> It's
mechanically very strong, and rarely
presents a problem unless
> you damage it in some way (forcing memory in
the
wrong way, pouring
> liquids into it, sandblasting it, drilling
holes
in it, etc).
>>
> And just how would you know?
Some years ago a friend turned up asking for help.
He'd purchased a motherboard that needed 168-pin SDRAM
(that's two little slots in the stick and two
corresponding
ridges in the connector to give you a hint). He' also
purchased two sticks of 184-pin DDR.
"I've managed to fit one but it doesn't
work".
FWIW the memory had survived, the socket had come
off slightly worse. With the amount of force he'd
used, he'd managed to not only to damage the ridges
in the connector but also remove at least one half
of one of the gold contacts.
Hehe. I have seen older Macintosh computers that have had memory forcefully rammed into
them. You see, most of the higher end, pre-G3 PowerMacs took 168pin, 5v DRAM. The memory
modules and the sockets look almost identical to PC style 3.3v SDRAM. Except one of the
little notches is about 1/8" to one side. Therefore, while the memory looks like it
should fit, it won't, or at least, not until it's REALLY pounded in there, and the
the tab in the socket crunches a notch into the edge of the memory module. I've even
seen some memory that had been filed to fit.
But other than that I've found DIMM sockets to be
very
reliable. I have one motherboard that I've used to
test SDRAM before selling it on, and despite having had
a large number of insertions and removals (all in the
same socket too ...) it's still going strong.
I used to work at an electronics recycler. To quickly test PC SDRAM, we simply plugged it
into a motherboard. I had this one ASUS Pentium III board that I used to test all the RAM
- and on days when we tested memory, I would literally test hundreds of modules at a rip.
Plug one in, hit the power switch, CMOS test passed - note amount of memory it registered,
turn off the board, yank the chip and toss it in the appropriate bin, plug in another,
etc.
That board survived many thousands of inertions/removals. Eventually the tabs on the
sockets would break off, but it didn't much matter, you could still plug and unplug
the memory. The board would continue to work just fine. Occasionally a socket would
require cleaning, which was done by spraying cleaner on a junk memory stick, and jamming
it in the socket repeatedly, then letting the board dry. I can't really remember ever
having to _replace_ the motherboard. It always just worked. Only one socket at a time was
used - alternating whenever one seemed to be getting dirty, but the first socket took most
of the abuse.
These "memory tests" were far from conclusive, and simply served to quickly weed
out the obvious duds and sort by capacity. But, the board survived. Computer hardware is a
lot more robust than most people give it credit for.
-Ian