Log in

View Full Version : A VERY BAD BUG discovered in the WM5 2.01 hx4700 BT module!


Menneisyys
08-22-2006, 11:43 AM
Upon request (http://www.pocketpcmag.com/blogs/index.php?blog=3&p=1165&more=1&c=1&tb=1&pb=1) (please also see the just-linked article for a more thorough elaboration on the background of all this and how the current screenshots should be interpreted), I’ve made some power consumption tests on the HP iPAQ hx4700 regarding the wireless units.

First, the Bluetooth (BT) unit. Please see the following screenshot taken with acbPowerMeter:

http://www.winmobiletech.com/kuvat/HX4700btuNITAndLEDPowerConsumption.bmp.png

It shows three cases (note that these tests were all done with the disabled Navi service. Without disabling it, there were considerably more and much higher, transient peaks in the chart as can be seen in this shot (http://www.winmobiletech.com/kuvat/EffectsOfNavi.bmp.png)):


BT unit off; ~102 mA
BT unit (and the blue LED) on (the first two peaks and the first half of the third; note that, in the third peak, I’ve also disabled the BT being discoverable. This had absolutely no effect on the power consumption); ~188 mA
BT unit on, blue LED off (the second half of the third peak); ~186 mA (as can be seen, the bright, blue LED consumes about 2 mA; see my disable-the-hx4700-LED posts in my blog (http://www.pocketpcmag.com/blogs/index.php?blog=3&p=928&more=1&c=1&tb=1&pb=1) on how the LED can be disabled).


What does this mean?

Unlike with ANY Pocket PC I’ve ever tested, the Bluetooth unit in the hx4700 consumes a LOT of power even when idling (and even with the BT LED/being discoverable disabled).

Please note that this is NOT a CPU usage issue – I’ve thoroughly tested whether this problem is caused by a badly written piece of code that would just “kick in” with considerable CPU usage. It is unlike the case with the Bluetooth Manager PAN bug of the Pocket Loox 720 I’ve described here (http://www.pocketpcmag.com/blogs/index.php?blog=3&p=972&more=1&c=1&tb=1&pb=1).

That is, while you can safely keep the BT unit on all Window Mobile devices I’ve ever tested (in this respect), the hx4700 with the new, 2.01 WM5 ROM is an exception. Only enable the Bluetooth on the hx4700 when it’s absolutely necessary!

What about the Wi-Fi unit?

The Wi-Fi unit, as with all the other Pocket PC’s I’ve ever tested (please do read my previous articles on power consumption issues if you haven’t done so!), behaves as supposed to. If you enable it and it’s in the “searching for networks” state, then, the net power Amperage to power the Wi-Fi unit will be around 223 mA (the average Amperage the device takes being around 325 mA), which means the power consumption of the entire device will be around 3.2 times more than with the Wi-Fi unit disabled. This screenshot (http://www.winmobiletech.com/kuvat/HX4700WiFiPowerConsumption.bmp.png) shows this case (the previous case with, later, the Wi-Fi unit enabled. Note that the peak shows the unit’s having found a wireless network).

What about the suspended state?

Note that when suspended, wireless units won’t consume any energy because suspending a non-Phone Edition device (like the hx700) also means suspending its Bluetooth unit – as opposed to the Phone Edition case, where suspended Phone Edition devices still have an actively ‘listening’ Bluetooth unit. This is why you don’t need to power on your Phone Edition Pocket PC when you want to use it as a, say, modem via Bluetooth.

Request

Could please anyone test the additional power consumption of the Bluetooth unit with the previous, 2.0 WM5 and the WM2003SE ROM versions? I’d really like to know whether this bug only affects the 2.01 ROM version of the hx4700. I think so – IIRC, I haven’t seen such bugs when I tested the WM5 2.0 ROM version upon testing the hx4700 for BT anomalies for my PL720 bug article.

hx4700 users with the 2.01 ROM version may also want to give it a try just to confirm the bug’s existence.

Just use acbPowerMeter to test the power consumption.

yildi
08-22-2006, 03:30 PM
Hi Menneisyys,

Here is my screen shot (well, I will try to attach it later, this editor does not seem to give me this possibility) with my 4700 under WM2003SE using the following protocol:
- Nothing for 15 seconds;
- Bluetooth only for 15 seconds;
- Bluetooth with the Think Outside keyboard active (typing text in Word) until the end.

http://beagle.u-bordeaux4.fr/yildi/files/powermeterWM2003SE.jpg

Murat

David_C
08-22-2006, 03:36 PM
I have not ran any tests when I was running WM2003SE, but I did notice that I used a LOT of power whenever I was running BT.

I would use about 10% every 30 mins. (it woudl go from say, 94% to 84% in about 30 mins).

Like I said before, this is purely an observation, no tests were run.

While running WM5 2.01 the same thing was happening.

I had just assumed that the BT feature ate up a lot of power.


Dave

Menneisyys
08-22-2006, 03:45 PM
Thanks to you both!

Good to know it's not a bug introduced in 2.01 but is clearly a hardware-related problem.

Strange the BT unit in the hx4700 comnsumes so f*ckin much power... The HP engineers / HTC have ROYALLY messed up something...

(Thanks again for the screenshot - Will also link in this thread from my blog / from other sites I've cross-posted the article to.)

buckyg
08-22-2006, 04:30 PM
Thanks to you all for explaining what I have seen. Soon after upgrading to 2.01 I went from an extended battery back to the regular battery. I had not run any tests or figured out why but sometimes noticed that the battery would drain fairly quickly. It was not consistent but I had not noticed it before. Yes, BT being the culprit makes sense.

The real shame of this is that I have to use BT in order to sync with my work laptop when not in the office. For the PPC, that is because I have to use AS 4.x and for my laptop, I have to connect it remotely to work using a VPN. (I am using AS 4.2, btw, still have to do this)

Guess it's back to the extended battery. I just kind of liked having the smaller look and feel with the regular battery...