Sunday, October 27, 2019

Shunt regulation of series-connected lead-acid batteries to equalize the voltage

There are many instances where series lead-acid batteries are connected - including:
  • 12 volt systems using two 6 volt "golf cart" batteries in series
  • 24 or 48 volts in a UPS (Uninterruptible Power Supply) consisting of several 6 or 12 volt batteries
  • A UPS system that requires much more than 48 volts - more on this, later.
One issue that arises with any system in which cells/batteries are series-connected is that the voltages across them will ultimately be unequal.  For example, in a hypothetical 24 volt system where the nominal float voltage would be about 27.1 volts it is likely that the two "12 volt" batteries will be slightly different.  If the voltages are "pretty close" (within 0.1 volt or so) there is probably on reason to be worried - but as batteries age, they will inevitably drift apart:  Before you know it you might end up with one battery at, say, 14.2 volts where it will be evolving gas (or if it is a "sealed" battery like an AGM, will start to lose electrolyte) and the other battery will then be around 12.9 volts where it will be chronically under-charged, leading to loss of system capacity.

In either case - if the battery is exposed to a consistently high or low voltage - the life of the battery will be reduced - possibly dramatically!

Figure 1:
A small pile consisting of four shunt regulators.
For convenience, these are outfitted with spring-loaded
alligator clips which allows easy installation and removal -
and eliminates the need to sandwich a ring lug on the
battery terminal and potentially increase the total series
resistance of the bank by doing so.
Click on the image for a larger version.
While there are certainly other considerations, keeping the series-connected batterys' voltages in check will certainly go a ways toward maximizing system longevity.

First, a couple words of warning:
  • Batteries are high-current devices:  Shorting can cause injury/fire, so be very careful!  Remove metal jewelry - particularly any rings, or at least cover them with tape.
  • Some types of batteries - such as "flooded-cell" lead acid - can weep a bit of sulfuric acid which can cause burns on skin and damage to clothing.
  • Some UPS systems are not isolated from the mains power and pose an electrical shock hazard:  Always assume this to be the case and take precautions (e.g. power down/disconnect.)
  • Even moderate/low voltages can cause electrical shock.
  • While this information is presented here in good faith, it is up to you to do research about its validity before implement it, taking responsibility for doing so.
  • It is up to you to research and implement safe procedures when working on this - or any - electrical gear and you are solely responsible for any damage/injury that may result.

"How do you know this?"

A quick check of manufacturers' specifications and recommendations will reveal that exposing batteries to either too-high or too-low voltage will compromise longevity, so it would seem to be "pound foolish" to ignore it happening.

As a "case study" I maintained, for several decades, several multi-battery UPS systems - the largest of which was a 300kVa system considering of two banks of forty 12 volt batteries in series that were wired in parallel (e.g. 542 volts nominal) - and with a total of eighty batteries, things are going to drift around!

Having an available BMS (Battery Monitoring System) I could easily track the voltages on the batteries - and inevitably, they started to drift apart as the batteries aged, were exposed to slightly different ambient temperatures (e.g. the ones higher up will be a degree or so warmer the the lower ones) and manufacturing variances.  Knowing full-well the implications of batteries that were drifting apart, I soon devised a simple shunt regulator, described on this page:
These "shunt" regulators are very simple:  Just several diodes (including a Zener) plus an LED that allows a selectable amount of "leakage" current so that "better" batteries (e.g. those with lower self-discharge/leakage current) aren't exposed to elevated voltages by the current of the "other" batteries.  By "tweaking" things one can balance the system so that all batteries will obtain very close to equal voltages.

Despite the fact that only the current through a few lowly LEDs (20-30 milliamps, maximum) is microscopic compared to the currents seen when the UPS was "on battery" and fully-loaded (several hundred amps) this system worked well for several years - but it required frequent adjustment as the batteries aged and their own internal leakage currents started to change:  The battery monitoring system's "live" voltage readout was an invaluable aid to allow tweaking - but this task eventually got "old".

A better shunt regulator:

The "LED+Zener+diode" arrangement has a sharper voltage-current "knee" than a simple resistor, but it was not quite "sharp enough" so I decided to upgrade to a circuit that would respond much more forcefully with increased voltage - and I chose to use the venerable TL431 "programmable Zener".  This chip is ubiquitous, appearing in almost every PC power supply ever made:  It has a temperature-stable onboard voltage reference, and it can handle up to 100 milliamps - several times the current of the original circuit.

This circuit is represented in Figure 2, below:
Figure 2:
Schematic and mechanical layout of the shunt regulators depicted in Figures 1 and 3.
Click on the image for a larger version.

This circuit is very simple in its operation:
  • U1, the TL431 will turn "on" if its reference voltage exceeds 2.5 volts, drawing current.
  • R4, the potentiometer, is adjusted so that the "reference" terminal is 2.5 volts at the desired shunt voltage.  A 10 turn potentiometer is strongly recommended as the setting of the precise voltage will be both "fiddly" and easily disturbed if a single-turn pot is used.
  • When the battery voltage is below the adjusted shunt voltage, U1 is "off", the circuit drawing a few hundred microamps.  This is likely to be less than the self-discharge current of battery itself.
  • When the battery voltage is above the adjusted shunt voltage, U1 will turn on:  The LED will illuminate - the brightness roughly proportional to the shunt current - and the bulk of the current will flow through R2.
  • Because the voltage will be high enough to activate the shunt only when the battery bank is being charged, these shunts will have negligible load when the bank is actually being used (e.g. power being drawn due to a power failure.)
  • R2 (and R1/LED1) limit the maximum current that is likely to be drawn by the circuit in the event that voltage cannot be drawn down below the threshold voltage.  This can occur during bulk charging of the battery banks and it can also occur if the sum of the threshold voltages of the individual shunt regulators is lower than the float voltage - something that could happen on a system that adjusts the float voltage with temperature (discussed below).

In practice, if we were to set the shunt voltage is set to 13.55 volts, the lead resistance connecting the circuit to the battery will result in a very sharp "knee", the circuit going from "off", drawing a few hundred microamps, to "on" and drawing nearly 100 milliamps - in just a few millivolts or 10s of millivolts  - depending on the gauge of the wire used to connect to the battery.
Figure 3:
Internal and external views of the shunt regulator.
These were built "dead bug" rather than on a circuit
board, using sleeving to insulate conductors that might
otherwise touch.  Braided silicone-fiberglass tubing
was used to insulate the large resistor as it is designed
to handle a bit of heat with the remainder of the circuit -
and the top of the braided tubing - being secured with
heat-shrinkable tubing:  The end of the LED and
potentiometer protrude from the end of the tubing.
Click on the above for a larger version.

In operation the it is suggested that the voltage threshold be adjusted to just light the LED at the ideal float voltage in order to force enough current through the charging system to assure that a small amount of current is flowing through each battery - or through the shunt regulator.  In systems that keep the batteries maintained at a constant temperature the float voltage will remain constant, but in some cases - where the batteries are in an uncontrolled climate environment - the temperature and the float voltage may vary - typically being slightly increased, by many "smart chargers", at low temperatures and decreased at high temperatures.

The information in Figure 2 suggests a voltage that works out to be about 2.26 volts/cell, which is a reasonable value for a temperature range between 10C and 35C (50F-95F).  If the charge voltage rises above this value, the shunt regulators will start to conduct - but the voltages across the battery will be equalized, provided that they were adjusted to the same voltage:  The heat produced - even though it may be just a few watts - will not be of detriment in extremely cold conditions to the performance of the battery.  At high temperatures the lower voltage being produced by the charger may not "trigger" the shunt regulator at the average voltage, but it will still keep any errant cells from straying too far in voltage from the ideal:  Because self-discharge and leakage currents of batteries increases at higher temperatures, it is arguably more important that measures to be taken to keep everything equal!

The diagram in Figure 2 shows two options -  Values for a circuit to be used with a "12 volt" battery and values for a "6 volt" battery - but the operation and set-up is identical in each case as described on the drawing itself:
  • Set the wiper of R4, the 10 turn potentiometer, the wiper is at the "ground" end.
  • Set an adjustable power supply for precisely the desired shunt voltage.  For normal "room-like" temperatures with Lead Acid batteries, 13.55 and 6.775 volts is recommended for 12 and 6 volt types, respectively.
  • Connect the unit to the power supply and adjust the potentiometer so that the LED just illuminates.   Note that even a few millivolts will make a significant difference in LED brightness which means that one adjusts several in on session and if they are "approximately" the same brightness, they will be really close to each other in threshold voltage.
  • It is strongly recommended that all units in a particular battery bank be adjusted to the same voltage.

It was convenient, at the time, to construct these circuits in a "dead bug" manner (see Figure 3) with no circuit board:  Once the layout was determined - and thoroughly documented - it took only a few minutes to assemble each unit, trimming/bending/insulating/soldering the leads in assembly-line fashion.

The entire circuit was covered with insulating tubing - but to cover the main heat-generating component, R2, I obtained some high-temperature, silicone-fiberglass tubing.  This tubing extends beyond R2 and a small piece of "normal" heat-shrink tubing is used to cover these components and hold everything in place with the tip of the LED and the adjustment screw of R4 protruding.

Because of the heat being produced - which could be well over 1 watt - R2 should be placed as far away from other components - particularly U1, which also generates heat.  If you choose to replicate this circuit on a small board it is strongly suggested that R2 and U1 be separated - and that R4, the potentiometer, not be placed too-near R2, either.

The leads connecting the unit are color-coded for polarity and in this case, they were fitted with alligator clips which provide a convenient connection to the battery terminals.

Long-term observation:

During the time that the UPS was active, batteries were replaced only when they degraded as indicated by the resistance measurements of the battery monitoring system:  By the time the UPS was finally shut down after 22 years of operation some of the batteries were "new" and some were as old as 13 years and still within their specifications of internal resistance (and when they did finally go out of spec in terms of internal resistance, many were load-tested and their remaining amp-hour capacity was still pretty good) - and much of this is attributable to the fact that these shunt devices did a very good job of confining the "float" voltages of all 80 batteries to within +/- 50 millivolts - most of that variation being due to not all shunt regulator units being more-precisely adjusted than that.

To be sure, the battery monitoring system did do in-situ impedance testing and a battery was pulled and replaced when its resistance exceeded a threshold determined by observation and correlation of the "failed" battery with its actual amp-hour rating measured after the fact:  By the time the internal resistance of the battery exceeded its mark (0.005 ohms for the particular 100 amp-hour, high-current UPS batteries that were used) it had dropped below about 80 amp-hours as measured at the 20 amp rate.

This long-term observation also showed that the LEDs were a useful visual indicator:  If an LED wasn't illuminated at least dimly it meant that the particular battery's leakage current had exceeded the average of idle current (shunt regulator current plus the battery leakage current) and that its terminal voltage was dropping - something that was usually a sign that that particular battery should be watched very closely.

What about an "Equalization charge"?

It is recommended by many battery manufacturers that an "equalization charge" be applied to the batteries periodically  to raise their terminal voltage, presumably "stirring up" the internal electrolyte of flooded-cell batteries.  In such situations, the shunt regulator will try to clamp the voltage, but since it is current-limited, the batteries will still see elevated voltage:  The shunt regulator should help divide the voltage of series-connected batteries to assure that this purposely-high potential will be the same across all batteries.  (Note:  One should not equalize AGM batteries as this can lead to internal gas pressure that can be vented and cause loss of electrolyte.)


You might ask:  "Don't I need to equalize the voltage of the individual 2 volt cells within a 6 or 12 volt battery for this to work?"  The answer is:  It would be nice if this was possible, but connections to individual cells is usually no possible.  Fortunately, in a single battery, individual cells are usually pretty-well matched as they were made at the same time and typically experience (pretty much) the same temperature throughout their lives.

If you have a system that uses individual 2-volt cells, the above circuit will not work at that low voltage - but there is a version of the TL431 (e.g. the TLVH431) that will work at the "2 volts" of individual cells.  Its maximum current rating is lower than the TL431, but it should be adequate for many applications.  The circuit in Figure 2 would have to be modified slightly to accommodate these changes (e.g. adjust values of R2 and R3 - R1 and the LED would be eliminated as the voltage would probably be marginal/too low for it to work) - but that might be the subject of another article.

* * *

Another related page at

This page describes a circuit of similar function, also based on the TL431, that is used for equalizing LiFEPO4 cells.  With component changes to adjust the voltage threshold, it could be modified for 6 or 12 volt batteries.  Because it uses power transistors, it can handle much more shunt current - but note that even with the large 300 kVA UPS, 50-80 milliamps, the current capability of the TL431 itself, was sufficient to keep relatively healthy batteries equalized.

One saving grace is that like most power systems, the UPS was rarely "on battery" which meant that even at just 50 milliamps or so, even an imbalance of a few amp-hours would eventually be equalized - assuming that the batteries themselves were run "completely down".  With a low-voltage disconnect built into a system that cut off the load at the 25%-30% level, this should never happen, so a bank that consists of multiple batteries - even those with slightly different amp/hour capacities (due to manufacturing differences, age) should "track" reasonably well upon recharge.


This page stolen from