T21 Memory dead after hibernation

[gaosan]

Hi, I got a T21 from a friend with XP installed. T21 comes with two RAMs, tatal 512M. I then installed XP patches and thinkpad patches download from IBM to enable Fn keys. After that, I run disk defrag, then went to bed.

Second day I found my T21 won' boot with famous 1-3-3-1 beeps. I googled and then found this forum. I realized it was memory problem. So I tried to unplug one RAM, same beeps. Then tried another RAM, my T21 boot as normal. I test the bad RAM on my dell notebook. The RAM is dead!!

I have no idea on what happened. So I added another RAM into T21( I have extra one from my Dell). So there are two RAM installed on T21. Tried several times to boot and power up/down with AC and battery connected. Every time it boot into XP, everything seems fine. Until I used Fn+F12, it went to hibernate mode.

Then power up T21, I got same 1-3-3-1 beeps again. I test RAMs with my dell computer again, and found one RAM dead again.

Why hibernation caused RAM dead ? I can not recall whether both dead RAM were from same memory slots or not? I believe it comes from the slot close to PCI. But I am not 100 percent sure.

Now I only have one 256M RAM. I use the slot close to front. I can power T21 up, boot to XP without any problem, then power down. Tried several times, there is no problem, but I won't try hibernation again until I know the reason. I use AC power only this time.

Any idea ?

[rkawakami]

Welcome to thinkpads.com!

Interesting problem... about the only two things that I can think of (neither of which should happen) is that the laptop is destroying some of the pins on the module or that the information about the configuration of the module is being altered. The first problem could occur if there was a defect on the motherboard which caused an overvoltage to be placed onto one or more of the memory module pins. The other thing that may be happening is that the Serial Presence Detect (SPD) information is accidentally being re-written. The SPD contains data on how the module is configured and what its capabilities are. This is read by the motherboard during power on. If the motherboard does not see the correct data in the SPD, it will complain (the 1-3-3-1 beep code). I have been able to confirm this happening with some other user's modules before:

http://forum.thinkpads.com/viewtopic.ph ... 816#200816
http://forum.thinkpads.com/viewtopic.ph ... 151#232151

I don't know if hibernation was used prior to those modules going bad. I suppose it could be possible that during the sequencing of the power-down the SPD is being disturbed, but again, that should never happen. What could help is knowning what your BIOS and Embedded Controller versions are. Boot the system with your remaining good module, access the BIOS (press and hold the F1 key down at power on) and read that info from the main BIOS screen.

If your system is indeed causing the SPD to be changed, then the modules may be recoverable. It would normally take a memory tester to do so. I don't know of any way that you could do this with a regular laptop. As you are fairly far away, it might not be cost-effective for you to ship the modules to me. If you can find a local computer shop which has a "bench-top" memory module tester, they might be able to confirm that this is what is happening.

[gaosan]

Wow, your reply is very informative. Thank you!! Ray.

I think it is unlikely hibernation cause extra turbulence in voltage which damage memory stick. I think it is likely SPD was changed for some reason. However I do not have memory tester to verify it. But why it is one slot ? not both ?

My two dead sticks are :
Crucual CR32M64s4w7e.16FG which is 256M PC133 CL2 and
Korea m464S3323BN0-L1L 256M
They are both decent sticks so I ruled out stick problem.

I did some research on SPD and realized that SPD is stored on a small serial EPROM on I2C bus. So any I2C programmer should be able to read it and write it. I have done some I2C test before at my work. I saw sometime EPROM were corrupted mysterious due to protocol violation or pull up register problem on SCL / SDA.

Does any one can confirm that both RAMs are connected from same I2C master ? I am curious on why only ONE slot damaged at a time, not BOTH.

I have managed to measure the resistance between PIN 144- PIN 141 (SDA) and PIN 144 - PIN 142(SCL) for two slots, values from 2.17K to 2.20K. According to I2C bus, these values are okay.

I could only measure the resistance from socket pins when the stick removed ( bottom side ), and then I insert a stick so I can measure it from the stick fingers (top side). That means the bottom side connection can not be verified. Measured resistance are ranging from 2.17K to 2.20K. According to

http://www.robot-electronics.co.uk/htm/ ... 2c_bus.htm

Those values are good.


Thanks for viewing.