Difference between revisions of "Problem with garbled screen"

From ThinkWiki
Jump to: navigation, search
(Solutions)
(Affected Models)
 
(23 intermediate revisions by 17 users not shown)
Line 8: Line 8:
 
It has furthermore been reported that there might be a relation between the temperature of the ThinkPad and the garbled screen, so that the display starts working fine when the ThinkPad has reached a certain temperature (like after 5 minutes of being powered on). Also by applying pressure to a certain point underneath the ThinkPad (on A30s its below the information sticker on the underside). This seems to affect the garbled screen even more, which is highly noticable on a LILO boot screen [http://www.drakonus.dragoncity.net/thinkpad like so].
 
It has furthermore been reported that there might be a relation between the temperature of the ThinkPad and the garbled screen, so that the display starts working fine when the ThinkPad has reached a certain temperature (like after 5 minutes of being powered on). Also by applying pressure to a certain point underneath the ThinkPad (on A30s its below the information sticker on the underside). This seems to affect the garbled screen even more, which is highly noticable on a LILO boot screen [http://www.drakonus.dragoncity.net/thinkpad like so].
  
If the screen is exported via VNC the remote screen will also be garbled.
+
If the screen is exported via VNC the remote screen will also be garbled. However, this may not be always so. At least in an A31 screen got garbled right after returning from a text-only session (ctrl-alt-F1, cntr-alt-F7), but the VNC session previously opened on the same machine did not experience any change. The reason for this is the fact that there are two separate channels leaving the GPU: analog for the VGA, and digital for the LCD/DVI. One can completely fail while the other remains fully or partially functional.
  
Most likely this is a problem with the ATI videochip, can people who experience this indicate which chip their machines contains?<br>
+
This problem occurred on some T4x series due to chassis flex.  If the laptop was held by the front of the unit, the weight at the rear would cause significant flex in the structure, which, in turn, caused the motherboard ("planar board") to flex, as well.  The USB ports may be affected, as well as the charging circuitry.  The plastic wrist rest may fracture near the rear the joint between the hard drive and optical drive, increasing the tendency to flex.  The soldering process was improved by the T43 line.
Looking at the effected machines, this is probably an issue specific to the [[ATI Mobility Radeon 7500]]
+
 
 +
The ThinkPads affected by this type of GPU problem - in the order of appearance - are as follows:
 +
 
 +
A30, A30p
 +
 
 +
A31, A31p
 +
 
 +
R40
 +
 
 +
R50, R50p, R51
 +
 
 +
T30
 +
 
 +
T40, T40p
 +
 
 +
T41, T41p
 +
 
 +
T42*, T42p*
 +
 
 +
The issue in question appears to have at least two possible causes, one of them being the chassis flex mentioned above which doesn't appear to be affecting the A3x/R40/T30 units. The second one is thermal cycling (repeated heating/cycles of the BGA chips) leading to solder connections - especially when dealing with lead-free solder - cracking. Re-balling the solder joints is the proper solution to this problem, while simple re-flowing may or may not present a long-term solution depending on a number of circumstances.
 +
 
 +
Leaving the machine on at all times and only turning the LCD off when not used has been suggested by many long-time ThinkPad users as one of the ways to minimize thermal cycling and prolong the life of the system. Putting the laptop to sleep should be avoided unless absolutely necessary as well.
 +
 
 +
* Late model (2005/2006) T42/p systems were produced using the same improved soldering technique that was applied to the T43/p units. While not completely failure-proof, the "dot" soldering does minimize the chances of BGA chips coming off the board. The downside to it is that it makes it more difficult - and in some cases impossible - to perform a proper re-balling procedure on the affected planar.
 +
 
 +
The list of ATi graphic chips possibly affected by these issues is as follows:
 +
 
 +
Radeon 7000, 7500, 9000, 9600
 +
 
 +
Fire GL 7800
 +
 
 +
Fire GL T2
 +
 
 +
A GPU failure will affect the video portion of the system, and often result in a non-booting machine with a black screen or a garbled one.
 +
 
 +
SouthBridge chip failure - very common on T43/p units as well as on all the other T4x systems, but not on A3x, R40 or T30 - can affect USB ports, keyboard, wi-fi as well as machine's ability to recognize a hard drive in the main bay.
  
 
==Affected Models==
 
==Affected Models==
*ThinkPad {{T40}}, {{T41}}, {{T42}}, {{R40}}, {{A30}}, {{A31}}, {{A31p}}
+
*ThinkPad {{T40}}, {{T40p}}, {{T41}}, {{T41p}}, {{T42}}, {{T42p}}, {{R32}}, {{R40}}, {{R50p}}, {{R51}}, {{A30}}, {{A30p}}, {{A31}}, {{A31p}} and to a lesser extent {{T43}} and {{T43p}}
  
 
==Affected Operating Systems==
 
==Affected Operating Systems==
Line 23: Line 58:
  
 
==Solutions==
 
==Solutions==
You can have IBM fix the problem if your ThinkPad is still in warranty.
+
* You can have IBM fix the problem if your ThinkPad is still in warranty.
 +
 
 +
* One reported workaround is suspending to ram after powering on and leaving it on power. This way the screen might still be fine after wakeup. The moment when you cold boot again, keep the laptop at the garbled boot screen for about 5 minutes, then do a normal reboot and press your thumbs.
 +
 
 +
* The problem can also be due to bad contact on screen and/or keyboard connectors on motherboard. Try pulling out the keyboard and pushing slightly the connectors, the screen should display again correctly. If yes try puting a foam hold over the connectors and pull back the keyboard.
 +
 
 +
* It may help to tighten the screws around the graphics chipset heatsink, or replace its thermal pad.
 +
 
 +
* Depending on the ThinkPad model, running the laptop on only battery power should reduce the garbled/corrupt effect. Due to the lowered processor speed, it should generate less heat which should reduce the corruption. It helps to have a high-charge capacity battery to prevent it from happening.
 +
 
 +
* In some models (At least the T40 with 9000 pro) the problem dissapear if you make an underclock. The default clock of the Mobility 9000 pro is 250mhz in the core and 200mhz in the memory. If you lower the core clock to 100mhz the problem dissapear. In Linux you can use a tool called rovclock [http://www.hasw.net/linux/] to make the underclock.  (Example for a Mobility 9000: rovclock -c 100 -m 200)
 +
 
 +
* Workaround: For X use the VESA driver instead of the radeon of framebuffer driver. For the virtual console the framebuffer doesn't show this problem either, see [[Talk:Problem with garbled screen#VESA Driver]]
 +
 
 +
* Workaround: Create a skeleton xorg.conf file with ''Xorg -configure'' and set the following line in xorg.conf '' #Option "NoAccel"  # [<bool>] '' to '' Option "NoAccel" # ["False"] '' It worked with Ubuntu 9.04 on a R50p and it allows for the screen to run in 1600x1200, which I could not get the VESA driver to do.
 +
 
 +
I recently experienced this problem; in my case it sometimes also failed to boot at all and produced one-long-two-short-beeps. 
  
One reported workaround is suspending to ram after powering on and leaving it on power. This way the screen might still be fine after wakeup.
+
My diagnosis was that one or more of the pins on the ATI Radeon chip had come unsoldered, such that it would make contact if the laptop was pressed or twisted a certain way, but not others.  I dismantled the laptop until I could reach the chip (it's under the fan assembly, so you can leave things like the PCMCIA assembly and right ultrabay on). Pressing on different corners of the chip made it work or fail. It's a surface-mount chip with the pins underneath (BGA?), so you can't resolder it from the top OR bottom.  I had to make a custom tip for my heat gun by bending some aluminum to a square slightly larger than the chip (1.25" square).  15 seconds at the 1000F setting successfully resoldered the chip for me.  We'll see about long-term reliability.  I hope you find this useful.  Obviously I would not do this on a computer which was still under warranty, but for a computer which is not, replacing the entire system board (IBM's procedure for fixing this problem) is simply not economical.
The moment when you cold boot again, keep the laptop at the garbled boot screen for about 5 minutes, then do a normal reboot and press your thumbs.
 
  
The problem can also be due to bad contact on screen and/or keyboard connectors on motherboard.
+
---
Try pulling out the keyboard and pushing slightly the connectors, the screen should display again correctly. If yes try puting a foam hold over the connectors and pull back the keyboard.
 
  
Depending on the ThinkPad model, running the laptop on only battery power should reduce the garbled/corrupt effect. Due to the lowered processor speed, it should generate less heat which should reduce the corruption. It helps to have a high-charge capacity battery to prevent it from happening.
+
I experienced similar problems with a R50p model.
 +
After some investigation I found out that the bolts holding down the cooler of the graphic chip were a bit loose.
 +
There are 4 bolts, two inside the case, two are fixating it from outside on the back of the case. One of the back bolts also fixates the keyboard.
 +
I tightened those 4 bolts rather hard, but not insanely hard.
 +
Because I was about to tighten those 2 inner bolts, I also tightened all the others inside the case.
 +
To avoid possible tensions, I tightened the other outer case bolts rather slightly.
  
UPDATE: In some models (At least the T40 with 9000 pro) the problem dissapear if you make an underclock. The default clock of the mobility 9000 pro is 250mhz in the core and 200mhz in the memory. if you lower the core clock to 100mhz the problem dissapear. In linux you can use a tool called rovclock to make the underclock.
+
Also see [[Problem with video related system lockup II]]
  
Instructions for gentoo and the mobility 9000:
+
---------------------
  
1) Download rovclock from this website:
+
This is a problem with the VGA chip - It is also a BGA chip which means that there are no pins that go through to the other side of the board like the good old days! - You can try and re-heat the BGA chip on its surface with a paint stripper gun for a minute you will more than likely see the chip drop slightly because of poor original production it wasnt sitting propwerly when flowed - the stripper gun sounds harsh but it works - you can get an adaptor in the shape of a straw made out of steel for the end of the gun and I would strongly recomend using it or you might melt some of the plastic connectors around it - I have been an engineer on thinkpads for 12 years so you can take my word on anything I write...
http://www.hasw.net/linux/
 
2) Compile it
 
3) Copy the binary to /usr/local/bin
 
4) Put these lines in your /etc/conf.d/local.start
 
  
cd /usr/local/bin
 
rovclock -c 100 -m 200
 
  
Sorry for my bad english, im only speak spanish
+
----
 +
After reading that this issue is related to excess heat, I investigated power management under Ubuntu Linux. Ubuntu users might want to look [http://ubuntuguide.org/wiki/Ubuntu:Feisty#How_to_enable_your_CPU.27s_Power_Saving.2FFrequency_Scaling_features here] for details on how to enable proper power management under 7.04 (Feisty). After making the suggested modifications my T41 runs significantly cooler.

Latest revision as of 16:40, 16 November 2016

Information about the problem of a randomly garbled screen.

Problem description

The symptom is a totally garbled screen (on the internal display) as seen in these pictures.

This happens directly after starting up: even the BIOS splash screen is unreadable. The screen can stay garbled for a number of boots and then (seemingly random) it will be just fine right from the start.

It has furthermore been reported that there might be a relation between the temperature of the ThinkPad and the garbled screen, so that the display starts working fine when the ThinkPad has reached a certain temperature (like after 5 minutes of being powered on). Also by applying pressure to a certain point underneath the ThinkPad (on A30s its below the information sticker on the underside). This seems to affect the garbled screen even more, which is highly noticable on a LILO boot screen like so.

If the screen is exported via VNC the remote screen will also be garbled. However, this may not be always so. At least in an A31 screen got garbled right after returning from a text-only session (ctrl-alt-F1, cntr-alt-F7), but the VNC session previously opened on the same machine did not experience any change. The reason for this is the fact that there are two separate channels leaving the GPU: analog for the VGA, and digital for the LCD/DVI. One can completely fail while the other remains fully or partially functional.

This problem occurred on some T4x series due to chassis flex. If the laptop was held by the front of the unit, the weight at the rear would cause significant flex in the structure, which, in turn, caused the motherboard ("planar board") to flex, as well. The USB ports may be affected, as well as the charging circuitry. The plastic wrist rest may fracture near the rear the joint between the hard drive and optical drive, increasing the tendency to flex. The soldering process was improved by the T43 line.

The ThinkPads affected by this type of GPU problem - in the order of appearance - are as follows:

A30, A30p

A31, A31p

R40

R50, R50p, R51

T30

T40, T40p

T41, T41p

T42*, T42p*

The issue in question appears to have at least two possible causes, one of them being the chassis flex mentioned above which doesn't appear to be affecting the A3x/R40/T30 units. The second one is thermal cycling (repeated heating/cycles of the BGA chips) leading to solder connections - especially when dealing with lead-free solder - cracking. Re-balling the solder joints is the proper solution to this problem, while simple re-flowing may or may not present a long-term solution depending on a number of circumstances.

Leaving the machine on at all times and only turning the LCD off when not used has been suggested by many long-time ThinkPad users as one of the ways to minimize thermal cycling and prolong the life of the system. Putting the laptop to sleep should be avoided unless absolutely necessary as well.

  • Late model (2005/2006) T42/p systems were produced using the same improved soldering technique that was applied to the T43/p units. While not completely failure-proof, the "dot" soldering does minimize the chances of BGA chips coming off the board. The downside to it is that it makes it more difficult - and in some cases impossible - to perform a proper re-balling procedure on the affected planar.

The list of ATi graphic chips possibly affected by these issues is as follows:

Radeon 7000, 7500, 9000, 9600

Fire GL 7800

Fire GL T2

A GPU failure will affect the video portion of the system, and often result in a non-booting machine with a black screen or a garbled one.

SouthBridge chip failure - very common on T43/p units as well as on all the other T4x systems, but not on A3x, R40 or T30 - can affect USB ports, keyboard, wi-fi as well as machine's ability to recognize a hard drive in the main bay.

Affected Models

Affected Operating Systems

  • all

Status

It is probably a problem of the graphics circuitry. In any case, it's a hardware problem and warranty will apply.

Solutions

  • You can have IBM fix the problem if your ThinkPad is still in warranty.
  • One reported workaround is suspending to ram after powering on and leaving it on power. This way the screen might still be fine after wakeup. The moment when you cold boot again, keep the laptop at the garbled boot screen for about 5 minutes, then do a normal reboot and press your thumbs.
  • The problem can also be due to bad contact on screen and/or keyboard connectors on motherboard. Try pulling out the keyboard and pushing slightly the connectors, the screen should display again correctly. If yes try puting a foam hold over the connectors and pull back the keyboard.
  • It may help to tighten the screws around the graphics chipset heatsink, or replace its thermal pad.
  • Depending on the ThinkPad model, running the laptop on only battery power should reduce the garbled/corrupt effect. Due to the lowered processor speed, it should generate less heat which should reduce the corruption. It helps to have a high-charge capacity battery to prevent it from happening.
  • In some models (At least the T40 with 9000 pro) the problem dissapear if you make an underclock. The default clock of the Mobility 9000 pro is 250mhz in the core and 200mhz in the memory. If you lower the core clock to 100mhz the problem dissapear. In Linux you can use a tool called rovclock [1] to make the underclock. (Example for a Mobility 9000: rovclock -c 100 -m 200)
  • Workaround: Create a skeleton xorg.conf file with Xorg -configure and set the following line in xorg.conf #Option "NoAccel" # [<bool>] to Option "NoAccel" # ["False"] It worked with Ubuntu 9.04 on a R50p and it allows for the screen to run in 1600x1200, which I could not get the VESA driver to do.

I recently experienced this problem; in my case it sometimes also failed to boot at all and produced one-long-two-short-beeps.

My diagnosis was that one or more of the pins on the ATI Radeon chip had come unsoldered, such that it would make contact if the laptop was pressed or twisted a certain way, but not others. I dismantled the laptop until I could reach the chip (it's under the fan assembly, so you can leave things like the PCMCIA assembly and right ultrabay on). Pressing on different corners of the chip made it work or fail. It's a surface-mount chip with the pins underneath (BGA?), so you can't resolder it from the top OR bottom. I had to make a custom tip for my heat gun by bending some aluminum to a square slightly larger than the chip (1.25" square). 15 seconds at the 1000F setting successfully resoldered the chip for me. We'll see about long-term reliability. I hope you find this useful. Obviously I would not do this on a computer which was still under warranty, but for a computer which is not, replacing the entire system board (IBM's procedure for fixing this problem) is simply not economical.

---

I experienced similar problems with a R50p model. After some investigation I found out that the bolts holding down the cooler of the graphic chip were a bit loose. There are 4 bolts, two inside the case, two are fixating it from outside on the back of the case. One of the back bolts also fixates the keyboard. I tightened those 4 bolts rather hard, but not insanely hard. Because I was about to tighten those 2 inner bolts, I also tightened all the others inside the case. To avoid possible tensions, I tightened the other outer case bolts rather slightly.

Also see Problem with video related system lockup II


This is a problem with the VGA chip - It is also a BGA chip which means that there are no pins that go through to the other side of the board like the good old days! - You can try and re-heat the BGA chip on its surface with a paint stripper gun for a minute you will more than likely see the chip drop slightly because of poor original production it wasnt sitting propwerly when flowed - the stripper gun sounds harsh but it works - you can get an adaptor in the shape of a straw made out of steel for the end of the gun and I would strongly recomend using it or you might melt some of the plastic connectors around it - I have been an engineer on thinkpads for 12 years so you can take my word on anything I write...



After reading that this issue is related to excess heat, I investigated power management under Ubuntu Linux. Ubuntu users might want to look here for details on how to enable proper power management under 7.04 (Feisty). After making the suggested modifications my T41 runs significantly cooler.