OK, that didn’t happen. But I’m definitely more bothered by the constant drone of a high-powered computer system than I used to be. Recently, I upgraded my system and removed a 760MPX board with dual Athlon MP 2000+ processors. I still had a use for this setup, so I put it into an Antec SX-630 enclosurewhich I had in reserve.
Unfortunately, it was loud. Very loud. Between the three 80mm fans (not counting the PSU fan), the Volcano 6Cu CPU coolers, and the fact that the SX-630 (unlike some of Antec’s newer cases) wasn’t really designed for quiet, it sounded like a large hive of bees had taken up residence in my office.
Something had to be done, but I didn’t want to blow a lot of money on the problem, money that could be better spent on… faster computer parts. I decided this would make an interesting experiment. How much could I improve the noise level of this case without shelling out a lot of cash? Read on and see.
Pinpointing the culprits
The first thing I wanted to do was get an idea of which components were generating the most noise. Cracking the case open, I gathered valuable information using the the highly advanced “touch the center of the fan until it stops” technique. I decided that, while the CPU fans were likely the majority of the problem, the case fans (one Antec included with the case, and two generic 80mm fans) weren’t exactly helping the situation.
Speaking of case fans, let’s talk about placement for a moment. On the SX-630, there are spots for two fans up front and one in the rear (again, not counting the PSU fan). One of the front fans mounts in the traditional spot near the bottom of the case, while another can be mounted in one of the 3.5″ drive cages to blow air across the drives.
Another thing to consider is the fan grills for the rear and lower front case fans. While these are great for keeping the appendages of small pets and children intact, they also reduce airflow and create turbulence and thus noise.
With my assessment of the problem complete, it was time to shop. Poking around online, I found a couple of Panaflo case fans and a couple of Copper Silent 2 TC CPU coolers. Yes, there are probably quieter fans and quieter CPU coolers, but recall the money limitations I set out earlier. I wound up spending just over $40, not counting shipping.
As you can see, the Copper Silent 2 TC has a larger heatsink with more surface area, as well as a larger fan than the Volcano 6Cu. Additionally, the fan is temperature controlled, which will hopefully translate to slower rotational speeds and less noise.
Once the parts arrived, it was time to begin the operation. I decided to do it in several steps. Some of these steps would consist of installing new parts, while others would involve modifying the existing configuration. At each step, I took readings with a sound level meter. From a foot away, I measured the noise level at the front, side and rear of the case.
After obtaining baseline measurements, I unplugged and removed the front fan that blew over the hard drive. Given the fact that nearly all 7200rpm hard drives do just fine without the benefit of a fan, I decided it just wasn’t necesary. One less fan is one less source of noise.
I mentioned earlier that the fan guards can create noise. For the sake of your fingers, it’s typically not a good idea to remove them, at least not without replacing them with a less-restrictive guard. In my case, however, the physical placement of the computer made it basically impossible for anyone to come in contact with the rear fan, and the front fans were covered by the front bezel. Out came the tinsnips…
I was surprised at how much removing the fan guards helped not only the noise but the airflow as well. In fact, I had to modify my testing procedures for the rear of the case, because the higher air velocity caused the sound level meter to read artificially high. From this point on, the rear sound level measurements were obtained by putting the meter above and behind the case, angled down towards the center of the back face.
Unfortunately, for obvious reasons I couldn’t take new “before” measurements from this new position, so be aware that the rear baseline measurements used slightly different methodology, and should therefore be taken with a grain of salt.
At this point, I took my second set of sound measurements. Next, I swapped out the two remaining fans for the Panaflo 80mm units, and took another set of measurements. Finally, it was time to install the new processor heatsinks.
Here, I hit upon a bit of a problem. It goes without saying that installing a heatsink the size of the Copper Silent 2 TC can be challenging in the best of circumstances. Additionally, while I can see the point behind them, the heatsink’s three hole retention clips make it absolute bear to install. The center hole is basically the same size as the plastic piece under which it fits, giving you absolutely no leeway in either direction. You must line the clip up as best you can, press down with great force and hope that you’re lined up. If not, back off and repeat ad nauseum.
The painful procedure I’ve just described was made worse by the fact that I was working on a dually board. Not only were there two heatsinks to install, but the component density on this particular board made for more than double the pleasure, double the fun. In particular, a finned heatsink glued to one of the voltage regulators interfered ever so slightly with one of the Arctic Cooling heatsinks. I wound up using a pair of vise grips to pinch the fins together on the voltage regulator heatsink. That and a few dozen curse words was sufficient to get the second CPU heatsink installed.
Here, I took a fourth set of measurements. However, these measurements were taken with the system at idle. Because the Arctic Cooling fans are temperature sensitive, I decided to take one more set of measurements after running Folding clients on each CPU for a half hour or so. I figured that would make for a good “worst case” scenario.
So the bottom line is, was it worth it? Man, after all that I hope so. Let’s take a look.
First off, note that I colored the rear measurement for the stock configuration a different color, as a reminder that I changed my methodology for subsequent measurements. Looking at the results, it’s clear that each step along the way helped some, though the biggest reduction (not surprisingly) came from the new CPU heatsinks.
As you can see, however, using the machine for Folding did push the noise level back up slightly. I assume that this is entirely the result of the CPU heatsink fans stepping up their rotational speed in response to the higher heat of the processors.
Subjectively, the before and after difference is substantial. Before, the drone emanating from the corner of my office was loud enough to be a distraction. Now, I only really notice the noise if I stop to listen for it, and that’s with the machine Folding 24/7.
Obviously, there are a million different ways to approach this problem, from the very cheap (modifying existing equipment without purchasing any new hardware) to the relatively expensive (rip out all the fans and build/install a water-cooling system). In my case, I tried to get the most bang for not very much buck, and I’m satisfied with the results.
In the course of this project, I came up with other things to try, such as stepping the fan voltages down to 7V or finding some type of foam or case lining to deaden the sound. In the end, I decided to stop here simply because I’m satisfied with the results. If you have a machine that makes more noise than you’d like, hopefully you can use some of these techniques to quiet things down without stealing too much cash from your all-important Video Card Upgrade Fund.