Avoid Built In Rechargeable Batteries
Insist on standard batteries
My first and best advice is that if it doesn't run on AA or AAA batteries, forget it. This advice will be ignored. Even nicer would be to run on both alkaline AND rechargeable AA or AAA batteries. Even devices that accept AAs do not, they take one, but not the other - very honourable exceptions, the old HP200LX, and also the TRG Handera 330 which can take AAAs or LiIon pack.
Avoid Custom Batteries
Custom batteries die, and then you can't find replacements. Forget any computer that requires a built in custom battery. After a few years, they are invariably unobtainable. They are unobtainable right from the start if you visit obscure places. Many devices have the battery built in, and it can not even be replaced. If the gadget is so cheap you just throw it away when it stops, this may be fine. If expensive PDAs tend to fall into this category, then I suggest avoiding all such models.
Final reason to avoid custom batteries. I have never owned a battery powered computer that didn't (eventually) have the batteries die. Equally, I have never owned a computer in which the batteries died, and for which I was still able to buy replacement batteries when they did die.
As a result of years of experience of batteries dying, I now utterly refuse to buy another computer with custom batteries. Not ever. No matter how good the computer is otherwise. (What, never? Well, hardly ever!) I do note that some third party companies will repack battery packs if the computer happens to use a standard rechargable cell in the pack, as many do.
Battery capacity (at affordable prices) isn't increasing all that fast (you just can't get all that far on chemical energy). That is one reason notebook makers are talking about esoteric (and possibly dangerous) gadgets like fuel cells.
Nickel cadmium was the standard before about 1995. An AA sized NiCd cell would provide anything from 400mA-hour to 900mA-hour (at say a typical PDA discharge rate around 50mA per hour). Nickel metal hydride NiMH tended to provide 1100mA-hour in 1995, and this increased by maybe 20% over the next few years, with perhaps another 10% to 25% to come. I expect them to peak around 2000mA, and this seems to be what has been reached in 2003. In 2005, some were claiming 2500mA, but I haven't tested this claim.
NiCd has a long cycle life (up to 1000 recharge cycles), good low temperature and high discharge rate performance (up to 20 times capacity), long shelf life and rapid recharge capability. This is great for toy cars and high power devices, but mostly irrelevant for PDAs. They self discharge at about 20% a month.
NiMH has a higher energy density, rapid recharge, long cycle (400 cycles) and shelf life, but the high discharge rate (about 5C maximum) isn't as good as NiCd (which doesn't matter with PDAs - but does with notebook computers or portable power tools), and poor charge retention (about 30% a month, so gadgets die when stored). Discharge characteristics are similar to NiCd. They sit at 1.2 volts and discharge is pretty flat until near 75%-80% of capacity. Then the voltage drops sharply to about 1 volt. If you take them below 0.8 volts you can get excessive gas discharge.
Both these styles of battery are high maintenance. You really need to dischange them and recharge about once a month.
NiCd are charged at a constant current. At about 80% charge the temperature starts to rise. Mostly you terminate charging by timer, temperature cutoff (if available), negative delta V (battery voltage will peak and then decline very slightly, and detecting this is the best method), or trickle charge at 0.02 to 0.1 C rate. NiMH are similar, but delta V voltage drop is much less or even zero, so rate of temperature rise is a better indication of full charge.
Notebook computers started going over to lithium-ion in 1995. Li-ion provides 50% more energy per volume and 80% more per unit weight than NiMH. The disadvantages are that demand exceeds production, they cost up to 50% more, and they need very capable protective circuitry for safe charging. They self discharge at around 10% a month, so store reasonably well. Discharge rates should never exceed 2C, unlike NiCd. There are some indications that Li-ion may have a service life of about 3 to 4 years, and a tendency to die after this time, whether in use or not.
You can charge NiCd or NiNH with a simple constant current charger. Set up a resistor so the charge is held at 15 hour charge rate say. Li-ion needs a current limited constant voltage. You charge at maximum current until it reaches maximum charging voltage (which depends on the manufacturer), then maintain battery voltage to within 50mV. Newer high capacity NiMH are more sensitive to charge rates, as makers change the electrode structure.
Battery Management ICs
Linear Technology, Micrel, National Semi, Benchmarq and others make suitable ICs for chargers. But get the voltage right, because a Li-ion will be damaged by even very small overvoltages.
Benchmarq, Integrated Circuit Systems and others make battery gauges, typically measuring both charge and discharge currents. Since discharge voltage is pretty constant on NiCd and NiMH, this works reasonably well (unless the battery is not used for a period, when they badly underestimate remaining charge). The Bq2050 handles Li-ion batteries by taking into account battery voltage, correcting for temperature and self discharge.
Most Li-ion battery packs now include appropriate charge rate controllers, since the temperature sensors need to be added to the battery pack. General opinion is that this is a better idea than building the sensing circuits into the equipment. Duracell have been pushing their Smart Battery Data (SBD) specification, which stores 34 battery parameters. Data is transmitted via the System Management Bus on a two wire I2C bus subset. Some PCs have BIOS support and an interface to Microsoft's Advanced Power Management (APM).
My own experience of Li-Ion is that they are no more reliable than NiMH. I got less than 18 months and 20 charges out of a Micron Li-ion battery in a Micron notebook computer. I suspect failure of some aspect of the intelligent charger.
Here are some standard battery sizes.
C (usually as 4500mA NiMH)
Sub-C (usually as 2400-2700mA NiMH or 1300mA NiCd for drill power packs) 43mm long, 22mm diameter.
1/2 Sub-C 26mm long, 22mm diameter.
4/5 Sub-C 33mm long, 22mm diameter.
D (usually as 8000mA NiMH)
2/3A (usually as 700mA NiCd) 28.5mm long, 16mm diameter. Used in Braum electric toothbrushes, for example.
4/5A (usually as 1800mA NiMH) 43mm long, 17mm diameter. Used in electric toothbrushes, laptop computers, drills.
4/3A (usually as 3000mA NiMH) 67mm long, 17mm diameter. Used in laptop computer battery packs.
Rechargeable AA tests
Four different brands were on hand for testing. A standard NiCd, and three NiMH from Innotec, Blue, and Varta.
Trickle charged all cells for 48 hours, and then measured their voltage over a period of time while they self discharged (please note that multimeters of the quality I use are not precision instruments, so don't take the exact values too seriously).
Time Now 1hour 8hour NiCd 1.312 1.300 1.284 Innotec 1.370 1.354 1.336 Blue 1.371 1.358 1.340 Varta 1.367 1.364 1.358
Time 24h 34h 48h 58h NiCd 1.264 1.247 1.247 1.243 Innotec 1.324 1.319 1.304 1.299 Blue 1.330 1.325 1.312 1.309 Varta 1.353 1.350 1.338 1.336
Trickle charged all cells for 58 hours, and then measured their voltage over a period of time while they discharged through a 10.5 ohm resistor to simulate heavy use in a pocket computer.
Time Now 45m 1h 1h45m NiCd 1.223 0.148 (dead) Innotec 1.280 1.223 1.212 1.197 Blue 1.282 1.202 1.197 1.183 Varta 1.236 1.188 1.203 1.160
Time 2h15m 2h45m 3h 4h 5h NiCd Innotec 1.184 1.178 1.173 1.156 1.109 Blue 1.175 1.172 1.169 1.160 1.149 Varta 1.082 1.161 1.073 1.083 1.044
Disconnected the load for four hours, and then reconnected and continued measuring with the 10.5 ohm load.
Time End 15m 40m 1h Innotec 1.133 0.506 0.630 Blue 1.172 1.146 1.142 1.137 Varta 1.143 0.836 0.916 0.889
Time 1h20m 1h40m 2h10m 3h15m 4h Innotec Blue 1.132 1.126 1.119 1.074 0.731 Varta 0.889 0.870 0.793
Charged all the batteries for 27 hours, and repeated the 10.5 ohm discharge test.
Time Noload 10.5ohm 15m 30m 45m 1h 1h15m NiCd 1.393 1.251 1.265 1.252 1.228 1.180 1.123 Innotec 1.383 1.310 1.283 1.261 1.240 1.220 1.197 Blue 1.419 1.331 1.265 1.225 1.199 1.179 1.167 Varta 1.411 1.294 1.327 1.315 1.297 1.283 1.279
Time 1h30m 1h45m 2h15m 2h30m 3h 4h 5h 7h NiCd 1.101 1.083 1.034 0.879 0.97 Innotec 1.155 1.16033 1.146 1.136 1.123 1.112 1.077 0.733 Blue 1.163 1.15779 1.141 1.132 1.125 1.108 1.092 1.079 Varta 1.259 1.25690 1.248 1.240 1.237 1.207 1.054 0.190
For a future set of tests, I'm going to build sets of custom voltage and current measuring gear, and take continuous measurements using a computer for control and recording. I got distracted by real life instead of doing the tests more carefully last time.
Widget.co.uk sell a solar battery for Psion, so you can use it in the field. The price seemed fairly steep (over forty pounds). However unless it charges an external rechargeable battery pack, I really doubt it will work. Small solar cells just don't do a good job of matching the power demands of a Psion.
Experiments with some small solar cells to follow.
An aluminium framed amorphous solar battery, nominally rated at 2 watts, and nominally 6 volts (no load output seems to be between 8 and 10 volts), while peak current seems to hit 150 mA (winter day, tropics). The size is a nominal 30cm x 15cm, with a cell area of 14cm x 29.5cm, or a surface area of 413 square cm (or 0.0413 sq m). If solar levels are really 1kw sq m (which is what cells are rated at), then this gets 41 watts ... however the efficiency is down somewhere in the 10% or less range, and you rarely get that amount of solar input even in the tropics. So 2 watts seems a fair rating.
Connected it to a Psion 5. It can run the LED, but the Psion internal circuits don't recognise it is supplying power.
Typical PDA Power Consumption Figures
Power consumption of a Psion 5, measured at the plug pack (6.3 volts DC, positive tip)
13.8mA when off (LED and maintaining memory) 70 mA on startup 38 mA with no activity 81 mA updating display 85 mA with backlight, no activity 131 mA with backlight and display updating 86 mA on D: with backlight 37 mA on D: but inactive
A typical PalmPilot has been reported to draw (very approximately):
80-90 mA on, backlight on, CPU busy, serial port active. 50-60 mA on, no backlight, CPU busy, serial port active. 30-60 mA on, processor busy (pen down, playing a game, etc.) 15-25 mA on, idle but displaying data (calendar, for example). 1.2 mA off, serial port DSR line terminated(~3K Ohms). 0.5 mA off (sleeping), Palm modem attached. 0.13 - 0.3 mA off (sleeping), out of cradle.
dividing into 900 mAH we get around:
10 hours continuously on with backlight and serial port active. 2-3 days always on, using memopad. 120-280 days off (sleeping) and unconnected (e.g. shelf life).
I'm finding that the [Compaq] 2010c's proprietary battery is becoming scarce and I'm looking for a cheap, easy solution. Can I use 2 rechargeable AA NiMH off the shelf or will I ruin the handheld ? I'd appreciate any input. Thanks. (NetNews Dec 2000)
Does any know where you can get a rechargable battery for the 2010C (Compaq). Mine will not hold a charge for more than a few hours used or just sitting around. I would really hate to have to get rid of the HPC. It has been a great HPC and I really don't want to part with it. But I have exhausted my search options.... BTW I did check with Compaq and got the "We no longer support that model.... So I am hoping that someone here can help (NetNews Jan 2001)
Let me know if you have a functional battery for the Compaq Aero 2100 series you would like to sell. kshope at ntwrld.com (NetNews Dec 2001)
LG Phenom Express. If you find replacement batteries tell us on the net ... (NetNews Aug 2001)
I need to follow up comparative value of Lithium throw away batteries, and also the Energiser e2 Titanium batteries (mid 2000) that are said to last 85% longer than standard alkaline.