• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Air cooled cm690. Any ideas on improvements?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

jameseboy

Member
Joined
Nov 15, 2010
Location
Livingston, Scotland
Hey all. Ive been noticing that my graphics card (hd4850) is running particularly hot recently. Now theres been no change to my system which generally runs rather cool. My cpu (q8200 oc to 2.9) generally idles between 29 - 31 degrees and under load is in around 37 - 42 degrees. Now for my graphics card, it used to idle at around 45 degrees and load at 56. however now, its idleing at 53 and under load is hitting the high 70's.

I was wondering if anyone could see anything i could do to the airflow in my case to help cool my card a bit more.
I have put images below to show my current fan status and airflow.

DSC01051.jpg

^^ Front panel off

DSC01050.jpg

^^ side. Note the 2 fans on the top are extractor fans as is the rear fan

DSC01053.jpg
DSC01052.jpg

If anyone can see anything i can do to help my airflow please let me know as im getting worried about my temps for my gfx card.
 
Sounds like a card issue, not an airflow issue. Clean the sink, maybe remount the gfx heatsink.
 
Yea, could be, it's an older card, so if its on the stock stuff it might just be that time in it's life cycle.
 
+1 to change the thermal paste. The stock thermal paste on the GPU are crap! It gets very dry with time.

I also heard that the transistors deteriorate over time. They need more and more power and therefore generate more heat to run at a given frequency. A mix of thermal paste + transistor deterioration could explain higher temperature.
 
I also heard that the transistors deteriorate over time. They need more and more power and therefore generate more heat to run at a given frequency.

Nah, transistors deteriorate, but I'm pretty sure they run a fixed voltage to the card so unless that was adjusted it should be the same heat because leakage levels can't change
 
Nah, transistors deteriorate, but I'm pretty sure they run a fixed voltage to the card so unless that was adjusted it should be the same heat because leakage levels can't change

Let me know if I am wrong.
When your computer is idling or under load the voltage is constant (if regulated by VRM) so if you're under load or idling, you'll keep the same voltage.
I agree with you on that part.


However, [I just apply some basic physic class. I haven't read a lot about how transistors work; I am just extrapolating on this part] when you're under load, the computer requires more juice (aka watt) to maintain the stability of the transistors (so they can keep running @ the frequency you set them at). With the deterioration, those transistor require moar juice. My idea to explain the difference of consumption is that their resistance is greater = more energy dissipated.
The voltage is just a part of the equation. Higher resistance = moar Watts = more heat @ constant voltage.
 
P=IE, unless the start pulling more current the heatload's not going to change (Cause voltage is constant). Combining this with ohm's law (V=IR...oh, I should mention I use E and V interchangably for voltage...bit of the old style bit of the new, it makes no difference)
P=V^2/R - more resistance, LESS wattage (for constant current), now if you apply ohm's law the other way around you get
P=(I^2)R, which means that for a constant voltage, an increase in current or resistance will lead to higher wattage...I can't really say how the cards are designed to behave, but I could see either side...but it shouldn't result in that bad of a jump imho
 
Yes you're right.

All of this stuff reminds me bad days :rain:

The aging transistor takes more and more time to switch thus it generate more heat.

Found that with a lil googly search:

"Gamer,
If you are still looking for ideas, I'd like to tell you a little of what I know relating to the dissapation of power by digital integrated circuits as related to basic physics. As you may or may not know, microprocessors are incredibly complicated and are driven by ideas from many other fields other than physics, so I will limit myself to a few things that are related to what you probably have learned in an introductory physics course.
First, you should take some time to familiarize yourself with the basics of how standard cmos gates are built (as circuits). As thenew3 pointed out, there is a brief time in which the transistors switch from low/high and high/low resistance. It is during this period that power is dissapated. Recall that (power = current^2 * resistance). Note that when a transistor has high resistance in a cmos circuit, it's complimentary twin will have low resistance, and thus very little power will be dissapated since relatively little current goes through it.
A logical question now would be what swifty proposes: how can we reduce the heat dissapated by the processor without doing the obvious (eg. reducing the operating frequency)
One way of addressing this is to try to limit the amount of time the transistors spend switching. One of the major factors in this (the speed of transistor switching) is capacitances present in the circuit. Recall the basics of simple RC circuits. Without having to wrestle with the hassles of fabricating VLSI circuits with Gallium Arsenide and other exotic materials (it can be done, but currently is not feasible for widespread commercial use), we can attempt to reduce the junction capacitance of transistors by the use of silicon-on-insulator technologies (the physics involved in characterizing the performance of any type of transistor, and thus the justification for the previous statement, is more involved than what is covered in introductory physics classes, so I won't go into it).
OK, that's all I have to say. Correct me anyone if you feel I am wrong."
 
Back to the GPU. Yes, your Thermal Paste could be getting old. The GPU fan also could be going bad. You might not hear it yet, but it's RPMs could be slower than they were ehn the card was new.

So TPaste, maybe the fan. Don't think is a component degrading......... Interesting discussion though... possibility? Yes..... probability? WAYYYY LOW.
 
One thing that would help is get rid of that tiny 92mm fan, or is it 80mm? Put a good fan there. Not sure what the top of your case is like, but putting a 120mm fan and cutting the mesh out could help.
 
@dsmlife , its actually the mounting plate that comes in the bottom of the cm690 case, i just popped the fan on and pushed it into the bays and it sits rather snuggly.

@Conumdrum, i have the 2 80mm fans at the top of my case as due to the size of my cpu heatsink i cannot fit 120 mm fans in, i was planning on cutting the mesh out when i get a dremel to get into it with(using a jigsaw i think will be messy) i think if i get some 20mm deep fans they would fit in better, however i had the 2 80mm ones kicking around.

As for the gfx card, im going to change the thermal paste when i finnish writting this message, shouldnt take too long to do. Probably change the paste on my cpu whilst im at it.
I presume when you say change the fan on my gfx card you mean get a new cooler? if so does anyone have any suggestions on a good performance one?
 
Back