Thinkpads Forum

Here's some interesting and important information concerning low density 256MB PC100 SODIMM:

Not all 8 chip modules are "high density"!

Said another way, there is at least one 256MB PC100 SODIMM that only has eight devices which can successfully be used inside a 600E, 600X, T21, A22 and probably any other "low density" system.

This thread: http://forum.thinkpads.com/viewtopic.php?t=67488 got me thinking about the claims of the eBay seller "1-800-4-memory". This seller has been mentioned here several times as being a good source for memory. They are offering this:

33L3069 modules

If you look at the picture they have it shows a module that has 4 chips on one side (and one can safely assume that it has 4 more on the other side) and a label on it which says "256MB SDRAM PC100 144pin LOW DENSITY". There's also a compatible Thinkpad systems list which includes PC100 equipment. Since they stated that they guaranteed that the module would work inside these systems and referenced the correct IBM FRU (33L3069), I bought one at $30. That's a little bit high in my book since I'm cheap, but this was a science experiment .

Arrived in three days in a small padded envelope via USPS First Class with Delivery Confirmation (shipping was "free"). The module I received looked exactly like the picture in the auction description. Put it inside a T21, booted a memtest86+ CD and it registered 256MB and passed! Installed the module in an A22p and it worked there as well. Even when combined with a low density 128MB or 256MB module, everything passes. Putting the module into either slot of a 600E (2645-8AU) and running the Easy-Setup memory test, it fails with the memory error code of:

DEV 002
ERR 83
FRU 2010

However, when booted to a memtest86+ diskette, it recognizes 288MB (256MB + 32 on board) and passes (!!??! - still trying to figure this one out ). Details about the module and my testing are here:

http://www.kawakami-ca.com/8_chip_low_d ... mb_sodimm/

Currently there's only the pictures I've taken of the module and memtest86+ test results. I'm planning to take the module to work on Monday and use the bench-top tester to get some more information about it. For now, these are my assumptions:

- the eight parts are 16M x 16 devices; i.e., 16M addresses and 16 data pins or 32MBytes total
- the refresh rate is 4,096 cycles every 64mS
- CAS latency might support CL2, based upon the parts being marked for 133Mhz (-75)

I've tried to get the datasheet on the individual memory component but so far have come up empty. They appear to be sourced from SpecTek, a company owned by Micron:

http://www.spectek.com/

but the part numbering system has changed a couple of times in recent history and the current part numbering system does not cover this particular device (S160032LLLTK-75A). The closest that I can get says the parts are 32M x 16. However, if there are 8 of those on the module, then the total memory size would be 512MB! And at the same time, the part number suggests that these are 8K refresh devices; what I would consider high density. Compounding my confusion, an apparent date code on the packages would lead me to believe that these parts were manufactured around June of this year.

Thanks for the detective work Ray. It's always nice to have an alternative source for RAM on these older machines.

Thanks for the test, I just bought one for $25.

I've been looking for a 256MB for my X20 (2662). I've tried two, both are 8 chips, both show 128MB only.

It is ridiculous to spend $50 for 256MB for a laptop I bought for only $100. Hopefully this $25 one will work.

Got the 256MB RAM today, put it into my X20 and it works like a charm. Yes it is 8 chips total.

Okay, I finally got my web site updated with the results of the bench-top tester analysis.

Here's the short scoop: It's really a 512MB module that's been modified to identify itself as a low density 256MB version.

The long story is on the web site. Hope your eyes don't glaze over .

Wow, I never thought that, even though I've tried two 256MB that appear to be 128MB only. So, we might just buy a 512MB!

rkawakami wrote:Okay, I finally got my web site updated with the results of the bench-top tester analysis.

Here's the short scoop: It's really a 512MB module that's been modified to identify itself as a low density 256MB version.

The long story is on the web site. Hope your eyes don't glaze over .

Hi, Ray, good work ! but... where in your web site is the story ? I had a look to the site, but couldn't find the ref.story....

Rgds.

The site: http://www.kawakami-ca.com/8_chip_low_d ... mb_sodimm/

The long story I'm referring to is the 7 screen captures from the tester I used and the detailed explanations of what I found stored in the SPD bytes. If you had previously visited that site but only see the 10 photos of the module and memtest86+ results, then force a refresh in your browser as you may be seeing a cached version.

A quick summary of the long story: A couple of the SPD bytes have been altered so that the 512MB high density module "looks like" a 256MB low density version.