This content requires the Adobe Flash Player and a browser with JavaScript enabled. Click here to get the latest version of Adobe Flash Player.

MANAGING LITHIUM CELLS IN AN ELECTRIC-TRACTION DESIGN

By Graham Prophet, Editor -- EDN Europe, 01 Mar 2008

In 2007, independently from the design that Jim Williams details in this article, Infi neon published an application note (Reference A) describing a multi-cell-battery-chargecontrol architecture for use in traction applications. Infi neon built a demonstrator in the shape of an electric racing kart, and the design problem was to maximise the available energy from the two battery packs while protecting the individual lithium ion cells from overcharge or deep-discharge damage.

As Infi neon’s Werner Roessler, from the company’s automotive- application-support team, relates, the problem is how you handle small mismatches between the cells. To avoid damage to cells from overvoltage levels, you can simply monitor the charge level and stop charging the entire pack when the fi rst cell reaches its fully charged level. Likewise, you can defi ne the end-of-discharge point as the point at which the fi rst cell reaches its minimum allowable voltage level. However, says Roessler, this approach can reduce the effective capacity of the battery pack—relative to its nominal capacity—by as much as 10%. The simplest form of charge levelling is to monitor each cell in the battery pack, to resistively discharge any overvoltage in that cell, and to continue charging until all the cells reach their nominal fully charged voltage level. However, this technique wastes energy. The architecture that the Infi neon application note describes comprises monitoring circuitry to track the voltage of each cell in the battery pack, plus a scheme that takes energy from a cell that is charged to a higher level than the rest of the pack to feed it into the complete pack. This achieves—almost—lossless charge balancing.

Infi neon’s kart uses a battery module of 30 lithium cells: 10 series-connected sets of three paralleled cells. Each set of three operates, in effect, as a single cell. Each inter-cell connection is individually connected to the monitoring and balancing circuitry. This architecture forms a 33V pack: Five such packs are connected in series to form a 165V “pillar”, to achieve the voltage level that the kart’s motors require. There are two such pillars in a parallel connection, which the control circuit can break if a pack in either pillar reaches the fully discharged level before the other. At the heart of the balancing scheme (Figure A) lies a transformer with a single primary winding and a secondary for each three-element cell. A MOSFET switch connects the secondary winding directly across each cell: A Hall-effect sensor monitors current in each of these secondary windings and feeds a voltage level representing that current to the analogue-to-digital converter of a microcontroller. Each 33V pack has its own microcontroller, and all of the microcontrollers communicate via a CAN bus, the lowest pack in each pillar having the “master” status.

A measurement scan comprises each MOSFET switch, one after the other, closing for a brief period under control of a pulse output from the microcontroller. Due to the magnetic characteristics of the transformer, the rise of current in the winding is nearly linear, and the rate of rise is proportional to the cell voltage. Consequently, the current resulting from a known pulse yields a measurement of cell voltage. When the microcontroller fi nds a cell with a voltage higher than the others in the pack, it closes the corresponding secondary MOSFET for a longer interval, generating a higher current in the secondary—but stopping short of saturation level of the transformer’s core. At this point, the MCU opens the MOSFET switch: the stored energy in the magnetic fi eld couples to the transformer’s primary, and from there—note the polarity of the windings—returns to add to the charging of the complete pack. Similarly, the MCU can “top up” a cell whose voltage is lagging the others, by “robbing” charge from each of the other cells in turn and feeding power back to the compete cell stack. Allowing for switching and magnetic losses, the circuit achieves balancing without wasting energy. Using the transformer coupling provided Infi neon’s designers with the means of making the measurements during the scan cycle; because a current pulse in any winding couples equally to all windings in the transformer, one ADC connection suffi ces to make all the measurements. Even the small amount of energy that the monitoring cycle requires is not lost: the energy in each pulse returns to the whole pack in the same fashion, and, as each cell receives an identical measurement pulse, by the end of a cycle the pack is in the same state as it began.

REFERENCES
  1. “Battery Supervision and Cell Balancing Module for Li-Ion Packs”, Rev. 1.1 Infi neon Technologies, Application Note, 21 May 2007.


 

Our Sponsors



Ads by Google