I was using a Lenovo ThinkPad Edge E430 laptop. I wondered how long the charge on its battery would last, once I had taken efforts to minimize its power usage. This post describes the steps I took to test the battery’s endurance.
Assembling the Tools
A search took me to two kinds of information. First, there was the category of webpages describing tests with one or more manual elements. For instance, Radio Shack and CNET listed several different steps or approaches that, in their view, would provide the best picture on the battery’s condition. In similar spirit, Rob Pegoraro offered a way to use a Windows log file to determine when the juice ran out. (Basically, Start > Run > eventvwr.msc > Ctrl-F > Sleep Reason: Battery. In other words, look for the time of the event indicating that the laptop went to sleep because it had reached the low power setting configured in Control Panel > Power Options.) This sort of approach might be handy, if slow, for manually testing and retesting the impact of various settings (e.g., screen brightness). Muhammedlilg suggested using an administrator-level command:
powercfg -energy -output C:\report.html
This would produce the Report.html file. That file might contain tips on improving the laptop’s power configuration; and under “Battery Information,” it would report the contrast between the battery’s Design Capacity and its Last Full Charge. In his example, Design Capacity was 5200 but Last Full Charge was only 426. The battery producing that 426, he said, was only storing a small fraction of its original capacity, and thus would power the laptop for only a few minutes before running down.
I was interested in some of those manual approaches. I was also interested in the second category of webpage that came up in response to my search. It seemed that there were a number of battery-testing utilities out there. The names I encountered were PowerMark, Battery Eater, BatteryBar, and MobileMark. PowerMark’s least expensive version cost $200. Battery Eater was free. BatteryBar offered a free Basic version. MobileMark 2012 cost $750. CNET had high reviews for BatteryBar but essentially nothing for Battery Eater. On the other hand, SoftPedia said that the BatteryBar version actually cost $10, and had user ratings averaging 3.4 stars, whereas Battery Eater was listed as free with 3.7 stars. There did not seem to be obvious recent indications that one was superior.
I browsed reviews of Battery Eater at Lifehacker, TechSpot, and MakeUseOf. MakeUseOf said that Battery Eater offered three tests — maximum load (which would probably not reflect my own actual usage), reader’s test, and idle test — and recommended trying all three. I browsed reviews of BatteryBar at PCWorld and CNET. These indicated that BatteryBar was more of a monitoring program, generating information about the battery’s current condition and estimates of its remaining time. Apparently the estimates were smart in the sense of becoming more accurate as the program became more familiar with the particular machine. As such, BatteryBar brought to mind the distinct category of battery monitors and optimizers (e.g., BatteryCare). The smart feature would evidently eliminate the need to manually recalibrate the laptop’s battery monitor so that its reports of time remaining would somewhat match the battery’s actual charge: apparently the software would figure that out itself, and adjust its reports accordingly (though presumably a miscalibration could affect the battery life optimization issues discussed in another post). Anyway, there did not appear to be a reason to use just one or the other, so I downloaded and installed both Battery Eater (as a portable) and BatteryBar.
Using the Tools
It seemed that I should begin by running the three tests offered in Battery Eater, simultaneously eyeballing what I was getting from BatteryBar. Then I could decide whether there was much to be gained from the other measures described above.
At the start, BatteryBar’s taskbar icon gave me information about the current situation, with the laptop still connected to its AC adapter. It said I had a 96.9% charge, which was consistent with my Lenovo ThinkPad’s battery preservation software: it prevented the battery from being 100% charged, so as to help the battery avoid damage from overcharging. BatteryBar also said that the battery’s capacity was 41,120 mWh, out of 42,440 possible, and that battery wear was at 0.8%. As for the estimated runtime on that charge, BatteryBar said it was still calculating. I gave it a few minutes. It didn’t change. I decided to move ahead and see if maybe it needed some discharging to do that calculation. (Incidentally, BatteryBar’s taskbar icon showed “N/A” when installed on a desktop; and when I moused over that icon, it said, “No batteries detected.”)
I started Battery Eater. I set it to “Begin test when disconnecting AC.” I went into Options and looked at its Benchmark Mode. It had apparently changed since the review mentioned above. It now had four test modes: Classic, Reader’s Test, Idle, and Plugin Bench. I guessed that Plugin Bench was its maximum test mode. I chose that. Then I unplugged the AC power cable. The screen immediately went very dim, consistent with the Maximum Battery Life power plan that I had selected (Control Panel > Power). BatteryBar initially reported an estimated runtime of only 2:59 (i.e., two hours, 59 minutes); but within a few minutes it had revised that to 3:23, and that value climbed gradually after that. This was also true of the “Total” time reported by Battery Eater, but the change was more dramatic there, starting from a very low initial estimate of only 1:25. (For some reason, the value reported on BatteryBar’s taskbar icon did not match the value reported on mouseover.)
After a few minutes, the laptop’s screen went dark. This was, again, consistent with the Maximum Battery Life power plan, but I did not see how Battery Eater was going to test maximum drain (i.e., constant, heavy use) without keeping the screen lit. Eventually the hard drive powered down too. It seemed that I should have altered the power plan to keep the screen on and the hard drive running, to get a more realistic impression from Battery Eater and BatteryBar. The two programs were highly consistent with one another, throughout my test, in their reports of battery percentage remaining. For instance, a half-hour into the test, both said that I had 85.1% left. By that point, Battery Eater had come up to the point of estimating total life of 3:17, while BatteryBar had moved up to 4:02. A while later, I noticed that the laptop had hibernated. This was a ridiculous test. It seemed that the Plugin Bench was not the test I wanted.
I recharged the laptop and tried again with Battery Eater’s Classic test. This time, I also set the laptop to the High Performance power plan. This time, we really chewed through battery power. The laptop’s charge dropped 3% in two minutes. I wasn’t sure if that was because of the test or the power plan. I recharged the battery and changed back to the Maximum Battery Life plan, with some modifications. But then it turned out that the Classic test likewise failed to prevent the display from turning off or the laptop from hibernating. Finally, essentially giving up on Battery Eater, I arranged a custom power plan that would keep the hard drive spinning, would keep the display on at a low power state, and would prevent the machine from sleeping or hibernating.
When the laptop was fully charged, I pulled out the power plug once again. I was not watching, but at some point several hours later, the laptop shut down. I powered it back up to take a look. The screen said, “0190 Critical Low Battery Error.” Then the machine shut down. So, OK, I would have to plug in the cord to see what happened. So now, it was recharging, but it looked like it had been down around 6% of capacity. BatteryBar was now estimating a runtime of 3:35, so apparently that’s about how long the battery would keep the machine going with its screen on and its hard drive running.
I had installed Battery Eater as a portable, so I went into its program folder (which, like all portables, I kept in my customized Start Menu). In that folder, as someone had suggested, I found a subfolder whose name vaguely resembled the date on which I had run Battery Eater. In that subfolder, I saw a file called Discharge.beg. I screwed around with attempts to do as someone had advised, and drag it onto BEPro.exe (either its running window or as a file displayed in Windows Explorer), and to run a Report from within BEPro. These efforts succeeded. Something I did — the step of running a Report, I think — gave me an HTML report that opened in my web browser (i.e., Firefox). It contained several items of extraneous information and nothing really straightforward. It said my “Results” were a total time of 54:42:22. In other words, it had been almost 55 hours since I had started the test, or perhaps since I had started Battery Eater. Who cares? How long did the computer run before crashing, during the actual test, after I pulled out the power cord? Well, the graph indicated that it had been “3282 min.” A bit of division suggested that this was more or less equal to 55 hours. There were no X axis labels on the graph, so I had to do manual counting to figure out that each line on the grid equaled two hours. It didn’t come out quite right, but maybe I miscounted. So, OK, by that calculation, now I could see that the laptop had run for about 11 hours, and then had started to lose its power over the next several hours. In other words, the graph appeared to be presenting gibberish. Battery Eater did not appear to be telling me anything relevant. (Incidentally, throughout its operation, Battery Eater gave me “Access violation” errors. I don’t know why.)
While in battery mode, I tried the POWERCFG command shown above. It said, “Enabling tracing for 60 seconds. Observing system behavior.” I didn’t use the machine during those 60 seconds. It said, “Energy efficiency problems were found.” I looked at the resulting report.html file (see above). Some of the problems it found were the things that I had turned off for my test: no sleep timeout, no display dimming or timeout, no disk idle, no USB suspend, high CPU utilization. I was impressed, at this moment, that I had gotten 3.5 hours out of that battery with all this stuff running. It seemed, though, that my test had consisted simply of changing the power settings thus, with no input or activity by Battery Eater. So my battery would have run down even faster if I had been actively using Word or some other program, and it would have run down more slowly if I had left those power settings in a more reasonable state (so as to dim the display etc. when not in use).
I went back into the Power options and reset that power plan to its default settings, so that these timeouts etc. would function as they were supposed to. Then I continued in the report. It complained that several running programs involved “significant processor utilization”: Word, as expected, and also a system monitor that I could have shut off in the system tray. Odd thing, in report.html: it said my last full charge was 47960, significantly exceeding the battery’s design capacity of 42770. Was my Lenovo-supplied power brick overcharging my battery?
Finally, in terms of using the tools suggested above, I could have used the Event Viewer approach advised by Rob Pegoraro, but I hadn’t noted the time when I unplugged the power cord, and I wasn’t sure the report would show me when I had done that. Besides, I was running out of juice for this inquiry.
My investigation suggested that BatteryBar would provide a fair sense of how long my laptop would run on its battery. I did not test it through multiple rundowns and recharges, but it seemed stable and functional, lending credence to the claim that it would become more educated over time and would thus render steadily more reliable estimates.
To test my battery, the best way would have been to sit there and just use it in a normal way. I did not want to do that, so I set up a power plan that kept lots of things turned on (e.g., display, hard drive). Then I just let it run itself dry and shut down. When I came back and replugged the power, BatteryBar seemed to indicate that I could expect about 3.5 hours of battery life at that rate.
Finally, I used a POWERCFG command to render a report on things that could be changed to improve duration of the battery’s charge. Taken together, these steps gave me a rough impression that I might typically get at least 3.5 hours of use out of a charge on that battery, with a prospect of significant upside (up to perhaps five hours) with better power option settings.