13 October 2003 (revised 19 December 2003)
I recently decided to switch most of my electronic gadgets that take AA batteries from disposable/primary (alkaline) cells to rechargeable/secondary (NiMH) cells. I have a lot of gadgets, so that means I need a lot of these batteries and fair number of chargers for them. NiMH battery technology is improving at a fairly rapid pace, and I expect to able to upgrade and replace the cells every one to three years. The chargers, on the other hand, are a longer-term investment, and charger performance is said to have a significant impact on battery performance. Fast NiMH chargers have to be somewhat sophisticated, with circuitry that detects cell condition and adjusts the charge rate accordingly. Poorly engineered chargers can damage batteries, shorten their lifetime, and/or leave them incompletely charged.
There's a fair amount of information (on the 'net and elsewhere) compairing the performance of various commercially available NiMH cells (most notably Dave Etchell's excellent "Great Battery Shootout" page). Unfortunately, for all the information available about cells, there's relatively little data comparing different charger products for them. So I've been measuring various properties of commercially available AA cell chargers. NiMH chargers are said to vary widely in how quickly and completely they charge and in how much they heat the cells (a rough indicator of the damage done by each charge cycle), so charger selection can have as much of an impact on performace as cell selection. This page summarizes my preliminary experiments. Note that all tests were done under somewhat uncontrolled (and definitely not laboratory-grade) conditions, and should not be regarded as authoritative or conclusive. In particular, all measurements should be considered relative rather than absolute.
So far I've tested seven different charger configurations. Five are "desk" charger models (which use external "wall wart" power supplies): the Lenmar MACH1-Gamma, the Maha C401FS (in "slow" mode), the Maha C401FS (in "fast" mode), the Maha C204F, and the Rayovac PS4 "1 hour" charger. The two others are lightweight "wall/travel" chargers with built-in AC power supplies: the RipVan100 Lightning Pack 4000N and the Sony BCG34. All can accept up to four AA or AAA cells. The Lenmar MACH1-Gamma, the Maha C401FS, and the Rayovac PS4 have four independent single cell charging "channels" (and can charge one, two, three, or four cells at a time); the other chargers have two (two cell) channels (and can charge either two or four cells at a time). The Rayovac can also charge a single 9V cell.
All batteries used in the tests were factory-fresh Maha Powerex 2200mAh NiMH cells purchased (in a single order) from Thomas Distributing. (The Powerex 2200 was, at the time of these tests, the highest capacity battery rated on the "Great Battery Shootout" page). It's entirely possible that these results would be radically different with different batteries, and chargers that perform well on some cell models could well do poorly with others. As time and resources permit, I may run tests with other high capacity NiMH cells in the future.
For most users and applications, the three most important questions about a battery charger are how completely it charges the cells, how long it takes to finish a charge, and how much damage it does in the process (the extent to which each charge reduces the useful lifetime of the cells). As is frequently the case in engineering, meaningful answers to these questions are suprisingly complex and involve tradeoffs and compromises.
How completely a charger fills batteries depends upon several factors, including the capacity, age, and state of charge already in the cells when the charge starts and the length of time the charger is left connected after the charge is "complete". Even when these factors are kept equal, the total amount of energy that can be extracted from a given fully charged cell depends to some extent on the manner in which it is discharged. Rechargeable cells are typically rated in milliamp-hours, mAh. A fully charged 1000mAh cell can in principle deliver 1000mA for one hour, 100mA for ten hours, or 10mA for 100 hours, and so on, but the reality is not quite so linear. Manufacturers of rechargeable batteries usually measure their products' mAh ratings using a load that discharges the battery over 10 hours (this discharge rate is more technically called "0.1C", where C is the ostensible one hour current capacity). This slow discharge rate tends to result in fairly optimistic ratings; higher loads (such as those imposed by the medium- and high- drain devices that get the most benefit from NiMH batteries) can usually extract less total energy out of the cell than the rated capacity would suggest.
Another issue is that mAh does not measure energy capacity per se. While NiMH cells deliver almost constant voltage throughout their discharge cycle, there is still a small reduction in voltage over time, which in turn reduces the amount of energy delivered to a constant current load. The mAh rating does not take into account this voltage drop and therefore is not a precise indication of the total energy that the charged cell can deliver. To get total actual available energy, the relevant measure is the total capacity in Watt-seconds (Joules), which unfortunately is not generally specified by the cell manufacturers (but is easy enough to measure). The energy to current capacity ratio is probably more influenced by the construction and chemistry of the individual cell than by its most recent charge. However, since the charger does alter the cell chemistry over time, I report both measured current capacity and energy capacity here; compare whichever metric you prefer.
In my measurements of charging completeness, I discharged the test cells at a faster rate (about 2.5 hours, 0.4C) than the standard (10 hours, 0.1C) used for rating the cells, and, as expected, this yields somewhat less total energy than the cells' official ratings. This may make all the chargers appear to be doing a less complete job than they actually are. Therefore, these measurements should be considered only relative to one another and not as absolute metrics of cell capacity under different charging protcols.
All the charges measured in these tests are advertised as "smart/quick" chargers, which means they initially charge at a high rate but monitor cell voltage and reduce or shut off the charge current once the cells are determined to be full. The chargers are designed to reach the full charge at various rates, ranging from one hour (the Rayovac PS4) to eight (the Maha C401FS in "slow" mode). All the chargers have a visual (LED) indication of charge state, with the intent that the charge is effectively "complete" once the quick charge cycle finishes.
Such chargers might be used in two ways: as "quick chargers" with the cells removed and put into service as soon as the initial charge cycle completes, or as unattended "overnight" chargers, with the cells allowed to charge for an additional period of time after the quick charge cycle completes (and presumably "topping up" to their full capacity).
To measure charging completeness under these two modes of use, cells from each charger were measured twice: immediately at the end of a quick charge cycle (as indicated by the chargers' LED displays) and after a full 24 hours.
Each charge-discharge cycle damages the batteries somewhat; excessive or improper charging can result in a reduced number of charge cycles over the batteries' lifetime and in reduced energy capacity. Different chargers are said to vary widely in this regard. One of the most commonly cited factors in charger-related battery damage is excessive heat. Charging tends to heat the battery somewhat, due to internal resistance and cell chemistry. Excessive current as the cells become more fully charged, among other factors, tends to generate excessive heat, and a properly designed charger must be careful to back off the current as the cells become increasingly "full." Better chargers use a variety of heuristics to estimate cell state, including the rates of change in temperature and voltage, the aim being fill the cells as close to capacity as possible without overcharging.
Maha (maker of the Powerex batteries used in these tests) rates their cells for a maximum charge temperature of 45° C. Other manufacturers may rate their cells for other temperatures, but 45° C appears to be close to an industry standard maximum for NiMH technology.
That said, I've no idea how to interpret temperature data to draw meaningful conclusions about any effective differences among the chargers tested; the results here are simply intended as raw data. In particular, while excessive heat in charging is certainly considered harmful to NiMH cells, I don't (personally) have the background to interpret what aspects of the temperature curves presented here might or might not constitute "excessive".
Two temperature tests were done for each charger: one with the cells in the original (semi-discharged) state, as they were delivered, and another after they had undergone a deep discharge (down to an unloaded voltage of about 1V). Each charger used the same cells for both runs, with the same cells measured (in roughly the same spot and with the same thermocouple/transducer, although I wasn't especially careful about this).
One rather surprising result of the temperature tests: in every case, while the charge cycle of a given charger after the deep discharge took longer than the charger's initial cycle with factory-fresh cells (as one would expect), the peak temperature was measurably (and in some cases significantly) lower in the longer cycle (starting from the deep discharge). This supports the possibility that a deep discharge prior to each re-charging could be beneficial for NiMH batteries.
Several figures are given for each of the seven charger/configurations tested. For energy and current capacity, two sets of results are given, one for a quick charge only and the other for a full 24 hour charge. The "Current Capacity" columns give the total number of milliamp-hours extracted from the charged cell (without regard to voltage); if the cell is charged completely, this should be close to the 2200 mAh rated "capacity" of the Powerex 2200 cells (but still less, as per the discussion above). The "Energy" columns give the total number of Watt-seconds (Joules) dissipated. The percentage figures in these columns are relative to the best measurements obtained among all the charger configurations tested (2069 mAh and 8701 Joules, overnight with the C401FS). "Length" gives the number of minutes from the start of charging until the quick charge cycle completed (starting with deep discharged cells) with two cells (out of four slots) in the charger. To measure the current capacity and energy, the two cells were removed from the charger, allowed to cool to room temperature, and discharged (in series, 2.4V nominal) into a 3.01 ohm resistive load (about 0.4C) while recording the voltage.
As a benchmark, Dave Etchell (see "the Great Battery Shootout") rates the Powerex 2200 cells at 2069 mAh and 9090 Joules/cell under aggressive charging and with discharge measured at approximately 0.5C. Observe that the best (24 hour) charge results obtained here are comparable to his figures.
Two peak temperature figures are given for each charger/configuration: one for partially charged cells (in the factory-charged state, as received from the vendor) and one for deep discharged cells.
Charging temperature was taken on the surface of one cell in each charger (measured at a single location) over the first two charging cycles of a fresh battery. Note that these measurements are of cell surface temperature, which is only an indirect indication of their (much more relevant) internal temperature. The temperature peaks generally occurred about 30-120 seconds after a charger switched "off" into trickle mode (the Lenmar being an exception, continuing the charge for about 30 minutes after its temperature peak).
|Performance with Maha Powerex AA220BH 2200 mAh cells|
|Cell Peak Temp.||"Quick Charge" Performance||24 Hour Charge Performance|
|Rayovac PS4||1.7A||40° C||38° C||56 min||1594 (77%)||6760 (78%)||2053 (99%)||8590 (99%)|
|Lenmar MACH1-Gamma||2.3A||38° C||34° C||65 min||1732 (83%)||7284 (84%)||1793 (87%)||7398 (85%)|
|Lightning Pack 4000N||1.0A (pulsed)||31.5° C||31° C||109 min||1795 (87%)||7550 (87%)||2033 (98%)||8554 (98%)|
|Sony BCG34||1.15A||49° C||47.5° C||114 min||1958 (95%)||8218 (94%)||1914 (93%)||7964 (92%)|
|Maha C401FS (fast mode)||1.1A (pulsed)||50° C||45° C||136 min||1975 (95%)||8345 (96%)||2069 (100%)||8680 (99%)|
|Maha C204F||0.54A (pulsed)||45.5° C||43° C||232 min||1920 (93%)||7990 (92%)||2005 (97%)||8341 (96%)|
|Maha C401FS (slow mode)||0.4A (pulsed)||39° C||38° C||423 min||2010 (97%)||8430 (97%)||2060 (99%)||8701 (100%)|
As the table indicates, all the chargers tested (except perhaps the Sony BCG34) provided a fairly complete charge after 24 hours. Time to reach a "full quick charge" and the completeness of that charge, however, varied quite a bit, ranging from less than one hour and 77% of full charge (the Rayovac PS4) to almost eight hours and 97% of full charge (the Maha C401FS in "slow" mode).
The Sony BCG34 travel charger was unique among those tested in that it delivered a measurably more complete charge after only two hours than it did after 24 hours. On further investigation and measurement, it became clear that this charger does not have a "trickle" mode, but rather just completely shuts off as soon as its quick charge cycle completes. Worse, the circuit continuously draws a few mA from the charged cells when it is in this state (this also occurs when the batteries are left in the charger with no power applied). This design flaw makes the charger effectively useless both for unattended overnight charging and as a compact travel charger in which cells can be stored, e.g., in luggage.
Among the five "desk" chargers/configurations tested, the Rayovac PS4 charger is reputed to charge at especially high temperature (enough to reduce the lifetime of typical cells), but my measurements do not appear to support that conclusion, at least on the Powerex 2200 samples I tested. The Rayovac PS4 did raise the cell's temperature more quickly than the other chargers (and appeared to charge at somewhat higher current), which might contribute to a shortening of cell longevity and/or higher internal temperature, but I have no data to draw any conclusions about this. By my measurements, on the other hand, the hottest charger (of Powerex 2200 cells) was the Maha C401FS in "fast" mode, which peaked (in the fresh cell) at 50° C (and kept it above 45° C, the manufacturer's rated maximum, for a good 15 minutes). At the same time, the C401FS (in "slow" mode) was also the lowest temperature desk charger measured. The Maha C204F fell roughly midway between the C401FS's fast and slow modes in both peak temperature and time to charge.
Overall, the RipVan100 Lightning Pack 4000N wall/travel charger was the lowest temperature charger measured, maintaining a cell temperature under 32° C throughout (it was also one of the fastest, peaking in less than 120 minutes even with deep discharged cells). The Sony BCG34, on the other hand, was the second hottest charger measured (approaching the C401FS in fast mode), reaching almost 49° C on fresh cells, and holding above 45° C for more than 12 minutes.
Note that there are small variations (about 1% overall) in the ratios of mAh:Joules among the various chargers and charge lengths. This is partly due to variations (mostly in internal resistance) between the cell samples used in the different chargers, but may also be attributable to differences in cell chemistry caused by different charging regimes (each run used the same cells in the same chargers).
Given sufficient time and attention, most of the chargers tested appear to be capable of bringing NiMH cells close to their capacity. The essential differences between them depend on how they will be used and other factors. In particular, some of the chargers heated cells close to or above their rated maximum temperature, which could cause long-term damage.
Other differences between chargers, which may be important for some applications, include whether they can charge individual cells or must charge in pairs, physical construction and ruggedness, and overall size and weight. An ideal charger would be rugged, small and lightweight, give a fairly complete quick charge in a short time and a very full overnight charge, be suitable for long-term "standby" charging, and keep cells comfortably below their manufacturers' maximum temperatures. Unfortunately, none of the chargers tested met all these criteria.
Lenmar MACH1-Gamma: This charger was unique among those tested in that it is equipped with a cooling fan that runs during the quick charge cycle to help regulate cell temperature. (The fan may have skewed the temperature measurements, however, since it cools -- and the thermocouple measures -- only at the surfaces of the cells, and so the true cell internal temperature may have been somewhat higher.) It was the second fasted charger tested, completing its quick charge cycle in about one hour (and, in fact, its temperature peak suggested that the actual quick charge may have taken only about half of that time). However, at 83% of maximum capacity, its quick charge was not as complete as that delivered by other (slower) chargers. Its 24 hour charge result was especially disappointing, and at less than 87% of the maximum, was the least effective 24 hour charge measured.
Rayovac PS4: This was the largest and heaviest charger tested (even its wall wart was significantly larger than those supplied with the other desk chargers), which makes it poorly suited for portable "travel" use. While it had the fastest quick charge cycle, it also gave a significantly less complete quick charge (78%) than slower chargers did. Its 24 hour charge, on the other hand, was quite complete, comparable to the other chargers tested. Although others have reported high temperatures with this charger, my measurements detected only moderate charging temperatures.
RipVan100 Lightning Pack 4000N: This was one of the smallest and lightest chargers tested. Its lightweight construction may be at the expense of ruggedness, however; cells had to be carefully seated or they tended to pop out (it seemed to be especially fussy with AAA cells). It delivered reasonably complete (87%) charges after a two hour quick charge cycle, and very full charges (98%) after 24 hours. It was by far the coolest running charger of all those tested. There is a built-in deep discharge mode, although for some reason the switch for this feature is rather hidden and only cryptically labeled. Although it reverts to a trickle mode after the quick charge cycle completes, the manual warns against using this for long-term standby charging. (The trickle mode runs at a moderately high current but with a low duty cycle, rather than the more usual trickle charge technique of a low current at 100% duty cycle). This charger is be sold under different brands in other countries (it is an OEM product manufactured in Taiwan by Hubgiant, model CHG2000PDSC, also available from Amandotech).
Sony BCG-34: This charger was also small, lightweight and fast, but basic design flaws make it unacceptable for most applications. It heated the Powerex cells well above the 45° C maximum, and discharges cells once its initial quick charge cycle completes.
Maha C204F: This is a small desk charger, with two (two cell) charging channels and a deep discharge cycle mode. It gave a fairly complete (93%) charge after a four hour quick charge, and a very full charge (97%) after 24 hours. The four hour quick charge time was on the slow side, while heat (45° C peak) approached the manufacturer's maximum (but tended to stay below it). The cell compartment has a plastic cover which is supposed to be kept open during charging; it's not clear why the cover is there at all (it can be easily removed, though).
Maha C401FS: This is also a small desk charger, with almost the same form factor as the C204F, but with four one cell channels and without the deep discharge feature. It has two modes, "fast", with a two hour quick charge, and "slow", with a seven to eight hour quick charge. Fast mode resulted in a very complete quick charge (95%) but unacceptably high heat, peaking at 50° C (hot to the touch). The eight hour slow mode, on the other hand, gave a 97% charge at low heat (below 40° C). Both modes gave the most complete 24 hour charges I measured. Like the C204F, the charger has a useless (but easily removed) plastic cover over the cell compartment.
I expect to use NiMH chargers in two ways: as "home" chargers for a large supply of batteries, and as "travel" chargers to be taken along in luggage and used on a small supply of batteries, e.g., in a hotel room. The requirements are somewhat different for these two applications.
Because of their relatively rapid self-discharge (1% to 3% of capacity per day), NiMH cells should be charged relatively close to the time they are to be used. The obvious way to do this is to start a charge cycle shortly before batteries are expected to be needed, but that requires inconvenient advance planning. Instead, for my "home" setup, caches of charged batteries are kept "topped up" after their initial charge with a low current "standby" charge until they are to be used.
Unfortunately, commercially available chargers are not by themselves well suited to this application. In particular, the smart/quick chargers tested here are not designed for continuous standby charging. The trickle mode into which these chargers revert after their quick charge ends runs at 24-100mA, depending on the model, which greatly exceeds the self-discharge rate of NiMH cells. Also, any power interruption puts these chargers back into fast charge mode, which can potentially damage fully-charged cells. Therefore, different chargers must be used for charging and standby.
My setup entails a moderate-size bank of primary smart/quick chargers plus a larger bank (enough for the cache of cells) of (inexpensive) low-current standby chargers. Upon discharge, cells are first recharged in a primary charger. After the quick charge cycle completes, the cells are transferred to one of the standby chargers, where they stay until they are needed. Note that with this arrangement, charge speed isn't an important concern, since fresh cells are always available at the end of the "pipeline".
For the primary chargers, I settled on Maha C401FSs (used in "slow" mode). For the standby chargers, I settled on a bank of Maha 2A4 wall chargers (not reviewed here; these are "slow" chargers that cost about one fourth the price of the C401FS), which charge 4 AA cells at about 45mA. 45mA is excessive current for continuous standby charging, so a household AC timer runs the standby charger bank at about 10% duty cycle, bringing the aggregate standby charge rate down to approximately 5mA. This should be sufficient to overcome self-discharge, but still low enough to not overcharge high capacity cells. (Thanks to Timothy Brown for suggesting the use of a timer.)
For travel, I'm more concerned about weight, volume and charge speed. I settled on a small set of RipVan100 Lightning Pack 4000Ns, which live in my suitcase and camera case. While the 4000N's two hour charge completeness (about 87%) is somewhat less than that of the C401FS in fast mode, I was more comfortable with its cooler charge temperature. Also, the 4000N's lower weight and internal power supply allows me to pack more of them in my luggage on trips that may run through many batteries. (The 4000N's built-in switching power supply can run on 120V and 240V, 50Hz and 60Hz, which makes it well suited to international travel).
At this time, my active battery cache consists primarily of Maha Powerex 2200 mAh cells (the same model used to test the chargers), but also includes cells from Sanyo (1850 and 2100 mAh), Jetcell (1850 and 2100 mAh), Lenmar (2000 mAh), and other Maha models (1800 and 2000 mAh). All cells are serial numbered and date coded when I put them into service, allowing me to monitor their performance over time.
I bought the Maha batteries and chargers from Thomas Distributing, and the Lightning Pack 4000Ns from RipVan100. (The latter is also sold by Amondotech.) I have no connection with any of these companies, but got good service from them (in October 2003).
Again, these results are intended primarily as raw data; interpret them at your peril. The temperature curve graphs below are normalized for 20°-50° C and 0-480 minutes, although I didn't always collect the full 8 hours of temperature data if the charge/temperature peaked significantly earlier than that. (The cells generally took up to about 30-45 minutes after peaking to fall to their steady state trickle charge temperature). Thermocouple measurements were recorded approximately every 3.6 seconds. (The C401FS in both fast and slow modes exhibited a notable "bump" about 30 minutes before peaking in both runs. I'm not sure what's going on there.)
Figure 1. Temperature curves for Rayovac and Lenmar desk chargers compared, fresh cells.
Figure 2. Temperature curves for Maha desk chargers compared, fresh cells.
Figure 3. Temperature curves for wall chargers compared, fresh cells.
Figure 4. Temperature curves for Rayovac and Lenmar desk chargers compared, after deep discharge (1 cycle).
Figure 5. Temperature curves for Maha desk chargers compared, after deep discharge (1 cycle).
Figure 6. Temperature curves for wall chargers compared, after deep discharge (1 cycle).
Discharge voltage curves looked very similar to one another for the various charging schemes: a rapid drop to the nominal voltage, a long "flat" run, and a rapid dropoff as the cell becomes exhausted. A representative example is given below.
Figure 7. Powerex 2200 discharge curve (.4C) after charging with C401FS (slow mode).
My measurement practices could be charitably described as "unrigorous." The ambient temperature in the room varied by about 3° C (from about 19.5° to 22.5°) over the course of the experiments. The thermocouples were attached to the surface of the cells (at roughly the middle of the cells' shafts) with electrical tape. Two cells were installed in the charger for each run (thus using half of each charger's four cell capacity); on chargers with single cell channels the second cell was inserted in an adjacent slot. Although I made some effort to calibrate the temperature readings to better than factory specs, the basic absolute accuracy of the Fluke thermocouple instrumentation was specified at +/- 2° C plus 0.5% (its repeatability is presumably significantly better than that but was not specified by the manufacturer). A much better test methodology would involve multiple battery samples and many more runs, and it's entirely possible that my samples were anomalous or that some critical experimental error went undetected.
Temperature samples were recorded approximately every 3.6 seconds via a K-type bead-style thermocouple connected to a Fluke 80TK voltage transducer. Energy and current figures were taken by measuring the voltage across the (known) resistive load over the course of the discharge cycles. Voltage samples were recorded approximately every 0.8 seconds across the load. Low resistance measurement of the discharge load was done with the four wire method, although instrument accuracy and calibration tolerances probably limit the absolute accuracy here to about 0.5%, perhaps worse.
Deep cell discharge between the runs and for energy measurement was done at about 0.4C, with a series pair of cells connected to a 3.01 ohm resistor network until the (loaded) single cell voltage dropped to significantly below 1V.
Quick charge and 24 hour charge energy measurements for the different chargers used new Maha Powerex 2200 cells that had been conditioned with three and four deep discharge/recharge cycles, respectively. (NiMH cells reportedly achieve closest to their maximum capacity at approximately this point in their life cycle).
A Fluke Scopemeter 199 was used for all data acquisition; the FlukeView software package collected the data into an ASCII file. The graphs shown here were produced with GNUPlot. Energy and current figures were calculated with some simple AWK scripts.
All images and text Copyright © 2003 by Matt Blaze. All rights reserved. You may not copy, modify or use these images or text, in whole or in part, for any commercial or non-commercial purpose without permission.
Home page: http://www.crypto.com
13 October 2003; revised 19 December 2003