BU-901: Fundamentals in Battery Testing

Discover what causes batteries to fail and why testing is still in its infancy.

No practical method exists to quantify all conditions of a battery in a short, comprehensive test. State-of-health (SoH) cannot be measured per se, it can only be estimated to various degrees of accuracy based on available symptoms. If the symptoms are vague or not present, a reliable measurement is not possible. When testing a battery, three SoH indicators must be evaluated:

  1. Capacity, the ability to store energy
  2. Internal resistance, the capability to deliver current, and
  3. Self-discharge, reflecting mechanical integrity and stress-related conditions

Batteries come in many conditions and a charge can easily mask a symptom allowing a weak battery to perform well. Likewise, a strong battery with low charge shares similarities with a pack that exhibits capacity loss. Battery characteristics are also swayed by a recent charge, discharge or long storage. These mood swings must be clearly identified when testing batteries.

Figure 9-1 demonstrates the usable battery capacity in volume that can be filled with a liquid, permanent capacity loss in the form of “rock content” that reduces the volume, and internal resistance in tap size symbolizing current flow

Conceptual battery Figure 1: Conceptual battery symbolizing the usable capacity, the empty portion that can be refilled, permanent capacity loss as “rock content” and the tap symbolizing power delivery as part of internal resistance. 

Courtesy of Cadex

The leading health indicator of a battery is capacity, a measurement that represents energy storage. A new battery should deliver 100 percent of the rated capacity. This means a 5Ah pack should deliver five amperes for 1 hour. If the battery quits after 30 minutes, then the capacity is only 50 percent. Capacity also supports warranty obligations with a replacement due when falling below 80 percent. Most importantly, capacity defines end of battery life.

Lead acid starts at about 85 percent and increases in capacity through use before the long and gradual decrease begins. (See BU-701: How to Prime Batteries.) Lithium-ion starts at peak and begins its decline immediately, albeit very slowly. Nickel-based batteries need priming to reach full capacity when new or after a long storage.

Manufacturers base device specifications on a new battery. This state is temporary and does not represent a battery in real-life situations because fading begins from the day it is made. The decrease in performance only becomes visible once the shine of a new device has worn off and daily routines are being taken for granted. An analogy is an aging man whose endurance begins to wear off after the most productive years (Figure 2).

Figure 2: Battery can be likened to a man growing old. Few people know when to replace a battery; some are replaced too early but most are kept too long.

Knowing when to replace a battery is a blur for many battery users. When asked, “At what capacity do you replace the battery?” most reply in confusion, “I beg your pardon?” Few are familiar with the term capacity as a measurement of runtime, and fewer know that capacity is used as a threshold for retiring batteries. In many organizations, battery problems only become apparent with increased breakdowns, which may be caused by a lack of battery maintenance.

Battery retirement depends on the application. Organizations using battery analyzers typically set the replacement threshold at 80 percent. (See BU-909: Battery Test Equipment.) Some industries can keep the battery longer than others and a toss arises between “what if” and economics. Scanning devices in warehouses may go as low as 60 percent and still provide a full day’s work. A starter battery in a car still cranks well at 40 percent, but that is cutting it thin.

Any battery-operated mission must plan for a worst-case scenario. Although manufacturers include some reserve when specifying runtime, the amount is seldom clearly defined. Critical missions demand tighter tolerances and the battery must be replaced sooner than when a sudden failure can be tolerated. (See BU-503: How to Calculate Battery Runtime)

Medical and military devices are considered critical and batteries are often replaced too soon. Rather than testing them, device manufacturers prefer to use a cycle count or a date stamp to mandate retirement. To cover all eventualities, the service duration on a date stamp is often limited to 2 or 3 years.

Medical technicians have discovered that many batteries for defibrillators have more than 90 percent capacity left when the mandatory 2-year date-stamp expires, replacing perfectly good medical batteries prematurely. In spite of this apparent waste, a US FDA survey says that “up to 50 percent of service calls in hospitals surveyed relate to battery issues.” Healthcare professionals at AAMI (Association for the Advancement of Medical Instruments) say further that “battery management emerged as a top 10 medical device challenge.”(See BU:803: Can batteries be Restored.)

Another application where battery capacity is important is in a drone. With a good battery, the device may be specified to fly for 60 minutes, but if unknown to mission control, the capacity has dropped from 100 to 75 percent, the flying time is reduced to 45 minutes. This could crash the $25,000 vehicle when required to negotiate a second landing approach. By marking the capacity on each pack as part of battery maintenance, batteries delivering close to 100 percent capacity can be assigned for long hauls while older packs may be sent for shorter errands. This allows the full use of each battery and establishes a sound retirement policy.

Many batteries and portable devices include a fuel gauge that shows the remaining energy. A full charge always shows 100 percent, whether the battery is new or faded. This creates a false sense of security by anticipating that a faded battery showing fully charge will deliver the same runtime as a new one. Batteries with fuel gauges only indicate SoC and not the capacity.

Battery failure is not only limited to portable devices. Starter batteries in vehicles have also become failure-prone. In 2008, ADAC (Allgemeiner Deutscher Automobil-Club e.V.) stated that 40 percent of all roadside automotive failures are battery-related. A 2013 ADAC report says that battery problems have quadrupled between 1996 and 2010.

ADAC, Europe’s largest automotive club, says further that each third breakdown involves either a discharged or defective battery. The report, published by German Motorwelt in May 2013, also mentions that only a few starter batteries reach the average age of five years, and this applies to all cars. The statistic was derived from the more than four million breakdowns that the ADAC car club typically receives in a year. The study only included newer cars; service-prone vehicles more than 6 years old were excluded.

BCI (Battery Council International) reports similar results. A 2010 study by the BCI technical subcommittee revealed that grid-related failures had increased by 9 percent from 5 years earlier. Experts suspect that higher electrical demands in modern vehicles lead to higher failure rates. (See BU-804: How to Prolong Lead-acid Batteries.)

Battery failure in Japan is the largest single complaint among new car owners. The average car is driven 13km (8 miles) per day and mostly in congested cities. The most common reason for battery failure is undercharge, developing sulfation. (See BU-804b: Sulfation and How to Prevent it.) Battery performance is key; problems during the warranty period are recorded as component failure and tarnish customer satisfaction.

A German manufacturer of luxury cars reported that one in two starter batteries returned under warranty had no problem. A German manufacturer of high-quality starter batteries stated that factory defects account for only 5 to 7 percent of all warranty claims. Battery failure during the warranty period is seldom a factory defect; driving habits are the main culprits. A careful assessment with advanced battery test instruments capable of looking at various failure symptoms can greatly reduce warranty claims.

The mobile phone industry experiences similar battery warranty issues. Nine out of ten batteries returned are said to have no problems. Rather than trouble-shooting a customer complaint because of lower than expected runtime, the clerk simply replaces the battery. This burdens the vendor without solving the problem; it may also lead to repeat complaints.

Dilemma of Battery Testing

Part of the problem lies in the difficulty of testing batteries, and this applies to storefronts, hospitals, combat fields and service garages. Battery rapid-test methods seem to dwell in medieval times, and this is especially evident when comparing advancements on other fronts. We don’t even have a reliable method to estimate state-of-charge, which is based mostly on voltage and coulomb counting. Assessing capacity, the leading health indicator of a battery, dwells further behind. Measuring the open circuit voltage and checking the internal resistance do not provide conclusive evidence of battery state-of-health.

The battery user may ask, “Why is the industry lagging so far behind?” The answer is simple: “Battery diagnostics are complex.” As there is no single analytical device to assess the health of a person, nor are instruments available that can quickly and reliably check the state-of-health of a battery. Like the human body, batteries can have multiple hidden deficiencies that no singular test method can identify with certainly.

A dead battery is easy to check and all testers are 100 percent accurate. The challenge comes in evaluating a battery in the 80–100 percent performance range while on duty. Regulators struggle to introduce battery test procedures. This is mostly due to the unavailability of suitable technology that can assess a battery on the fly.  The battery is labeled “uncontrollable” for good reason; this gives it immunity.

The battery world devotes much effort on the super battery, but this improved battery is incomplete without being able to check performance while in service. Improving performance and reliability does not rest in a better battery alone, but in tracking the performance as it ages.

Professor Mark Orazem compares the complexity of testing batteries with the Indian tale in which blind men touch an elephant to learn what it is (Figure 3). Because each man only feels a part of the body, disagreements arise among them. Battery testing is complex even for the sighted man but progress is being made. Better technologies will eventually immerge.

Indian tale Figure 3: Indian tale reflecting the complexity of estimating battery state-of-health.

Story of blind men trying to figure out an elephant through touch. The tale provides insight into the relativism and opaqueness of a subject matter, such as a battery.

Last updated: 2016-05-27

*** Please Read Regarding Comments ***

Comments are intended for "commenting," an open discussion amongst site visitors. Battery University monitors the comments and understands the importance of expressing perspectives and opinions in a shared forum. However, all communication must be done with the use of appropriate language and the avoidance of spam and discrimination.

If you have a suggestion or would like to report an error, please use the "contact us" form or email us at: BatteryU@cadex.com.  We like to hear from you but we cannot answer all inquiries. We recommend posting your question in the comment sections for the Battery University Group (BUG) to share.

Or Jump To A Different Article

Basics You Should Know
The Battery and You
Batteries as Power Source


On March 17, 2012 at 3:06pm
lll,,l;p;;;kkk wrote:

wot is a error to my batery

On April 4, 2012 at 2:38pm
Gary Heyer wrote:

Over my “yars”, I have found a small hygrometer (Specific Gravity tetster) to be very useful. If a battery is weak, I check each cell. My main objective is, is it just one cell, or are all uniformly down? If it’s just one, that is a sign to me that the one cell is shorted, and the batt is usually at the end of its days. If they are all down equally, then it MAY be a sign of battery age, but I would next check the vehicle’s charging system, and also whether there might be some constant discharge, like a trunk light that stays on.

On July 4, 2012 at 4:57am
Soren wrote:

@ Garyen, Your method is not useful when you are faced with RVLA batteries.

On July 4, 2012 at 4:59am
Soren wrote:

@ Gary, Your method is not useful when you are faced with RVLA batteries.

On July 11, 2012 at 1:27pm
Simon wrote:

Lots of interesting info.
Like to know if anyone has experienced an electric bicycle with a 36V - 14 amp Li+Polymer battery which only lasted 13 months, 1400 miles.
Am very disappointed at this battery performance and wonder if other e-bikers have similar

On May 15, 2014 at 5:00pm
Steve wrote:

Is it possible for a car battery to be internally damaged in a collision yet show no obvious external damage?  If so, how would you test for the damage?

On October 9, 2014 at 11:48pm
Edward wrote:

Steve, it is possible that the car battery to be internally damaged in a collision without obvious external damage. you can test the inner-resistance or capacity and voltage to clarify


On October 22, 2014 at 11:56am
eric wrote:

Hi thanks for all the great information. I have a questions for you. I have a few APV’s (advanced personal vaporizers) that use single 18650 VTC5 batteries and a few that use dual 18650 batteries in parallel, it seems that the dual 18650 APV’s last even longer then if I had used an APV that only holds one 18650 which I use till depleted then put another 18650 in it and run that one down? So, will running dual 18650 batteries in parallel will they last longer then if you used a similar device that only holds one battery which you use till its depleted then replace with a fresh one? thanks

On October 22, 2014 at 6:50pm
Edward wrote:

Dear eric, I can not understand you clearly,  but i can tell that do not use different cell in parallel connection

On March 14, 2015 at 1:02pm
joseph Alyx wrote:

Well its good to learn the different ways on this platform, pls keep it up.

On March 14, 2015 at 1:07pm
joseph Alyx wrote:

Its great learning here, pls keep it up.

On June 16, 2015 at 6:52am
Chuck McCown wrote:

I have spent the last 36 years around large stationary batteries at telephone companies.  I have seen many battery testers come and go.  None of them has ever been able to predict when a cell is starting to go bad or really give me anything useful.  What is useful is the following:  Kill the mains power and run off of battery power.  Check the voltage of each cell every few minutes.  Do this for several hours if you can and then graph it.  Now-a-days the rectifier/charger can do this for you automatically.  A bad cell will have lower voltage than the others.  I like them all to be within .05 volts of each other.  .1 worst case.  Also, the slope of the discharge voltage of the whole string can be mathematically determined, and when combined with the amps of the load, you can actually extrapolate the amp hour capacity of the string.  This Battery Discharge Testing is rock solid, proven and terribly reliable.  You can also find resistance connections this way too if you measure voltage drops across the straps.  You only have to do that one time at the start of the test.  It may be an old fashioned method but it works.

On December 6, 2015 at 5:31pm
Mike Smith wrote:

I have recently purchased a new replacement 2000mAh (7.4Wh) lithium-ion battery for my phone. The new battery appears to have less capacity than the battery I am replacing. So out of curiosity I decided to try and check the capacity of each battery. After charging each battery to full capacity 100% on my phone (4.2 volts with a volt meter) I applied a load of 60Ω and plotted the discharge curve down to 3.7 volts using a 10bit AtoD converter on a microcontroller and a low on resistance FET to terminate the discharge at 3.7 volts to prevent damage to the battery. The results were as follows. The old battery after being fully charged and then left for a couple of weeks had a capacity under the above test conditions of 765mAh. The old battery after being fully charged and then immediately tested had a capacity of 837mAh. The new battery after being fully charged and then immediately tested had a capacity of 727mAh.

My questions are when the manufactures come up with a capacity for a lithium-ion battery is the capacity quoted the capacity between the 2 voltage limits 4.2 and 3.7 if not how do they test below 3.7 volts without damaging the battery? Is the capacity predicted and not tested? Is the above test far? Can you suggest improvements? What are your feelings regarding the condition of the 2 batteries tested?

On April 24, 2016 at 12:06pm
Farouk Eshragi wrote:

If cost were not an issue, perhaps a semiconductor based long term monitoring of battery cell performance can provide a solution, at least in circumstances when the cost of the battery compared with the system it is serving in is low, and the number of battery cells are limited. Reading the data wirelessly from the device may offer a good solution for long term monitoring of data that could include full information about flow of current, the voltage to which the battery is exposed, the length of the charging and discharging cycles, peak currents, and peak temperatures, and with this information, the state of the battery can be deduced?

On December 2, 2016 at 10:50am
Gary Vreman wrote:

This was a helpful article.  I have a polar axis grid tied solar array with a 4A linear motor that draws occasionally through the day and charges a 8aH universal battery with 15W panel.  I find that the battery only lasts a year before I am fighting maintaining charge.  This may be due to sulfating so I plan on making a datalogger to quantify the battery states and see if a periodic 6A charge provide better long term capacity.  Additionally I will know if the 3 old batteries I have could be repurposed for a less critical system.  Data over time seems to be the best indicator, and like you said what level is needed.