Added 1GB of memory, problem occured

[krizze]

Damn, guess I should have waited, and read this thread Added a kingston 1gb module, it worked for around a week, but now my A1 is dead.
viewtopic.php?f=24&t=903

[SNOboy]

after doing some reading on the atom cpu i think alot of these problems are from using 667 ram, the cpu is a new breed and has alot of low power commands. timing is a big issue with this cpu and running on a fine line, thats how they get away with fanless cpu's. The fan on my intel board that uses the atom 250 1.6mhz cpu is actually on the northbridge chip not on the cpu i will have to open up the One to see if its the same which i think it is. they clearly state that the required ram is 533mhz and not anymore. I think the chipset does not drop the 667 ram speed down to the lower 533mhz. Also if you look at the specs of the atom cpu you will see it has even more thermo protection than normal cpu's and will shut down cpu to protect at high temp or if there is a problem on the fsb.

This is from Intel
The processor has only one signaling mode, where the data and address busses and
the strobe signals are operating in GTL mode. The reason to use GTL is to improve
signal integrity.

The processor supports two kinds of signalling protocol: Complementary Metal Oxide
Semiconductor (CMOS), and Advanced Gunning Transceiver Logic (AGTL+). For FSB
data and address bus, only AGTL+ is used.
The termination voltage level for the processor CMOS and AGTL+ signals is
VTT = 1.10 V (nominal).

***Due to speed improvements to data and address bus, signal
integrity and platform design methods have become more critical than with previous
processor families.***

The CMOS sideband signals are listed in Table 6.
The AGTL+ inputs, including the sideband signals listed in Table 6, require a reference
voltage (GTLREF) that is used by the receivers to determine if a signal is a logical 0 or
a logical 1. GTLREF must be generated on the system board. Termination resistors are
provided on the processor silicon and are terminated to its I/O voltage (VTT). The
appropriate chipset will also provide on-die termination; thus, eliminating the need to
terminate the bus on the system board for most AGTL+ signals.
The AGTL+ bus depends on incident wave switching. Timing calculations for AGTL+
signals are based on flight time as opposed to capacitive deratings. Analog signal
simulation of the FSB, including trace lengths, is highly recommended when designing
a system. Just my 2 cents, SNO

[number78]

Well, I tried Kingston 667Mhz yesterday. I got one pseudo-boot, but in the bios, the top line of every field flashed on & off. So I reseated RAM, put back together, and A1 wouldn't boot. Just fan on, fan off. Took out RAM, worked. Put RAM back in, didn't work.

So i'm getting some A-Data 533Mhz soon, and try that. But I can definetly say Kingston 667 Didn't Work!

(Also PXE comes up if no other boot device found, I.e. SSD is unplugged)

[iDq]

Here are the specifications of the two memory models from Kingston:

* 533 MHz -> http://www.valueram.com/datasheets/KVR533D2S4_1G.pdf
* 667 MHz -> http://www.valueram.com/datasheets/KVR667D2S5_1G.pdf

Note the diference at the power consumption.

Perhaps, the power consumption of the 667 MHz memory (2665 mW) can cause this issue. Or, perhaps, the installation of the 667 MHz memory damages the onboard 512 Mb memory.

[iDq]

the mobo doesnt underclock 667mhz memory and loads of people have had problems with kingston stuff.

my recommendation would be crucial or corsair 533mhz

A 667 MHz memory should work at 533 MHz (FSB speed). I think.

[glibdud]

Here are the specifications of the two memory models from Kingston:

* 533 MHz -> http://www.valueram.com/datasheets/KVR533D2S4_1G.pdf
* 667 MHz -> http://www.valueram.com/datasheets/KVR667D2S5_1G.pdf

Note the diference at the power consumption.

Perhaps, the power consumption of the 667 MHz memory (2665 mW) can cause this issue. Or, perhaps, the installation of the 667 MHz memory damages the onboard 512 Mb memory.

I'm sure those numbers assume that the RAM is running at full speed. Since it would be running at 533MHz in the A1, its power consumption will likely be closer to that of the 533 stick in practice.

[iDq]

Here are the specifications of the two memory models from Kingston:

* 533 MHz -> http://www.valueram.com/datasheets/KVR533D2S4_1G.pdf
* 667 MHz -> http://www.valueram.com/datasheets/KVR667D2S5_1G.pdf

Note the diference at the power consumption.

Perhaps, the power consumption of the 667 MHz memory (2665 mW) can cause this issue. Or, perhaps, the installation of the 667 MHz memory damages the onboard 512 Mb memory.

I'm sure those numbers assume that the RAM is running at full speed. Since it would be running at 533MHz in the A1, its power consumption will likely be closer to that of the 533 stick in practice.

Not necessarily, Ithink, different memory chips -> differents power requeriments (an absurd but clarifier example [I think]: you can set the frequency of a Pentium III to 25 MHz and the power consumption will be different than a Pentium II operating at same frequency).

Samsung modules operating at the same frequency have different power requeriments. You can check it here:

http://www.samsung.com/global/system/business/semiconductor/product/2007/6/11/DDR_DDR2/DDR2SDRAM/Module/SODIMM/M470T2953EZ3/ds_512mb_e_die_based_sodimm_rev14.pdf

Note: obviously all the modules works under the same voltage: 1.8 v (DDR-II spec.)

EDIT: I also think that the power consumption will likely be closer to the 533 module.

[glibdud]

Not necessarily, Ithink, different memory chips -> differents power requeriments (an absurd but clarifier example [I think]: you can set the frequency of a Pentium III to 25 MHz and the power consumption will be different than a Pentium II operating at same frequency).

Samsung modules operating at the same frequency have different power requeriments. You can check it here:

http://www.samsung.com/global/system/business/semiconductor/product/2007/6/11/DDR_DDR2/DDR2SDRAM/Module/SODIMM/M470T2953EZ3/ds_512mb_e_die_based_sodimm_rev14.pdf

Note: obviously all the modules works under the same voltage: 1.8 v (DDR-II spec.)

EDIT: I also think that the power consumption will likely be closer to the 533 module.

Sure, they're not going to be exact, but judging by the fact that the model numbers are so similar, I'm sure the chips used are very similar as well.

Years ago I read that often, different speeds of RAM from the same manufacturer might contain the same electronics... the difference being that one run might be more stable at high speeds than another due to tolerances, so they test and label each stick accordingly. Don't know if that's true anymore (or really if it was ever true in the first place), but that's sort of what I'm basing my guess on.

Sounds like we mostly agree on the conclusion, though.

[Thomas8675309]

For what it's worth, I installed a stick of Kingston 1GB 200-Pin DDR2 SO-DIMM DDR2 533 (PC2 4200) Laptop Memory (about $22 from Newegg), and it has worked a perfectly ever since.

I was discouraged to buy Kingston. It might work but it could damage your board. As to how, the tech guy didn't give specifics.

I find it very hard to believe that Kingston RAM could damage my motherboard. Nonetheless, consider me your guinea pig. No problems to date, but if I have any in the future, I will report back.

Just an update - I've now been using my Aspire One with the extra Kingston 1GB DDR2 533 stick installed for over a month with pretty heavy use. Not a single problem so far. Price is now down below $20 in the U.S.

[thewtex]

I had the same problem, and I purchased Kingston memory, so I was pretty discouraged when I came across this this thread. However, I reseated the SSD ribbon, which is quite tricky, and it seems to be working fine.