View Single Post
  #10  
Old February 14th 04, 08:29 PM
John Lewis
external usenet poster
 
Posts: n/a
Default

On Mon, 09 Feb 2004 18:09:56 GMT, "Homie" wrote:

Many semiconductors are designed to run at temps as high as 125c....... but the
cooler they are kept , the better the slew rates, the better the slew rate, the better
the thermal efficiency, the better the thermal efficiency, the cooler they run....the
cooler they run, the longer they last .......
Cool huh?

Agreed... However, having adequate design margins to run within timing
spec at 125 degrees C is not the same as being specified for operation
at 125 degrees C. Consumer silicon is normally spec'd for continuous
operation at a maximum 70 degrees C case temperature, unless
explicitly stated otherwise . For example, the latest P4 Northwood
processors are spec'd by Intel for 70 degrees C and the P4 EE at 64
degrees C, (both core temperatures) for 'reliable' operation.

[ Above these temperatures, built-in hardware thermal-protection will
arbitrarily cut-in at an unspecified (by Intel) core temperature to
lower the internal clock rate to preserve silicon integrity. ]

Beware of case temperatures in excess of 80 degrees C on high-density
devices on the latest silicon processes. Thermal stresses on the
silicon vias and thin-dielectrics. Also, the device density magnifies
the failure incidence. It just takes one via or one transistor in the
100million or so in a CPU or GPU to generate a useless lump of
silicon.

Devices rated for continuous 125 degrees C operation are low-density
very conservatively spec'd and normally radiation-hardened too. I
doubt if Spirit or Opportunity incorporate any large devices on 0.15u
(or smaller) processes......

John Lewis


Homie

--
Mainboards, Videocards & CPU pin repair.

http://motherboardrepair.com


"Wayne Youngman" wrote in message
...
Hi,

I just bought myself a new Digital Thermometer today, and the first thing I
could think of testing was the *South-Bridge* of the NF7-S system I just
sold.

Since the machine didn't have any hard-disks I just let it load up memtest
x86 and loop for a while.

15 minutes later. . . .the reading from LCD display was. . . .68°C lol!

I know we discussed this before (a few months ago), but isn't that too hot?
I did try to find some pre-made heatsinks in the *U.K* but no luck. Why
didn't anyone start selling heatsinks for the nForce2 boards?

The South-Bridge is situated very close to the AGP slot so I have no idea
how people are using Zalman heatsinks, surely they get in the way of the AGP
card, especially if it has a meaty heatsinks itself?
--
Wayne ][
new specs coming soon!