[Q] Signal Lost with Screen Sleep

mephissto

Member
Mar 6, 2010
36
3
0
Clermont-Fd
@neotekz : I'm not in a weak 3G signal, and I still have the issue.

Someone on the Google code thread said that the issue disappeared when he flashed uikl3 radio. If someone want to try, as I can't do it myself right now :/
 

dosmage

Member
Jun 2, 2008
7
2
0
Goleta
I've been meaning to do more troubleshooting, but I was busy all day yesterday and today. Here's my two cents on the problem. I believe the issue is most definitely not a radio firmware issue because when I flashed to Cyanogenmod I didn't flash a new radio. Cyanogenmod 9 running 4.0.4 exhibited the same issue as the stock 4.0.4 rom. This problem is not signal strength related as the radio doesn't recover, reports that the mobile network is disconnected, and Opensignal Maps doesn't show radio information. It's obviously power management related as the issue occurs only when the phone is put in to standby. Here's where my knowledge of GSM is lacking. During the problem you can't scan for mobile network carriers and the more disturbing thing is that the APN settings are empty and you can't create new APN settings. Therefore I believe that perhaps the SIM card is being powered down in some way. I tried to find a SIM card app in the market, but all I could find was anti theft related apps. I haven't had a chance to figure out a way to test my theory and as I said I am not very GSM savvy.
 
  • Like
Reactions: memoreks

DaLoona

Member
Mar 27, 2010
49
6
0
I too had this "problem", as soon as it went into stand-by/sleep, I couldn't get incoming calls, no sms, whatsapp or anything else.
Had it on the OTA 4.0.4, which I got from XDA and I flashed by the way, currently on AOKP 29 + Franco r.116.
And as others before me posted, raising the minmum cpu speed from 350MHz to 729MHz did the trick. Standby now works for me.

And if it worked before and still on old 4.0.2 but not on newer versions of ICS I dont think it is a hardware fault, but something screwed up in the code.
 
Last edited:

fishko

Member
Feb 9, 2010
8
0
0
Same problem here since the update on 4.0.4... Went back to 4.0.2 and the problem was gone... now testing on 4.0.4 cpu min speed 700mhz ... seems everythig fine.
 

warhummer

Member
Sep 20, 2008
24
4
0
Southern MD
Interesting turn of events

I just flashed Android Revolution's 4.0.4 version and it appears I do not have the signal-drop-while-sleep anymore. The signal bars do occasionally turn grey, but they always had full bars and the "E" moniker (I'm firmly in a T-Mobile Edge-only area). I do have SetCPU installed, but it's not set to change the CPU speed (I think). I'm going to go back an verify in the meantime...my "scientific method" is getting sloppy in my old age.

I let phone sleep and made repeated phone calls at random points throughout the day, as well as text messages and all went through as expected.

If anyone has a similar experience, please let me know.
 
Last edited:

fishko

Member
Feb 9, 2010
8
0
0
Same problem here since the update on 4.0.4... Went back to 4.0.2 and the problem was gone... now testing on 4.0.4 cpu min speed 700mhz ... seems everythig fine.
No signal loss anymore with 700 mhz min cpu speed. Yet another issue :) ... 5-10 %/h battery drain in stand by instead of 2-5%/h with cpu on 350 mhz...

Sent from my Galaxy Nexus using XDA
 

DaLoona

Member
Mar 27, 2010
49
6
0
No signal loss anymore with 700 mhz min cpu speed. Yet another issue :) ... 5-10 %/h battery drain in stand by instead of 2-5%/h with cpu on 350 mhz...

Sent from my Galaxy Nexus using XDA
With AOKP 29 and Franco r.116 Im getting ~2 %/h @729MHz in standby.

Edit:

This is not to be biassed by the way, but to show you can improve batterylife by trying different ROM's and kernels, this combno just works great for me :D

Sent from my Galaxy Nexus using XDA, killing iPhones along the way.
 
Last edited:

nicholasbgr

Senior Member
Jun 29, 2010
213
19
0
@neotekz : I'm not in a weak 3G signal, and I still have the issue.

Someone on the Google code thread said that the issue disappeared when he flashed uikl3 radio. If someone want to try, as I can't do it myself right now :/
This was the first thing i tried since I'm from Brazil and UIKKL3 is the brazilian radio and it didn't fix it for me.
What I find curious is that the issue is a little different on my phone, the signal doesn't dies for good, it just becomes intermittent.
Setting the minimum clock to 700mhz did seem to "fix" it for me but I'm afraid the battery hit wont be gentle.


I have been having this issue since 4.0.3 on my GSM Nexus. I have tried multiple radio's and multiple kernels. I was having to change the min cpu to 700mhz. I tried FuGuMod's Kernel and it seems to have fixed my signal issues. I am able to keep the min cpu at 350mhz. If you search the thread he explains what he thought was causing the issue. Once he disabled that feature in the kernel my signal issues went away.

http://forum.xda-developers.com/showthread.php?t=1438417

I think its a mix between software and hardware issues. There is some software setting that the hardware doesnt like. But since this doesnt happen to everyone running a certain kernel and only some people there must have been a bad batch of hardware from samsung.

I think I am going to try a warranty exchange with Samsung.
Here's what I found about the feature he advises to turn off (Smart Reflex)

Smart Reflex

The variations in the silicon manufacturing processes are responsible for the difference in performance and/or power consumption of the device. Adaptive Voltage Scaling- AVS, implemented with SmartReflex(TM) technology, tends to narrow this difference.

Dedicated SmartReflex hardware implements a feedback loop - without processor intervention - which optimizes the voltage levels to account for differences in the manufacturing process, temperature and silicon degradation.
To enable smart reflex

# echo 1 > /debug/voltage/vdd_core/smartreflex/autocomp
# echo 1 > /debug/voltage/vdd_mpu/smartreflex/autocomp

To disable smart reflex

# echo 0 > /debug/voltage/vdd_core/smartreflex/autocomp
# echo 0 > /debug/voltage/vdd_mpu/smartreflex/autocomp


Impact on Power Consuption on Android System

This feature is hardware controlled after initialization. Initialization is done from the kernel. Android Power management is completely unaware of smart reflex.

Power consumption is reduced since the operating voltage is lowered for each operating point in which the device is operating in.
To minimize the power consumption in runtime

The maximum power reduction in runtime is achieved when “enable_off_mode”, “sleep_while_idle”and “smart reflex” options are enabled and dvfs option is set to “power save”.

To minimize the power consumption in suspend

The maximum power reduction in suspend is achieved when “enable_off_mode”, “sleep_while_idle”and “smart reflex” options are enabled and UART console timeouts are set.

Please check the power consumption analysis at http://processors.wiki.ti.com/index.php/Android_Benchmarks_DevKit2.3.4v2.1#Power_Management

http://processors.wiki.ti.com/index...d-2.3.4-DevKit-2.1_PortingGuides#Smart_Reflex
 
Last edited:

rob61280

Senior Member
Jul 28, 2005
272
20
0
Liverpool, U.K.
I tried XXLA2 and UGLC1 and the signal drop persisted with both. I then went back to UGLA4 and I have not experienced a dropped signal in 24 hours.

Your mileage may vary depending of your location and operator, I live in the U.K. but the North American radio suits my phone better, its worth trying a few of radios before assuming you have a hardware fault.

Also in the past fugumod kernels have helped when I had this issue on 4.0.3

R
 

j18ter

Member
Apr 3, 2012
5
4
0
Had the opposite experience here, after installing the yakju OTA upgrade, which left me with the XXLA2 radio, it seemed the signal was dropping quite low, but it never went away completely, not even when the device goes to sleep. Still, it seemed lower than under 4.0.2, so I tried UGLC1, which is supposedly optimized for North America. This doesn't work for me at all, regularly losing the signal completely when it goes to sleep. So I went back to XXLA2 from Google's 4.0.4 factory build for now.
[update] It turns out both radios drop signal when the device goes to sleep, not always, but enough to make this 4.0.4 upgrade a show-stopper. So what is the best work-around? Rooting and controlling the CPU frequency?
 
Last edited:

memoreks

Senior Member
Apr 5, 2007
51
17
0
King's Lynn
Update from Google

This was posted to the issue on Google's website:

Comment #19 on issue 28133 by [email protected]: Signal lost when Galaxy Nexus is in Standby/idle - Android 4.0.4 Only

---------------------------------------------------------------------------

Hi folks,

In order to look into this issue on our side, we'd like to collect bugreport.txt files from those who received the OTA update to and remain on Android 4.0.4.

If you're experiencing a loss of signal during sleep after the 4.0.4 update, follow the instructions below:
support[dot]google[dot]com/mobile/bin/request.py?&src=6279961&contact_type=android_bugreport
(Note that this link is only valid for reports about signal loss)


We appreciate this information as we look into your reports,
Ethan

-------------------------------------------------------------------------

I've unfortunately rooted mine to fix it, as it is my work phone and thus I can't mess too much with it, but anyone who can recreate this issue please update the bur report via the link above, and Google will be looking into it.
 

dosmage

Member
Jun 2, 2008
7
2
0
Goleta
As previously mentioned in this post, in the open issue with Android someone has asked us to generate bug reports and submit them. I've reflashed my phone and submitted one as well as what appears to be 3 others before me. If you are experiencing this problem I highly encourage you to do so they can get an idea of the impact.
 

Hidjes

Senior Member
Dec 18, 2011
170
24
0
Bilthoven
You have to prevent to let the phone go into deep sleep.
You can do this by setting (in SetCPU for example) not at a minimum of for example 200GHZ but a little bit higher, so it is awakened all the time. GL
 

j18ter

Member
Apr 3, 2012
5
4
0
I tried the other recommended fix, namely to root the phone and then use adb to change the following kernel variable from 1 to 0, i.e.,

su
cd /sys/kernel/debug/smartreflex/sr_core
echo 0 > autocomp

Contrary to what an earlier post may have implied, this doesn't require using a custom kernel. I've done this with the stock kernel in 4.0.4, and since then the signal hasn't been dropped. I'm guessing this setting will be lost after reboot, so a more permanent way to set this is needed. I don't know the impact on battery consumption yet.

J
 

memoreks

Senior Member
Apr 5, 2007
51
17
0
King's Lynn
Battery life

While I haven't got proper metrics to back it up (I have a day job!), I haven't found the battery life of 4.0.4 with the 700mhz fix applied, to be any different to 4.0.2 battery life.

As a rough guide, my GNexus with a 4.0.4 ROM and the "700mhz fix" applied has lost on average 2.4% per hour today, with 3-4 phone calls and a bit of light usage. This is kind of what I expect from a modern smartphone without going crazy on custom kernels etc.

If your GNexus is disconnecting from the GSM network for long periods of time with an unfixed 4.0.4 ROM, then its bound to improve battery life as it is not keeping in contact with a cellular base station, and is powering down the relevant radio parts of the phone.

If anyone has the spare time, then it would be interesting to compare battery utilisation between a working 4.0.2 and a fixed 4.0.4 installation, and see if the 700mhz increase makes much difference. My guess is that it won't.
 
  • Like
Reactions: j18ter

allio

Member
Jan 25, 2012
5
1
0
While I haven't got proper metrics to back it up (I have a day job!), I haven't found the battery life of 4.0.4 with the 700mhz fix applied, to be any different to 4.0.2 battery life.

As a rough guide, my GNexus with a 4.0.4 ROM and the "700mhz fix" applied has lost on average 2.4% per hour today, with 3-4 phone calls and a bit of light usage. This is kind of what I expect from a modern smartphone without going crazy on custom kernels etc.

...

If anyone has the spare time, then it would be interesting to compare battery utilisation between a working 4.0.2 and a fixed 4.0.4 installation, and see if the 700mhz increase makes much difference. My guess is that it won't.
I think you're broadly right, that 4.0.4 with the 700MHz "fix" probably behaves similarly to 4.0.2.

If your GNexus is disconnecting from the GSM network for long periods of time with an unfixed 4.0.4 ROM, then its bound to improve battery life as it is not keeping in contact with a cellular base station, and is powering down the relevant radio parts of the phone.
In my experience, this isn't true. The constant attempts to re-establish the signal actually prevent the phone from going to sleep at all. I rebooted my phone half an hour ago and it has probably one minute of screen on-time since then, but it's been awake 95% of that time due to the constant disconnections and attempts to reestablish the network link. Battery life is plummeting like a stone as a result - 6-7% in half an hour.

I believe the 4.0.4 update has enabled some more aggressive power saving techniques (smartreflex) that weren't used in 4.0.2. This is resulting in greatly improved battery life for most people. Unfortunately, on certain devices (such as ours), it's proving to be problematic.

Clearly it needs a fix from Google - I just hope the fix results in us getting the same power savings everybody else does, and doesn't leave us out in the cold by simply turning off the feature.

I've submitted my log and starred the bug report.
 
  • Like
Reactions: j18ter

rootusr

Senior Member
Dec 5, 2010
383
39
0
UPDATE: not :( It's not working, after a some idle time I'm losing connectivity again...seems that the only option is to modify CPU frencuencies

Having the same problem with Stock rooted YAKJU since the update to 4.0.4.

I can't understand why since seems to be a Kernel issue but i flashed UGLA4 and the problem went away... so, is this really a Kernel problem?


I already report this to google.
 
Last edited: