I’ve constantly been receiving requests (see for example this thread) to publish more and more power consumption measurements of current Pocket PC devices. Here you are :)
First, please read at least this article (if you haven’t already done so) to understand what this article is all about. I also highly recommend the other articles I’ve linked in from the Recommended links section.
--HTC Wizard (a.k.a. imate K-Jam, T-Mobile MDA Vario, Qtek 9100, MDA Vario, XDA mini S, SPV M3000, VPA Compact II, Dopod 838) radio power consumption
As promised, I’ve run some long-lasting tests to find out more about the power consumption of the radio unit itself. It’s about 7.5% a day (I’ve measured it for two days; not counting in the 2.5%/day battery life depletion needed to refresh the dynamic RAM memory) in inactive (no SIM is inserted but the radio is online and is ready to initiate emergency calls) mode. Again and again, these figures, along with the 9% (there, not counting in the 2%/day RAM refreshing-related depletion either) of Universal, are relative (show how the power consumption of these devices’ radio compares) and only show how they fare at a given geographical location. That is, they can ONLY be used to compare each device’s being power-hungry, NOT as an absolute result as “your device will consume that much power a day in YOUR home”.
Wizard’s 7.5% a day is an excellent result and is slightly better than the 9% of Universal, particularly taken into account the bigger battery size of the latter.
--Standard power consumption test suite: now, for the Universal and the hx4700!
I’ve also re-run my standard test suite including the HTC Universal (a.k.a. i-mate ?JasJar, Qtek 9000, O2 XDA Exec, SPV M5000, MDA IV/Pro) and the HP iPAQ hx4700 (with WM5 version 2.01 – see FAQ here on this particular ROM version) now to find out how they behave.
The test suite I’ve tested:
Backlight tests: (all with no wireless units (GSM radio, Bluetooth, Wi-Fi, IrDA) switched on, no CPU usage. This finds out the absolute minimum power your device at least will consume when it’s switched on, runs at the default (automatic) CPU speed.)
No backlight at all (Note that in this scenario these devices are far from being usable, particularly the Axim x51v, which has an, in my opinion, slightly less visible screen in external light than the other three devices (which all have a screen using the same Sony technology and, therefore, behaving in a very similar fashion)
(Absolutely) minimal backlight
Maximal backlight (may be useful to know for example when watching videos on Sony screens because of the very high contrast and saturation of these screens (the x51v's not-really-saturated, not-contrasty screen has a definite advantage here) or in outdoor circumstances when you need to use the maximal backlight level available to see anything)
Wi-Fi searching for networks (also testing possible power saving modes to find out whether they’re of any help)
CPU usage vs. power consumption tests with Resco Audio Recorder (recording in q:3 32 kHz 19 kbps Speex mode (my favourite one with the best size/quality ratio but, unfortunately, with pretty high, about 60% CPU usage) - please see _Everything you may need to know about sound recording on the Pocket PC_ for more information.)
Note that, this time, I’m also stating the original charge level at the start of the test. As is also stated for example in this pretty nice article by Brighthand forum member tanbam, there may be slight (not much) differences in power usage depending on the charge level of the battery because of the proportional Voltage level decrease as the battery is discharged. (Note that the Voltage decrease with Li-Ion batteries is in no way as visible as with other kinds of batteries. That is, a Li-Ion battery that only holds little charge will still have only a bit less Voltage than a battery full of charge. This is why there aren’t major differences in the low-charge and the fully-charged cases, Amperage-wise).
As can clearly be seen, what I’ve stated in my first battery consumption-related articles and tests is still topical:
when there is some remarkable CPU usage (see the Resco test), the power consumption figures skyrocket, particularly with the iPAQ and the x51v, particularly because both are 624 MHz devices, as opposed to the 520 MHz Universal and the PL720 (the latter consume decidedly less power in these cases). It’s, therefore, essential that you use some kind of CPU underclocking, especially on 624 MHz devices, when you run an application that constantly uses the CPU.
Fortunately, CPU underclocking is already supported by the x51v (in the Processor tab in the Settings/System/Power applet) but, unfortunately, not in the hx4700. Note that the Resco results are almost the same in the 624 and the 520 MHz modes; it’s only in a heavily (on the verge of usability underclocked) 208 MHz case that it’s visible decreasing. The case is a bit different with games – please see my former articles on the power consumption of the x51v and on my remark on what applications / games run OK when you (radically) underclock the device.
the Pocket Loox 720 is by far the best device, power consumption-wise
the lowest backlight level (with the – this is very important! – automatic backlight mode enabled as can be seen in here – it’s not enabled by default!) with the hx4700 consumes very little, as opposed to the other devices
when idling (without backlight – again, the hx4700’s backlight is far more battery-friendly than that of the other devices when used sparingly), the PL720 consumes the less; then comes the Universal, the x51v and, finally, the worst-behaving hx4700.
the various Wi-Fi power saving modes aren’t worth anything (at least when the device is actively searching for networks). In Wi-Fi, it’s again the PL720 that turns out the most power-saving (in my personal tests, it consumes about 25% an hour with BT enabled – connected to a ?StowAway BT keyboard -, in Web browsing mode, with the lowest backlight and connected to a Wi-Fi network).
BT and IrDA: Note that the chart doesn’t contain Bluetooth and IrDA beam activation power consumption data because I’ve already published them all. In a nutshell: you can freely activate and, with Bluetooth, actively use them (it’ll only result in an additional 1-2 mA power consumption, except on the PL720 in BT PAN mode, where you have taken into account what I’ve stated here), except for the hx4700, where it adds another 80-90 mA power consumption.)
finally, the Universal has turned out to be pretty good surprise: I’ve expected far worse, based on real-life usage. (Read: I need to recharge it on a daily basis with even moderate Web browsing – no hard-core gaming or stuff at all! – and mail reading while, using it in entirely the same way, the HTC Wizard doesn’t need recharging for days. Compared to the Wizard, it really sucks battery life-wise; compared to other VGA devices, it fares pretty well.) It fares decidedly better than the x51v and the hx4700 in almost every respect, particularly in the CPU usage and the Wi-Fi tests.
--The comparison / result chart
Device:
X51v A12
HTC Universal
HP iPAQ hx4700 2.01
PL720
Initial charge
100%
78%
67%
86%
No backlight
75
65 (30 mA when the lid is closed and the screen dimmed)
94...100 (interesting it’s only some 10 mA’s less than with the absolutely lowest backlight level). Note that these results are slightly higher than those of tanbam, who measured about 10 mA less in this case. The cause for this is unknown - may be a 2.0 vs 2.01 ROM version issue?
57
Lowest backlight
127
128
117 (note that this wasn’t the lowest possible value but a value that results in a similar backlight level than the x51v and the Universal; setting (by enabling the auto backlight mode) the absolutely minimal backlight it consumes 102...110 mA.
this
in
333...334
auto
in
283...288
200
192...249
255...330
182...189
405
341
300