Reflow.IT – Toshiba Satellite Pro P6100 with graphics memory errors

I had this Toshiba Satellite Pro P6100 laptop sitting around. It had been tested some time ago, and it had errors in the video graphics memory. I had used VMT which is a Video Memory Stress Tool, to run diagnostics on the video graphics memory .


This is what the screen looks like. What we see at the top of the screen with the greenish lines are artifacts on the display due to these memory errors. Now the memory errors are not stable, they appear at different addresses and affect different bits, so it could be a graphics processor problem or a memory problem. This laptop had a separate card which contains the graphics processor and its two memory chips but also interfaces to the keyboard, mouse pointer buttons and the system board. Due to my previous success with using my SMD Rework station to reflow a laptop system board, I thought I should try this again.


I removed the graphics add-on board, and taped up the areas that I did not want to heat up. I also attached a thermocouple temperature probe, just like I had done previously.

I worked my way around the board, heating the three chips gradually, and then concentrated first on the graphics processor, and then afterwards on the two memory chips. After letting it cool down, I then reassembled the laptop and was rewarded by a clean display with no signs of artifacts.


Another successful “Reflow.IT“. I suspected that due to the nature of the problem where the memory errors occurred at different addresses and different bits that it was an intermittent connection problem with the chips. The reflow helps to melt the solder and let it solidify “hopefully” making good contact on all the chip connection points. It looks like it worked. Now what do I do with a laptop with a Pentium 4 and 512MB of ram. It runs Windows XP and has a 40GB disk drive and has a serial port on it, which is quite sadly lacking from most modern day laptops.

Repair.IT – Acer Aspire S3-951 laptop disk errors

This slimline Acer Aspire S3-951 laptop came in the other day. It was having problems booting in Windows 7. Windows would want to launch Startup Repair, but this would fail, and then it would reboot and the same process happens again and again. It appears that the internal disk storage has problems. Judging from the thin laptop I asked whether it had a SSD, a Solid State Drive – the answer was don’t know, ok. The only way to find out is to open it up.

Sure enough, after opening up, I see the sata disk is a SSD – but because I can only see the bottom of the SSD, I don’t know what brand and model it is. Eventually I managed to remove it after undoing a few screws, and moving wireless cables about and find that it is a Micron RealSSD C400 256GB. Usually the SSD’s that are commonly available are 120GB, 240GB and 250GB. The 256GB SSD’s are a fair bit more expensive, over $200 in most cases, whereas a 120GB can be had for about $80. Compare this with a slimline sata hard disk with 500GB capacity that can be bought for $65.

The SSD was put onto my test computer which runs Ubuntu Linux and allows me to make a raw disk image of the data – the first step that I usually do, is to make an image so that if it does fail sometime in the future, at least I have a copy of what was readable. Sure enough, during the copy there were numerous errors – each being a bad sector / memory location. 5,824 bad sectors actually – which is quite a lot.

The next thing to do is to find out whether or not the disk image is usable – so it was copied to a larger hard disk. Now this disk is thicker, i.e. 10mm whereas the SSD was 7mm – so it doesn’t fit into the laptop properly, so I had to run it without putting all the covers back on. I managed to get Windows to boot normally, which came up to a login screen. I didn’t have the password, nor did I have the power adapter so this would have to wait until both are available.

Last night, the laptop’s owner came by with the power adapter and the password. I was able to log in, and of course charge the battery. The owner did not have recovery disks, so I connected a USB DVD-Writer, and ran the Acer eRecovery Manager to make the recovery disks, 4 DVD’s actually. Once that was done, I made a Windows Repair Disk (a CD) just in case I had later problems.

Now, if I wanted to move this to a smaller SSD, there are two ways. The first way is to shrink the system partition to make it smaller, and then move the hibernation partition, so that everything fits at the beginning of the disk, such that it could be copied to a smaller disk. I checked in Disk Management, that the system partition could only be shrunk by about 80GB, not enough to fit everything onto a 120GB SSD, however I could put it onto a 240GB SSD. Anyway, it was past midnight, so I had to leave it.

This morning, I considered the options. The second way is to get a smaller SSD, then use the Recovery disks and install a new Windows on the smaller SSD, then backup the old Windows partition and restore it onto the new SSD. It sounds good in theory but not in real life, with many people having problems with doing this. If it was Windows NT – it would have been simple, but this is Windows 7. While checking prices of 256GB SSD’s – all of them being $212 or higher, at least those that were in stock at my usual suppliers, I then thought to check Umart – they had a Transcend 256GB SSD for a special internet price of $135 and it showed as in stock. Fantastic, I quickly placed an order for it – and when I got an email that it was ready – drove down to West Ryde and picked it up.

Back to my test computer, the new SSD was connected and I set it up to copy the disk image back to this SSD – which is the same size as the original SSD – no need to make any changes at all. Some hours later (it takes time to copy 256GB of data), it had completed, so I quickly installed the new SSD back into the laptop. Powered it up and told Windows to start normally and I could log in. I enabled the disk check on startup, then rebooted and let Windows do the chkdsk, which found a few problems which it corrected then automatically restarted. Now that is all for this “Repair.IT” – oh, not yet, a quick photo.


Here is the old Micron SSD sitting on top of the laptop. I notice that the firmware on the SSD is 0009 – which is quite old. There were a number of firmware updates to address various issues – maybe this might have contributed to the early failure of the SSD. Better check my other computers to make sure that the firmware is up to date – I only have one SSD, but my sons both have SSD’s in their computers. Firmware updates can address wear issues, encryption problems and basically anything, so it is best to keep the firmware updated if possible. However, the vendor generally suggests making a backup before doing the firmware upgrade. When is the last time you made a backup of your computer?

Review.IT – Asus Nexus 7 tablet slow charging problem – confirmed fixed.

Here is an update on the Asus Nexus 7 with the slow and intermittent charging problem. Previously I had replaced the USB Charging Connector Flex Cable. Yesterday, while I was out and about, I used the tablet and let it discharge until last night it got to 10%.

I put it on charge and then this morning, I checked the Battery Monitor Widget.

Nexus 7

This is what I saw on the display. The battery level last night was 10%, and during charge it rose steadily – this is what it should be doing. The previous charge was from 80%, but with this latest charge from 10% to full charge working so well, I can now confidently say that the problem is confirmed to be fixed.

Charging started at about 11:35 PM and was completed by 2:20 AM – under 3 hours. The charging problem used to be that it could take days to charge if it did charge, and other times plugging in the charger would indicate that it was charging but was actually draining the battery. This fix cost just over $7 in parts and is relatively easy to do. I will sometime in the future take a closer look at the flex cable that I had removed to determine what might be wrong with it.

Repair.IT – Asus Nexus 7 slow battery charging problem – possibly fixed!

Previously, I had examined an Asus Nexus 7 tablet to investigate the slow and intermittent battery charging problem.  Many users on the internet have complained on numerous forums about this problem.  There were a number of reported resolutions, one being to update the firmware, others to replace the charger, replace the charging cable and replace the tablet. I had tested the battery which appeared to charge reasonably well, using my special charger. Then this Nexus 7 was left for a while, during which time it sat on the backburner – which is a list of non-urgent jobs.

Anyway, a month ago, I came back to this Nexus 7 after finding some further information that suggested that replacing the “USB Charging Connector Flex Cable” had in some cases fixed this slow charging problem. I checked the price of these on eBay, as we do, and found that I could get one for just over $7 delivered. This was reasonable, and certainly cheap enough to try, so I proceeded to buy it, and in due course, last week – it arrived.

USB Charging Connector Flex Cable

I had earlier removed the battery from the Nexus 7. It is stuck on, with double sided tape in two places – I used an old phone card, one of those cheap calling cards, which is thinner than a credit card, and use it to raise the battery up, and pushed the card in to break the adhesive bond. Doing it in this manner, allows the battery to come out without bending or otherwise damaging it – which could cause undesirable results, like explosions or burning or such like. Lithium batteries should never be punctured and if found to be bulging, should not be used.

Last night, I decided it was time to replace this cable, time to see if I could “Repair.IT“. Inside the Nexus 7, the speakers are first removed, three screws hold it in – two black and one silver one. The speakers can stay connected, just moved out of the way. Next the multi-conductor flex cable was unlatched, by raising a black tab – then removed five screws, two of them very small. This cable was now able to be removed, and the new one installed in much the same manner as the removal.

I connected the battery again, then put the tablet on charge using a 2A USB charger. After a short time, I was able to see that the battery was charging. A few minutes later, it was showing that it would fully charge in 30 minutes. The battery had earlier been retested and had about 80% capacity. As it was close to midnight, I decided to let it charge and I would check it in the morning.

This morning, I was happy to see that the battery was showing full charge at 100%. Using the Battery Monitor Widget, an app I had installed previously, I could see that the charging rate in the graphics was very good and had gone to full charge and stayed there. Before, it used to charge very very slowly and if it ever got to full charge, it would drop afterwards even with the charger connected. This new behaviour is definitely an improvement and is more like what it should be doing. Time will tell if this is now fixed, but it does appear to be fixed. I will use the tablet throughout the day and charge it each night and if it continues to charge – then I will be confident of saying that this Nexus 7 slow and intermittent charging problem has been resolved.

Research.IT – Samsung Galaxy S III GT-I9300 not booting

My brother sent me his Samsung Galaxy S III, a 16GB GT-I9300 android phone – it wasn’t booting. Based on his report of the symptoms, I did some research on the internet and concluded that it was most likely a eMMC failure.

When I got the phone, the first thing I did was to plug in my charger. As expected a battery symbol comes up – not expected is that it doesn’t change, normally it will show a battery level going up as it indicates that it is charging.

I tried turning it on, the Samsung screen comes up, then stays there – doesn’t go away.

Ok, that isn’t good. Next I try to go to the Recovery Settings – press and hold volume up, menu and power. The screen will eventually go off, then the Samsung screen comes up then let go.

Now this seems to indicate that the phone is unable to access partitions in its memory.

/efs, /system, /cache, /data are separate partitions in its memory which is laid out as disk storage – if these are not accessible, the phone cannot boot and is essentially now just a brick. Further research on the internet shows up a lot of information that is contradictory in both nature and content. The only promising thing about the research is that I learnt more about this android smartphone environment. There were a few references to try to correct this by erasing the Nand flash memory, however doing so will require access to a pit file, a partition information table which is similar to a MBR in disk terms.

Even finding a pit file was difficult. I was able to find a zip file on the internet that contained not one, but three pit files. Trawling through more forums, I was able to find out some critical information – the pit files are for 16GB, 32GB and 64GB phone versions, so now I was able to identify the correct one to use, the 16GB one – of course.

Another forum page suggested that we try Nand erase then Re-partition, others say never do this. Other pages suggested that it was possible to fix it by doing this and then installing the stock firmware. In preparation, I downloaded both the android 4.1.2 and 4.3 stock firmwares for this phone which was from HK. I also downloaded Odin 3.07 and 3.09 – some sites say to use 3.07, others to use the later one – I also got 3.10 just in case.

Ok, now to put the phone into downloading mode.

Now we press and hold volume down, menu and power.

Then connect the phone to the computer after making sure that the Samsung USB drivers had been installed.

Now run Odin 3.07, load the pit file, select Re-Partition and Nand Erase All, load the PDA firmware which is the phone firmware for 4.1.2


Now the time to click Start… and… Fail – ok, not unexpected.


What it means is that the Nand flash memory is basically not accessible, cannot be erased, or written which means that my original conclusion was valid – the eMMC is dead. The only way to fix this is to replace the memory chip.

I also did try with different Odin versions, as other forum pages suggest, but the result was still the same. One possibility is that the memory chip being a bga chip might have suffered a fate similar to graphics chips in laptops – perhaps one thing that I could do is to heat it up with my smd rework station, the same way that I did on the Dell laptop – maybe that might fix it if the memory chip hasn’t actually failed, so something to try…

Reflow.IT – Dell XPS M1530 Laptop

This Dell XPS M1530 laptop came in a while ago.  I was busy at the time, so it got put on the back burner. Anyway, I was asked about it recently and of course, I hadn’t forgotten about it because I see it every few days, but I had left it – quite a long time.

The problem is that it doesn’t boot, or even get to a bios screen. It sits there with the fan and hard drive running, and just not boot. It will however, do a couple of things. If I press and hold the D button, then press the power button, it will come up and do a diagnostic on the lcd screen. Mainly going through a few different colors, white/grey, red, green, etc.

If I press the Fn button, then press Power, it will go into a diagnostic. It shows three possible lights using the blue keyboard status leds, NumLock, CapsLock & ScrollLock. The left one is flashing, and the middle and right ones are solidly lit.  From the internet, this indicates a CPU Processor fault. Solution is to – 1 Reset the CPU, 2 Replace the CPU, 3 Replace the system board.

Now this laptop had come in previously and I had replaced the system board. This was because one of the heatsink mounting tabs had come off the system board which meant that the heatsink was not making good contact with the graphics chip allowing it to overheat and therefore fail. Replacing the system board again, is no guarantee that the problem won’t happen again, so this time I decided to try a reflow.

Some background information – a number of nVidia graphics chips in laptops had problems whereby they would prematurely fail. It was known to occur in Compaq, HP, Dell and other brands so it wasn’t the brand, but actually the chip. It seems that these chips being a BGA (Ball grid array) had the wrong alloy of solder balls on it, so that when soldered onto the system boards, after some time, these solder balls somehow didn’t make contact. The only permanent solution to this was to remove the graphics chip and install an upgraded chip that did not use that alloy – of course, it is difficult to find out what alloy we need, and certainly, the manufacturers won’t tell us – but that is what we hear.

An alternative is to remove the chip, clean all the solder alloy off it, then reball it, by melting new solder balls to the grid array. I haven’t been successful in doing this because my cheap equipment isn’t up to the job or maybe I don’t have the experience. So plan B, is to perform a reflow. A reflow means to heat up the graphics chip to a high enough temperature, like 250 degrees, to allow the existing solder balls under the BGA chip to melt, and then allow it to cool, and hope (fingers crossed) that it cools down with all of the joints intact. Again, I haven’t successfully performed a reflow as yet, but maybe now I might get lucky.

I removed the bottom memory and cpu cover, removed the memory, then unscrewed the 7 screws that hold the heatsink on, remembering to remove the heatsink fan power cable. I removed a small piece of plastic from near the graphics chip. Then cleaned the thermal material from both the cpu and the graphics chip. The cpu was easy to clean, but the graphics chip has lots of chip capacitors on it, so we don’t want to scrub too hard, and it is often easier to leave the residue, but remove it from the contact surface.

I hooked up a temperature probe – a thermocouple to the graphics chip side, then it was time to “reflow.IT“. I have a smd rework station, which is a hot air machine, set to 300 degrees C, then by heating the graphics chip and surrounding area for a little while, then get closer to the graphics chip and go around and around it. Just basically aim the hot air onto the chip to get it evenly heated. The thermometer would go up when the hot air is in that area, then drop down. I did this for about five minutes gradually getting the chip hotter and hotter – the temperature only got to about 200 or so, but the chip itself was probably hotter than that. Anyway, after I thought it was ready, I stopped and let the chip cool down on its own.

I put some Arctic Silver 5 thermal paste onto the cpu and graphics chip – spread it around with a small piece of plastic (cut from an old credit or loyalty card). Mounted the heatsink, attached the fan cable, installed the memory and put the cover back on. Lastly, which I didn’t mention before, insert the battery. Oh, by the way, I had the laptop open with the keyboard facing down on the work bench when I was doing the heating up – you don’t want the heat to get to the lcd screen otherwise that will be damaged. A better way is to take the system board out of the laptop, but it was easier to leave it in.

Ok, acid test time – open up the laptop, press the power button…  My goodness, I see a bios screen, then shortly it starts booting up – success! My first successful reflow. I realize that I didn’t take a photo. By now, Windows had started and was asking for a password – ok, I messaged the owner for the password. In the meantime, could I try this with the old laptop which I still have lying around?


This time, I took a photo to show the area around the graphics chip. The graphics chip is a nVidia chip, the one on the left. The cpu is the one on the right bottom. To the top left of the graphics chip, there is a hole, where a heatsink mounting nut should be. If this works, I might have to use some epoxy and put that nut back on.

I set the hot air rework station to 350 degrees. Then started heating the graphics chip – this time only for about 4 minutes or so. The temperature probe was showing that when the air was directed at the thermocouple, it was hitting 245 degrees, so this time, it should be hot enough – but of course, don’t have it too hot for too long. Ok, done – leave it to cool down, and check if a password had come through. Yes, it had – enter password and log on.

Success! This laptop comes up with a popup message each time you log in – it is a throwback from when it used to be on a Windows domain. I had a look and found that it was a script running from the Startup folder in the Program groups – deleted it, and this should fix this particular problem.

I left it to reboot, then reassembled the other laptop, put some memory in – didn’t have a hard drive to put in, but won’t need it. Plugged in a battery, then fingers crossed, pressed the power button, and… voila! I get a bios screen – that is two successful reflows today. Ok, now I need to dig out a few more laptops and try this out some more.

[Note] Previously, I had been trying to perform reflows with an infrared rework station, however I could never get the chips hot enough to be able to remove a BGA chip. Then a while ago, I had the need to work on some surface mount boards, to remove chips which needed a hot air rework station – that was when I got this rework station and the infrared station got relegated to the garage.

[Edit] Updated diagnostic light indicators.