View Single Post
  #3  
Old February 17th 14, 06:55 PM posted to microsoft.public.windowsxp.hardware
Paul
external usenet poster
 
Posts: 18,275
Default Fixing inconsistent RAM detection

BillW50 wrote:
On 2/17/2014 4:31 AM, jphilippa wrote:

I have two identical 1Gb RAM chips installed, and they are not always
detected. I have even had both BIOS and WIN say that RAM is 1.5Gb [as
well as 1Gb and 2Gb]. Swapping the chips around seems to fix it for THAT
session, and then what is detected at next reboot is an
adventure/uncertain.


Try cleaning the RAM contacts with a clean pencil eraser. Hopefully that
would take care of it. Don't use other cleaners until you try the eraser
trick first. As this one seems to last the longest.


I generally recommend against this, for a first step.

Contact systems:

Tin on tin, or gold on gold.
Tin on tin, relies on the scraping of protective oxide
off the outside of the metal. This gives a gas-tight connection.

Gold on gold involves precious metals on both sides. At least,
as long as there is some gold left. Gold "slides" against its
gold partner. Friction level is lower. Oxides don't form on gold,
which is why it is used.

Mixing gold and tin, is not recommended.

To clean a RAM slot:

1) Turn off computer power. Unplug, if you are unsure of whether
+5VSB is still running. Wear an antistatic strap. A commercial
antistatic strap, has a series resistor, and brings the chassis
of the computer, to human body potential (so the RAM doesn't
get zapped by static). Clip the strap onto an I/O screw on the
back of the computer.

2) Pull the stick of RAM. Place it in an antistatic bag.
Touch the inside of the bag with your finger, to bring it
to the same potential as the rest of your gear.

3) Using a strong light and magnifying glass, inspect the
socket for bent or broken pins.

4) Inspect the DIMM. Look for surface dirt.

a) If the DIMM and socket are clean, simply reinsert. The
wiping action of the contacts, is sufficient to establish
a connection. In many cases, the DIMM simply was not down
into the socket fully. Verify the cam ejectors are in the
full upright position.

b) If the DIMM is dirty, use isopropyl alcohol and a cloth
to remove the debris.

c) Only under the most dire circumstances would you apply an
abrasive. If the 10u gold plating is completely shot, perhaps
grinding the **** out of it with your pencil eraser is your
only option. Be prepared to have to do this, every few months.
It is habit forming.

*******

The BIOS uses two methods to sense the DIMM.

1) SMBUS, readout of SPD data. The BIOS uses this
as the estimated size. But, it uses a "trust but verify"
philosophy.

2) The second (sequential) step, is "poke and peek sizing".
The BIOS sets up a preliminary mapping, and does safe access
to the DIMM. It writes a value, and checks later whether it
can be read. If an address line is not making contact, it is
this test that catches it. Values for bytes are selected so
that the test resists "floating bus" false results.

It's even possible, for the test in (2), to correct for
an entirely wrong SPD chip. Some DIMMs were made once, where
the wrong SPD was soldered to the DIMM. And yet, the computer
was able to deal with the DIMM all the same.

The BIOS also does a memory test, but this is far from satisfactory.
I've had entire dead DIMMs, be missed by the "long" BIOS test. No
beep error code, machine tries to start, and crashes.

In this case, pressing on the DIMM or re-seating it
would be a first step. With the power off while you are fooling
with it. Some motherboards now (LGA2011) use DIMM slots with only
one ejector instead of two. And the user can never be sure the DIMM
is secure in these (poorly designed) DIMM slots.

Example of a proper antistatic strap.

http://www.radioshack.com/product/in...ductId=2260808

Have fun,
Paul


Ads