 $
Distributed power source: 1000 Ah (40 mA delta leakage) 300 cells in series Always plugged in the wall	On/off duty cycle during balancing: 40 % Power of waste heat: 24 W Cost: 300 $	On/off duty cycle during balancing: 1.3 % Power of waste heat: 22 W Cost: 3000 $
EV, PHEV: 100 Ah (4 mA delta leakage) 100 cells in series Charged daily, plugged in the wall 12 hours a day, 8 hours charge / 4 hours balancing	On/off duty cycle during balancing: 24 % Power of waste heat: 0.8 W Cost: 100 $	On/off duty cycle during balancing: 0.8 % Power of waste heat: 2.7 W Cost: 1000 $
Public transportation EV: 1000 Ah (40 mA delta leakage) 100 cells in series Charged every 4 hours, plugged in the wall 30 minutes each time	On/off duty cycle during balancing: > 100 %* Power of waste heat: 8 W Cost: 100 $	On/off duty cycle during balancing: 100 % Power of waste heat: 3 W Cost: 1000 $
HEV: 10 Ah (0.4 mA delta leakage) 100 cells in series Charged 1/2 the time while driving SOC kept at 50 % +/- 20 % Balanced once a week for 10 minutes, by going to 100 % SOC	On/off duty cycle during balancing: > 100 %* Power of waste heat: 0.08 W Cost: 100 $	On/off duty cycle during balancing: 0.8 % Power of waste heat: 0.44 W Cost: 1000 $
E-bike 10 Ah (0.4 mA delta leakage) 10 cells in series Charged nightly SOC kept at 50 % +/- 20 % Balanced once a week for 10 minutes, by going to 100 % SOC	On/off duty cycle during balancing: 2.4 % Power of waste heat: 0.01 W Cost: 10 $	On/off duty cycle during balancing: 0.1 % Power of waste heat: 0.25 W Cost: 100 $

*) Note that when balancing On/Off ratio would have to be greater than 100 % (which is impossible), balancing must be enabled at all times, not just at the end of charge. In order to do that, the BMS must remember which cells need balancing, and how much balancing, and do balancing even when not charging.

Power wasted in heat, dissipative vs. nondissipative, for various applications (lower is better).

This is the spreadsheet used in these estimates. [xls]

Applications in which nondissipative balancing is best

Nondissipative balancing is recommended in applications in which dissipative balancing:

• If low current, it would take too long
• If high current, its heat generation could be problematic

Some BMSs use on-chip dissipative balancing, which is limited to small balancing current (on the order of 20 mA / cell), due to heat dissipation limitation of the IC. Therefore, chip manufacturers (notably Texas Instruments) are eager to propose nondissipative balancing as a way to overcome the limitations of on-chip dissipative balancing.

Other than the above applications, dissipative balancing is recommended.

Balancing algorithms

Balancing algorithms (whether nondissipative or dissipative) can be based on:

• Voltage
• Final voltage
• SOC history

This table compares the three methods.

	Voltage based	Final voltage based	SOC history based
Principle of operation	Balances whenever charging, regardless of SOC, attempting to match cell voltages	Balances at one end of charge (nearly full of nearly empty), attempting to match cell voltages	Balances all the time, attempting to match SOC, based on previous history of cells
Pros	Very simple method	The voltage at one end of charge (nearly full or nearly empty) is strongly affected by SOC, so it is more indicative of true SOC If done when nearly full, the charging current may be reduced, so varying voltage offsets due to variations in cell resistance are less of an issue; in any case, the charger is mostly off during balancing, so cell resistance is not an issue then	Can balance with less current (or in fewer cycles), because balancing can occur at any time
Cons	Using Li-Ion cell voltage as an indication of SOC is not effective because their voltage curve is quite flat at mid SOCs Plus, the voltage under load (the current tends to be high throughout the SOC range) depends of the internal cell resistance, which varies from cell to cell	Balancing at one end of the SOC is problematic because the time available for balancing (between the time the battery is mostly charged, and the source of charging is removed) is short, so the required level of the balancing current is an order of magnitude higher.	Requires more computing power, and memory to store the history of each cell
Notes		In most cases, balancing is done when the battery is nearly full, because in general batteries are regularly charged fully, but rarely are they emptied all the way	Texas Instruments is a leader in this solution

Nondissipative balancing techniques

Nondissipative balancing can be:

• Cell to battery: energy is removed from the most charged cells and sent to the entire battery
• Battery to cell: energy is removed from the entire battery and sent to the least charged cells
• Cell to cell: energy is transferred between 2 adjacent cells: from from the most charged one to the least charged one

Nondissipative balancing topologies: (L to R) Cell-to-battery; battery-to-cell; cell-to-cell.

This table compares the three methods.

	Cell to battery	Battery to cell	Cell to cell
Type of converter	Isolated DC-DC converters, low voltage to high voltage	Isolated DC-DC converters, high voltage to low voltage (or, and bulk DC-DC converter with N switched outputs)	Non-isolated DC-DC converters, low voltage to low voltage
Number of converters, for an N-cell battery	N	N	N-1
Direction and operation	Fed by a cell when it has excess charge Feeds the battery	Fed by the battery Feeds a cell when it has insufficient charge	Fed by a cell when it has higher voltage than the adjacent cell Feeds the adjacent cell
Pros	More efficient: high voltage output rectifiers Simpler: low voltage transistors, controlled from same low voltage side as the cell electronics Fastest balancing when only a few cells are low: the majority of the converters are operating	Fastest balancing when most cells are low: the majority of the converters are operating Can be implemented with a single, bulk, high power DC-DC converter, and many switched outputs to the cells	Fewer converters Highest efficiency per converter (in the 90s) Simpler, less expensive converters All DC-DC converter connections are at low voltage relative to each other
Cons	Not terribly efficient (in the 80s)	Requires high voltage transistors Requires isolated control from cell electronics (low voltage side) to drive transistors (high voltage side) Inefficient (in the 70s): low voltage rectifiers (synchronous rectifiers may help, for additional cost and complexity)	More wires A mid-pack opening (such as a fuse or a safety disconnect) blows up 2 converters Takes longer to balance, as energy has to go from converter to converter, until it reaches the destination Overall efficiency is quite poor, as losses occur at each step, from converter to converter

There is also a bidirectional topology that can switch between cell-to-battery and battery-to-cell functions. The former is used when most cells are low, the latter when most cells are high. The added cost and complexity of this topology makes this topology worthwhile only for a few applications.

Conclusions

• The power handled when maintaining a Li-Ion battery pack in balance is usually minimal: typically on the order of 0.1 to 10 W per cell in series. Therefore, in most cases, the question on whether nondissipative or dissipative balancing is better is purely academic.
• In the few applications in which the power handled is noticeable (> 10 W / cell), it is possible that dissipative balancing and nondissipative balancing generate similar amounts of waste heat (due to the stand-by power of nondissipative balancing circuits).
• In general, the significantly higher cost of nondissipative balancing trumps any energy saving arguments, especially when those savings are minimal.
• The argument in favor of nondissipative balancing comes down not to its efficiency, but to its ability to transfer a significant amount of charge quickly, in applications where only a short time is available for balancing, or frequent gross balancing is required.
• Nondissipative balancing is worthwhile alternative to dissipative balancing that uses on-chip resistors (which is limited by the IC operating temperature to only 10s of mA).
• Algorithms based on SOC history increase the effectiveness of balancing hardware (typically by a factor of 3), regardless of whether balancing is done actively or dissipatively
• For nondissipative balancing, the Cell-to-battery topology is best, as it has few disadvantages, and is most efficient, especially when only a few cells are low.