Heat & Speeds

Resources, guides, discussions for modding your system.

Heat & Speeds

Postby Tulatin » Fri Mar 14, 2003 10:52 am

Atang1, you say that CPU's have to run at slightly high temperatures to be efficient so ;
1. Is this why when some people Overclock using a phase-change cooler they get crappy results (after all the CPU is running at 5F).

2. Why some CPU's have to warm up - for example, i've seen someone playing UT2003 on their comp from a cold boot. At first it runs really slowly, but after 15 or so minutes the framerate is good - but the CPU temperature is also 45c

3. Why when i play May Payne, it starts off choppy with the cpu at 30c and then runs smooth at 39c?
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby atang1 » Sat Mar 15, 2003 7:06 am

The theory that explains all this phenomena in the world of semiconductor is strain due to the difference in material cosefficient of expansion. And the theory of atomic spacing distortion, or technically called dislocation of atoms.

When temperature change due to applying electricity, the gate oxide expands with the silicon. Except that oxides expand less than silicon or metals in genral. This sets up strain from dislocation of adjacent atomic spacing. this cause the dark currnt to increase and the transistor heats up. Since the gate current increases, the speed of the transistor increases. Intel is now doing double gates to increase the speed of saturating the cmos transistor based on this theory. Two gates saturates the source and drain area much faster. The other one is dislocation saturates the source and drain area much faster.

There are, of course, manufacturing processes that left dislocations in the silicon or gate oxide, that we can anneal them away. And we can run the cmos transistors warmer to have more dark current and the transistors saturates faster.

Once, you understands the technical theory then you can create situation where your computer can run faster. Or even, you can fix your cpu or chipset or memory etc. to prevent them from total failure. My job is to let people understand the theories and let them know the simple procedure to cure their semiconductor devices. As you can tell I have substantial experience in the semiconductor field, having worked in the industry since the early pioneering days along iwth Dr. Noyes(Dr. Grooves boss) and many others. I had mu own company to fool around with.

Your many question involve many other material, and they will be answered on their own merits.
atang1
 

Postby atang1 » Sat Mar 15, 2003 7:39 am

Phase change means the substance changes from solid to liquid or gas.

The phase chnage pads are made of wax. Wax is a composition of bees wax and resin. The melting point shifts with the amount of resin up to over 350 degrees F. So, the phase change pads on heatsinks must be phase changed within the operating temperature of the cmos transistor operating temperature range of -20-80 degrees C.

However, this is the failure mechanism of waxes. It is working if it is a solid; but if it goes into the liquid phase, the spring on the heatsink squeezes out the liquid easily.

Then the heatsink sits on the silicon chip bare. When the cpu cools down, the solid wax shrinks and separates from the silicon chip. Never use phase change pads on bare silicon chip cpus.

This is not true when you use phase change pads on cpus with heat spreader, such as ppga cpus. The phase change pads stay mostly solid. And the condition can be tolerated for a long time. I believe that phase change pads came in when the cpus were using ceramic packages or heat spreaders.

It is not difficult to remove the wax with a single edged razor blade and apply some thermal grease with zinc oxide and silicone. Silicone is the real heat transfer media, which, however, may dry up on you.

The fact that Arctic Silver has to be 72 hours later, to be effrctive, it tells you not to trust it. Zinc oxide has a coefficient of 0.8 and the best other thermal grease is 0.7. The coefficient has to be multiplied by the thickness of the thermal grease to have the total heat transfer. So, use thermal grease sparingly. The others being plastic glue, rubber adhesives, carbon fibers, silver in solution and polymers, are not as effective as silicone, whichever can be applied the thinnest. And it is the thickness of the thermal grease that counts. Remember, in the same breath, some heatsinks are down to 0.35 in thermal resistance coefficient now.
atang1
 

Postby atang1 » Sat Mar 15, 2003 8:09 am

There is a sweet spot for your cpu when temperature is just right for the cmos transistors to run at the faster speed. Cold temperature is not it(the sweet spot). People change core voltage to get to the sweet spot. Changing core voltage increases the cpu temperature.

As the cpu wattage increases, people had to defend themselves from catastrophy by using the largest heatsink and fan. Unfortunately, as the heatsink gets bigger, the mass(weight without gravity) gets bigger. It take 15 minutes to reach temperature equalibrium. Or it take 15 minutes for the cpu to reach the sweet spot of fastest speed.

It is a fact of life that we are living with. But fortunately, Taiwan heatsink and fan designers are being asked to keep mass at a minimum by using thinner fins and base in heatsink design. And the fins has to have a tunnel to force velocity up to be effective. But the velocity has to stay within laminar flow. Turbulent flow is ineffective like an egg beater. Look at some of the latest heatsink? Copper can be made thinner in fins and base even if the mass is larger than aluminum. Manufacturing method of aluminum heatsink is limited on how thin the fins can be and how close they can be extruded? Cut aluminum fins will be more expensive to make?

There is hope yet in heatsink designs as we learn the hard way by experience. Hard knocks or street smarts?
atang1
 

Postby atang1 » Sat Mar 15, 2003 8:20 am

You are getting the street smarts when you noticed that your cpu should have at least 39 degrees C read by your sensor software. Others will find different temperature readings to be ideal speed for their cpu.

Temperature reading by sensors are all over the map. Silicon diode in the cpu chip temperature reading theoretically will be exact. But alas, the fudge factor used in temperature presentation software is not exact.

However, if you are a mechanical engineer, a trend is sufficient. We will know our way, without being exact. Just document the changes with time and you can be scientific.

Hopw this answers all your inquisitive mind.
atang1
 

Postby Tulatin » Sat Mar 15, 2003 1:07 pm

that's a lot of posts, but yes it did answer my thoughts. When i was referrig to phase-change, i meant the cooling systems such as the Chip-Con Promethia or the Vapochill.

With this 'sweet spot' of temperatures, it seems that the stock cooling (retail boxed) seems to find it the best.
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

Postby atang1 » Sat Mar 15, 2003 9:30 pm

If cpus are using bipolar transistors and run 200 degrees C., when water changes to steam; it is a phase change cooling device, such as standing water pipes with a small amount of water in the pipe. Steam cooled by air at the other end and changed back to water, sliding back to the cpu. They used them on real time computers. But even real time computers don't use them anymore, because of efficiency.

If you use airconditioner to cool a cpu, it is just an air conditioner(60-75 degrees F air), already compressed gas into liquid and as it evaporates it cools. It is not considered a phase change device caused by the heat of the cpu. Only phase change pads qualify as the name implied.

Many people post ideas on the motherboards.org. Most lack engineering knowledge. To each his/their own. Not enough time in the day to correct any of them. Let them eat cake/have fun.

On bulletin boards, such as "gone in 60 ns" threads, some fundamental concepts are defined.
atang1
 

Postby triniphen » Thu Jun 05, 2003 5:30 pm

atang1 wrote:If cpus are using bipolar transistors and run 200 degrees C., when water changes to steam; it is a phase change cooling device, such as standing water pipes with a small amount of water in the pipe. Steam cooled by air at the other end and changed back to water, sliding back to the cpu.

If you use airconditioner to cool a cpu, it is just an air conditioner(60-75 degrees F air), already compressed gas into liquid and as it evaporates it cools.

Most lack engineering knowledge.



1. water boils at 100C, at 200C it explodes.
2. the pipe is a condenser, if the condensed material returns to its orgin it is refluxing, if collected elsewhere is it distilling.
3. air conditioners outputting air at 60F are called fans.
4. the gases boiling point is below ambient temperature hence it absorbs heat from the surroundings and boils, taking the heat with it (endothermic).

im not trying to be mean i just like accurate descriptors.
Last edited by triniphen on Thu Jun 05, 2003 5:33 pm, edited 1 time in total.
A7N8X-Raid/Aud/Lan/SATA
AMD XP2800+
Radeon 9700 Pro
Samsung DDR400 512MB
WD Raptor 36Gb SATA RAID 0
Maxtor 60GB ATA133
Plextor 52/32/52
Viewsonic Af90+ Flat CRT 19"
triniphen
Brown Belt
Brown Belt
 
Posts: 384
Joined: Thu Jul 12, 2001 12:01 am

Postby Tulatin » Thu Jun 05, 2003 5:32 pm

Phase-Change System: Water in a vaccum, which circulates between the CPU and a compressor. Kind of like a CFC cooling system.
Tulatin
Enlightened Master
Enlightened Master
 
Posts: 14664
Joined: Wed Dec 11, 2002 2:29 pm
Location: Canada

eep

Postby triniphen » Thu Jun 05, 2003 5:37 pm

even worse, the BP would be lower, i hope theres one hell of a condenser on that thing.
A7N8X-Raid/Aud/Lan/SATA
AMD XP2800+
Radeon 9700 Pro
Samsung DDR400 512MB
WD Raptor 36Gb SATA RAID 0
Maxtor 60GB ATA133
Plextor 52/32/52
Viewsonic Af90+ Flat CRT 19"
triniphen
Brown Belt
Brown Belt
 
Posts: 384
Joined: Thu Jul 12, 2001 12:01 am

Next

Return to System Modification

Who is online

Users browsing this forum: No registered users and 2 guests