Battery Testing Procedures

Introductions:

The testing procedures used in all of the portable battery reviews are outlined here. It is very important that the procedures are published openly so that any and all test results can be reproduced by others. In addition if professionals are able to identify flaws with the procedures please bring them to our attention so that adjustments can be made. If a critical flaw is found then we will retest the in-scope batteries and update any related review articles as required to ensure our data is completely accurate.

The following documentation cites multiples pieces of equipment used in the testing of the Portable LiPo Power Bank batteries (AA & AAA testing does not follow this procedure), for more information on the noted equipment please see the Battery Testing Equipment page.

 

Background on Portable Power Bank style batteries

Power bank style batteries are testing multiple times using different level of loads. These types of batteries are designed to be used in a very different way then many common LiPo batteries such as those found in and used for RC Vehicles. In an RC Vehicle, the load on a battery can range drastically from 0.5A up t0 6 or 8A. The result is that a a 6000 mAh battery could be drained in 30-60 minutes depending on the load. The capacity of a battery, such as 6000 mAh is determined by the amount of load that would be needed to drain the battery in 1 hour. Now most common Lipo batteries get this rating based on the discharge of not 1 hour but 20 hours or 20C. This gives us the starting point of where the testing begins. In general all batteries are tested using the calculated current for 20C and 10C in some cases future testing will include 5C. These numbers are used where they best represent a real life load that would be put on them by a USB device at 5V. In an RC Car, you could see the electronics pulling a load of 10A or 10A. but you will never see this from a cell phone or tablet, meaning that trying to place this load on these style batteries does not reflect real life use. The values that are used in our tests, documented in each review, will represent the most common values that would be found in consumer electronics that would use portable LiPo batteries. For USB based devices tests will be done between 500mA up to around ~2200 mA. Other batteries that are designed for different use cases will be tested with loads that most fit their typical use case in real world use. Any notable variance from these procedures will be documented in the reviews as applicable.

 

Procedures Updated August-31 2014

Power Bank Testing Procedures V2

All tests results documents in reviews that used this new procedure version will be noted as such.

At this point you know why things the way we are, now describe the the how. Using the equipment outlined we take each battery pack and run through 4 or 5 tests at each of the selected currents, for smaller batteries this may be .5A and 1A, for Larger batteries this may be .5A, 1A, 1.5A, 2A, 2.1A,2.4A and so on. If, after 4 tests at eat current, the results are consistent a 5th test is not run. Lets use a 10,000 mAh battery with a max output of 2.1A as our example going forward. The first set would be at .5A, then 1A and 2.1A just as an example. The test procedures for each set are then carried out as follows:

  1. Take the Re:Load load generator, connected to the USB Tester and plug it into a setup battery. This is not the battery being tested.
  2. Dial in the the load on the setup battery until the USB Tester reads the needed value, in this case 500mA.
  3. Unplug the setup battery, plug in the battery being tested, and document the start time.
  4. wait until the battery is drained and read the total mAh output recorded by the USB Tester
  5. Document the results

In version 1 of the procedures i had to use a timelapse to properly capture the duration of the test and use the duration and known current to calculate the batteries output. After further discussion with the designer of the USB Tester a feature previous not known to me now allowed me to use the USB tester to capture the battery output on the device and have a second power source to power the OLED Backpack and tracking algorithm. The new feature allowed me to separate the power sources so the power source for the OLED Backpack was not taken into account or passed through to the dummy load generator. The meant the USB Tester started at 0 until my test battery was plugged in, then it would start tracking until the test battery was dead. Allowing me ti just read the total output from the display instead of calculating based on time.  This allowed for a much more accurate measurement of the output along, less steps for each test and less failed tests such as not properly recording the start time or the timelapse recording failing causing me to loose the whole test.

Known Variances/margins for error

Based on the procedures above, the USB tester is able to capture the exact amount of current that travels through the device. Even if the load varies by a bit above or below the set value, which is known to happen based on the device and environment, the variance is still now fully accounted for. Previously in the the V1 procedure i had to constantly check to ensure the current draw was consistent or the results would be skewed. preliminary testing shows that the results are not noticeably different meaning the previous testing procedures were not bad. This new procedure is just easier and more accurate. Since there is no longer a time lapse delay and no longer a load variance to account for we can consider the test results under this procedure to be more accurate. The USBTester device itself has a PC Application, i will be looking at this application further to see if i can use it to collect logs that can be shared as part of the data published n the reviews.

Test Duration: No variance or margins to account for.
Test load: No variance or margins to account for.
USBTester: While i don’t currently have documentation on known resolution variances for the tester my own testing has shown there to be little to no variance when compared to multiple Digital multi-meters. Since the aggregation and totals are done in software it is possible there is some very small variance, but for the purposes of these tests i would now consider this negligible

 

Testing Procedures Version 1 - No Longer used

Previous documentation kept for historical reference only.

Power Bank Testing Procedures (v1 – No longer used)

All tests results documents in reviews that don’t note V2 or later were done using this procedure

At this point you know why things the way we are, now lets wrap up the how. Using the equipment outlined we take each battery pack and run through 4 or 5 tests at 20C and then 4 or 5 tets at 10C. If, after 4 tests, the results are very consistent a 5th test is not run. Lets use a 10,000 mAh battery as our example going forward. The 20C value for this battery would be 500mA (10,000 / 20). The test procedures are then carried out as follows:

  1. Take the Re:Load load generator, connected to the USB Tester and plug it into a setup battery. This is not the battery being tested.
  2. Dial in the the load on the setup battery until the USB Tester reads the needed value, in this case 500mA.
  3. Unplug the setup battery, plug in the battery being tested, and document the start time.
  4. Start the timeplase recording of the USB tester
  5. Document the results

Now I will elaborate on steps 4 and 5. Using a Raspberry Pi with a Pi Camera Module, it is setup to take a a photo every 2 minutes of the USB testers (and the rest of the hardware) throughout the test. Since these tests take many hours they are commonly done over night or otherwise unattended. When when the batteries are all dead, we then review the photos and fine when each tester went dark. Checking the time stamp of that specific photo we can compare it to the starting timestamp and we now have the duration of the test.

Since we know the start and stop time we have the duration, and since we know the exact load place on the battery for that duration we can calculate the effective capacity. In this case lets say the test took 11 hours, of the expect 20 hours. We now simply divide the expected time by the actual time and we will find that in this test the battery put out 55% of the expected power. We can also calculate that this equals 5,500 mAh (11 hours at 500mA).  You may think this sounds like a crazy example but as you review the batteries tested you will find that for some, 55% efficiency is actually realistic, but that is a topic for another page/article.

Known Variances/margins for error Based on the procedures above, we are documenting the following variances or margins for error in the testing.

Test Duration: Since the camera only takes a picture ever 120 seconds, we will assume a maximum margin for error on each test of 2 minutes or 120 seconds.

Test load: Due to variations in the batteries, especially at higher loads, along with the difficulty in precision with the dial on the Re:Load 2 load generator, we are documenting a margin of error on the load of 10mA. Unofficially, margin in most tests are within 0-3mA of the target load. However, as noted we will officially state that our tets have a margin of error of up to 10mA