This article was written by Bob Smith from QEFITesting to find the cyclic Life of Lithium Polymer Batteries.
This is the second part of a two-part article dealing with the cyclic life of lithium polymer batteries.
Part 1 - dealt with the general specification of LiPos and their performance, and with the testing equipment I have available to
carry out cyclic testing of them.
Part 2 - will cover the test procedure and the results of testing a series of packs.
The equipment I detailed in Part 1 operates to the following pattern. I try to start all of the tests at the same
point in the charge/discharge cycle and although this could be anywhere I try to begin with an empty pack since this is
easy to eastablish by checking the pack/cell voltage under load (nominally 3 cells at 3 volts each). The first stage in the
cycle is then a normal balanced charge using the recommended pattern of a 1C constant current up to 4.2 volts per cell
followed by constant voltage as the charge current reduces down to 10% of the 1C value. This normally takes around
70 minutes and is followed by a 10 minute pause or dwell to allow the pack to stabilise. The discharge follows at a rate
which is normally the manufacturers’ maxmum recommended value e.g. 20C, and the discharge is stopped when the
poorest cell in the pack reaches 3 volts. This is again followed by a 10 minute dwell before the whole cycle begins
again. You can see a pictorial representation of this single cycle in graph1. The equipment is designed so that this cycle
is automatically controlled and repeats as long as needed to reach the required number of cycles. The data produced by
each cycle in terms of the pack voltage, elapsed time, discharge capacity, and pack temperature is recorded by data
logger and periodically downloaded to a computer.The set-up I am using with the logger gives me around 27 hours of
recorded data with each cycle taking about 1 hour 40 mins so I get 16 or so cycles in each batch which means six days
for the 100 cycle test.
One of the problems associated with this type of long term testing is the vast amount of numerical data you
produce and I am only able to include examples of the data in the form of a typical HiBox plot and a set of
temperature/time figures. Even here you can see that the HiBox plot of only 20 cycles is crammed to the point where it
is difficult to interpret the data visually. Remember that the temperature data is a dual purpose record since it also
provides a back-up discharge capacity. Since the discharge is at constant current the product of that value in amps and
the time period of discharge in hours is the discharge capacity in Ah. In the event, the main recording system via the
HiBox proved to be sufficiently reliable that no back-up was needed.Data Interpretation.
The basic intention of the whole programme was to determine the loss of capacity suffered by the test packs
over the 100 charge/discharge cycles. Of the two traces on the plot, the blue voltage values are basically a check that
the process is proceeding as intended and that the maximum charge voltage does not exceed 4.2 volts per cell, and that
the minimum discharge voltage did not drop below 3 volts per cell. The values on the chart are for a 3 cell pack so are
12.6 and 9 volts respectively. The balancer built into the equipment did control the discharge cut-off on the basis of the
worst cell, i.e discharge was ended when the lowest voltage cell dropped to 3 volts.
The second green trace is for the discharge capacity and is accumulative, i.e. each step in the trace is the
discharge magnitude for that particular cycle in Ah but the scale reading is the total of all the discharges since the test
commenced. Over multiple cycles the vertical scale would be difficult to interpolate accurately but the HiBox software
has a floating cursor which gives the reading of any point on the trace to two decimal places (10 mAh).
Once a hundred cycle test is complete, each of the discharge capacities is measured and a graph plotted of this
capacity against cycle number. There is, of course, a fair amount of variation in these values which arises partially from
experimental deviations and partially from timing and other minor errors. In spite of this there is clearly a downward
trend in the plot which indicates a loss in capacity. If a mean line is plotted through the fluctuations (which are only
some tens of mAh in value) then, in all of my results to date, this produces a straight line with a negative (reducing)
slope. This implies that the capacity of a LiPo battery degrades with use, that the rate of degradation is constant and
cumulative, and that the quality of battery can be assessed by the magnitude of the degradation.Test Results.
The data I have gathered could be used in several different ways to measure the performance of the packs but
in this article I am only referring to the loss in capacity over 100 full charge/discharge cycles. You could start with the
manufacturer’s specified capacity but this is usually measured at a very low rate of discharge and I have tested the
packs at a much higher rate. All batteries produce a lower capacity if they are discharged under high loads so it would
be unfair to base the assessment on the capacity on the lable. The figures in the table are therefore calculated by taking
the the capacity of the mean line at 100 cycles and subracting this from the capacity of the mean line at 1 cycle and
expressing this as a percentage of latter value. In graph 2 you will see that the indicated values are 2.08 Ah at 1 cycle
and 1.92 Ah at 100 cycles giving a loss of 0.16/2.08 x 100 = 7.7%. In terms of the use of these packs it is obvious that
we want this value to be as low as possible.
The table of test results presented here has been compiled from two sets of batteries. I had already completed
the tests on several packs before I was invited to write this article, and in addition the editor contacted a number of
manufacturers/retailers to submit packs for testing specifically for the article. The combination of makes is now fairly
broad but remember that new makes of LiPo batteries seem to be hitting the UK market on what seems like a monthly
pattern, so there are still some gaps. There are also some slight anomalies with the discharge rates. Although the
majority of the packs are 2100 to 2400 mAh packs rated at 20C discharge, some are of different capacities and ratings. I
attempted to be as consistent as possible with the 20C test and used this value even if the labelled rate was higher, but it
would have been unfair to use it where a pack was rated at less than 20C and in this case the specified maximum was
used. In all cases the packs survived the test regime although some were producing very low readings by completion.Analysis.
There are several points to make about the results.
1) In terms of the cyclic life the results are probably better than I expected. All of the packs lost capacity but
some lost very little and some a significant amount. Although the losses appeared to be linear over this 100
cycle span, I suspect that they would become progressive (at an accelerating rate) if the testing had been
extended, especially with the poorer performing packs.
2) There are also large differences in the comparison between the rated capacity (the value on the label), and the
initial test capacity (cycle 1) at maximum loading. Remember that the percentage loss over 100 cycles is based
on the latter value rather than the former, but it is equally important to know how much you will get from a
pack under load when it is new. It is clearly not always possible to rely on the value on the label.
3) There is absolutely no indication that LiPo packs should be bedded in by slow charge/discharge cycles when
first used (as we were advised to do with NiCd and NiMH packs).
4) I hope that the data provided in this article is of use to the modellers reading it, but I have to add a word of
caution. I believe my equipment and procedures are sound and reliable, but they are not up to the standards of
a commercial testing laboratory in areas such as calibration. In addition to this, I am reporting on the results of
testing a single sample of each pack type and statistically this is obviously not ideal. I am repeating the results
I obtained from my tests and I hope you have found it interesting.
OverTec, Jesmond Dene Trading Estate, Forton, Nr Lancaster, Lancs PR3 0AT – Tel 01524 793328 websitehttp://www.overlander.co.uk
West London Models, 214 High St, Harlington, Middlesex, UB3 5DS – Tel 020 8897 2326
Puffin Models, Unit D3 Backfield Farm, Wotton Rd, Iron Acton, Bristol. BS37 9HD – Tel 01454 228184
Ripmax Ltd., Ripmax Corner, 241 Green Street, Enfield, EN3 7SJ
Tel: 020 8282 7500 Fax: 020 8282 7501 Website http://www.ripmax.com
BRC Hobbies, P O Box 226, Whickham, Newcastle upon Tyne, NE16 4WU
Tel 0191 4887879, website -www.brchobbies.com
Electrolite RC Ltd., 6 Westmoreland St. Harrogate, N.Yorks, HG1 5AT email email@example.com
ModelPower, 3 Church Walk, Mancetter, Atherstone, Warwickshire, CV9 1PZ
Telephone: 01827 711501 email firstname.lastname@example.org
J Perkins Distribution Ltd., Northdown Business Park, Ashford Road, Lenham, Kent ME17 2DL
Telephone: 01622 854 300 email email@example.com
RCM Direct, 9 Kenilworth Close, Hemel Hempstead, Hertfordshire, HP2 4EY
Telephone: 0845 205 5050 email websitehttp://www.rcmdirect.co.uk
Graupner GmbH & Co. KG, Postfach 12 42, D-73220 Kirchheim/Teck, GERMANY.
Website http://www.graupner.comPhotographs etc.
Page 1 - An assortment of the packs tested.
Page 2 - Graph 1
Page 3 - Graph 2
Page 4 - Typical Temperature/Time data record
Page 5 - Typical HiBox plot of cyclic voltage and capacity.