SPCR

I did some ducting to get rid of the relatively noisy stock cooler on my factory overclocked 8800GTS while retaining the feature of exhausting the hot air directly out of the case. It required two duct iterations to get the job done: the first duct failed miserably while the second improved one is working quite nicely. Otherwise the system comprises of a Q6600@3Ghz, 4GB memory, Samsung 500GB HDD in the Scythe enclosure sitting on foam + draft proofer legs and the overkill PSU. Here are the pics:

First the failed duct. Exhausting done by a 92mm Nexus fan through removed PCI slot covers. Not enough airflow going through the heatsink fins, fine for 2D use but 3D temperature testing had to be aborted when the GPU core temp was at 90C and rising steadily.


This is the successful duct on its own. It was made of clear bendable but somewhat sturdy plastic cut from a desk cover. The separate pieces were mainly attached to the 120mm Nexus with double sided tape and to each other with strong tape. The idea is to slide the fan + duct under the HR-03+ to get localized airflow through the heatsinkink fins.


Duct from left perspective. It is not attached to the HR-03 in any way. The exhaust path is sealed all the way to the PCI cover area where some leakage is possible. The fan is controlled with a Fanmate, acceptable temps (63C idle, 82C load) were reached with the Nexus at 800RPM.


Duct from the right. The right edge of the duct is intentionally cut to a bent shape to avoid the motherboard headers and connectors. The slot covers above the exhaust area will act as intakes to get some airflow over the HSI heatsink. It is possible that some exhausted hot air will be taken in but I don't think this is a huge problem.


Overall system view. Please note that the duct removes virtually all chances of using expansion boards. Only one PCIE slot is left free and that is between the graphics card and the CPU. For me this is quite acceptable as I have no need for expansion boards, at least not at the moment. All three standalone fans are controlled by fanmates taped to the back of the case, GPU fan is running at 800RPM, back exhaust Nexus at 700RPM and the Scythe 1200RPM nominal CPU fan at 600RPM.


Everything considered I'm happy with the duct, the system is quite quiet regardless of the load. One possible problem with the duct is the location of the fan: it will be sucking hot air over itself when operating which will probably shorten the bearing lifespan. I trust that the BIOS will give the warning beep when the time comes

Very nice built,congrats for the hand made duct.
So the fan is sucking the hot air from the heatsink,in your configuration ,what if you reverse the fan and cover a larger amount of the video card and heatsink with this plastic material,so that the air will be directed to the heatsink itself and out the back.
Do you think this is possible??since you have some experience with duct fixing.
I had this idea but didnt had time to experiment.

I like the duct a lot. Thanks for sharing. What voltage are you running to your CPU? What load temps are you getting on it?

Very nice, but you need to make a seal around the edge of the heatsink with the fan - it would be much more effective this way.

What's that black box thing in the lower right corner of your case?

TK421, thats a hard drive enclosure, meant to reduce vibration and seek noise from a hard drive. its called a Scythe Silent Box...not the cheapest things in the world, but well worth it for silence.

Thanks for the feedback to you all!

lor77 wrote:what if you reverse the fan and cover a larger amount of the video card and heatsink with this plastic material,so that the air will be directed to the heatsink itself and out the back.
Do you think this is possible??

Basically I think it's doable but there are a few problems. Sealing the graphics card <-> heatsink area is going to be painful with all the motherboard components and gaps between the graphics card and the MB. This kind of a duct wouldn't be easily removable.

Also, if the fan is reversed the hot air will be flowing over the memory & HSI heatsinks which may not be good. In all honesty I have no idea about the thermal properties of those chips. Anyway it would be nice to know if it works, please tell us when you get the time to do some experimenting

miahallen wrote:What voltage are you running to your CPU? What load temps are you getting on it?

I'm running the CPU on the basic voltages given by the MB. The CPU also undervolts and steps down the multiplier correctly when not loaded which is nice. CPU-Z reports 1,12V on idle, 1,216 when fully loaded. Load temps are currently at 70C after I increased the RPMs of CPU fan and exhaust fan to about 700 and 780 respectively to account for not having the balcony door open anymore (it's too cold outside!).

Bobendren wrote:Very nice, but you need to make a seal around the edge of the heatsink with the fan - it would be much more effective this way.

It might be but then again it would not be easily removable. The 120mm fan is also a lot bigger than the HR-03+ heatsink area so it would be a compromise anyway. I don't know if some kind of half-sealing would necessarily produce good results as many fans / fan installation methods leave the airflow area non-sealed. Open-framed Arctic fans and different silicone mounting methods for fans come to mind. Still, if someone has done testing with sealed vs. unsealed fans on HR-03+ I'd be interested to hear the results.

miahallen wrote:What voltage are you running to your CPU? What load temps are you getting on it?

I'm running the CPU on the basic voltages given by the MB. The CPU also undervolts and steps down the multiplier correctly when not loaded which is nice. CPU-Z reports 1,12V on idle, 1,216 when fully loaded. Load temps are currently at 70C after I increased the RPMs of CPU fan and exhaust fan to about 700 and 780 respectively to account for not having the balcony door open anymore (it's too cold outside!).

What's wrong with that mobo? It should OV just fine, but if you do so, make sure you disable EIST & Speedstep in the BIOS, they'll cause problems. However, considering your temps, I wouldn't advise it. They are to high with that modest OC, I'd double check your mounting.

Thanks again for sharing you work on the duct, I'll have to take some pictures, but I created a simpler version for mine. I'll try to get some photos up in the next couple days.

miahallen wrote:What's wrong with that mobo? It should OV just fine, but if you do so, make sure you disable EIST & Speedstep in the BIOS, they'll cause problems. However, considering your temps, I wouldn't advise it. They are to high with that modest OC, I'd double check your mounting.

Thanks again for sharing you work on the duct, I'll have to take some pictures, but I created a simpler version for mine. I'll try to get some photos up in the next couple days.

Huh? Nothing wrong with the MB, I just don't see any reason to overvolt as the CPU is stable at the speed I want it at. Overvolting would only increase the load temps and disabling EIST/Speedstep just for the heck of it seems a bit stupid considering the impact on idle temps.

I also don't think the load temps are too high for a Scythe fan at that RPM and as the CPU is stable and nowhere near throttling why go for lower temps?

Waiting for the pictures

I'd be concerned about the lifespan of the CPU if it's hitting 70C peaks. With the case exhaust and duct fan extracting...you may be a bit restricted on your inlets. If you can duct even some cooler room air to the inboard side of the CPU heatsink you should get some temp drop. Maybe a couple of pieces of 1" plastic hose from top holes to the CPU heatsink?

ronrem wrote:I'd be concerned about the lifespan of the CPU if it's hitting 70C peaks. With the case exhaust and duct fan extracting...you may be a bit restricted on your inlets. If you can duct even some cooler room air to the inboard side of the CPU heatsink you should get some temp drop. Maybe a couple of pieces of 1" plastic hose from top holes to the CPU heatsink?

This is something that I'd actually like to look into. Do you have any authorative links describing the temperature/life expectancy of modern CPUs? After a bit of googling I couldn't find any "official" info on what is the actual maximum recommended core temperature. Lots of opinions and speculation though I might do some intake opening at some point, the P182 has considerable problems with the main chamber airflow.

Overall I'm not that worried about my temps as the only practical situation where the temps are at ~70C is when I'm rendering something. Gaming usually loads ony one core to 100% and that doesn't increase the CPU temp by much.

Also, a couple of things that I'm wondering about: has anyone actually had a broken CPU due to sustained highish temperatures? I ran my old computer's XP2200+ higher than recommended temperatures (around 80C max perhaps, I don't remember exactly) without any problems and the system is still running fine after about 5 years. And if high temperature delta between off <-> idle <-> load is bad for processors in general then how come there is very little talk about GPU life expectancies?

A Q6600 will last for a very long time at 70C. Depending on the stepping, its throttle temperature is either 85C or 100C. As a rough rule of thumb, a 10C temperature increase cuts the MTBF in half. Intel's target MTBF is over 10 years. I sure don't worry about my 24x7 70C E6600 lasting only 5 years...

Many people here, and most people on other forums, obsess about CPU temperatures, and especially idle CPU temperatures.

mellon wrote:

ronrem wrote:I'd be concerned about the lifespan of the CPU if it's hitting 70C peaks. With the case exhaust and duct fan extracting...you may be a bit restricted on your inlets. If you can duct even some cooler room air to the inboard side of the CPU heatsink you should get some temp drop. Maybe a couple of pieces of 1" plastic hose from top holes to the CPU heatsink?

This is something that I'd actually like to look into. Do you have any authorative links describing the temperature/life expectancy of modern CPUs? After a bit of googling I couldn't find any "official" info on what is the actual maximum recommended core temperature. Lots of opinions and speculation though I might do some intake opening at some point, the P182 has considerable problems with the main chamber airflow.

Overall I'm not that worried about my temps as the only practical situation where the temps are at ~70C is when I'm rendering something. Gaming usually loads ony one core to 100% and that doesn't increase the CPU temp by much.

Also, a couple of things that I'm wondering about: has anyone actually had a broken CPU due to sustained highish temperatures? I ran my old computer's XP2200+ higher than recommended temperatures (around 80C max perhaps, I don't remember exactly) without any problems and the system is still running fine after about 5 years. And if high temperature delta between off <-> idle <-> load is bad for processors in general then how come there is very little talk about GPU life expectancies?

http://www.tomshardware.com/forum/22174 ... ture-guide - great read on Core 2 CPU based temps.