Commentary
3/17/2004
09:33 PM
Fred Langa
Fred Langa
Commentary

Langa Letter: Cool and Quiet, Part III

Fred Langa wraps up his exploration of the causes--and cures--of noisy PCs. His step-by-step process, including successes and failures, ensures you won't make mistakes as you work to make your PC cool and quiet.



In Cool and Quiet, Part One, we started with the basics of quieting a PC's noisy fans, and saw how PC noise reduction can actually be rather easy and cost just a few dollars. A high-quality, nearly silent PC fan can cost under $10, for example.

We continued our exploration in Cool and Quiet, Part Two and took a close look at some of the science of PC fan noise, including the odd fact that two slow-spinning fans can be quieter and yet provide more cooling than one fast-spinning fan. We also looked at the most common and effective PC noise-reducing technologies and hardware, including inexpensive, snap-in (no tools needed) add-on fan speed controllers that let you slow down--and quiet--almost any PC fan.

In fact, by the end of Part Two, we'd covered nine major elements of PC noise and its control:

  1. Most ongoing (nontransient) PC noise usually comes from the cooling fans.
  2. There are three main components to fan noise: air turbulence (a "whoosh"); tip noise (a whir or whine), and mechanical noise (clicks, grinds, rattles).
  3. The simplest way to reduce all fan noise is to reduce fan speed.
  4. Reducing fan speed also reduces airflow, which reduces cooling ability, a potential danger.
  5. You should monitor your PC's internal temperatures using a tool like the free MBM5 before, during and after any changes to your PC's cooling system.
  6. To maintain a safe and effective level of cooling with lower fan speeds, you'll need to do one or more of the following: increase the size of the fans; increase the number of fans; increase the size of heat sinks.
  7. It's counterintuitive, but increasing the number of fans in your system may reduce the total system noise: Replacing, say, a 30dB factory-installed fan with two 20dB fans will cut the noise by 7dB, but can still produce the same or greater air flow.
  8. You easily can attain truly "whisper quiet" operation (around 30dB or less) for all your PC's fans combined, using standard, inexpensive, off-the-shelf fans and components, good fans can be amazingly inexpensive.
  9. The simplest way to reduce a PC fan's speed is by reducing its voltage from the usual 12v to 5v. This can be done by directly connecting the fan to the PC's 5v power supply, or by stopping down the normal 12v via a resistor. And again, ready-made parts for this are easily available, and very inexpensive.

If you haven't done so already, please read Parts One and Two now, as they contain the full background and additional details on the above that make today's article--Part Three--make sense.

Five Changes
In this article, we'll wrap it all up: I'll show you exactly how I modified my system in five successful ways--and in one unsuccessful way. I'll also show you my final results both for sound and for temperature, even after running the system through a worst-case thermal stress test.

If you recall from Part One, my initial concern was with a clicking noise produced by my system's main case exhaust fan. The PC--a brand-new and well-made 3.2GHz P4--has thermal sensors and a fan-control system (in the BIOS) that can adjust the speed of the case fans according to need: When the system is working hard, the fans operate at full speed, with all the whooshing and whirring that's common to most PC fans. But when the system isn't working flat-out--which is most of the time--the BIOS intermittently cuts the power to the fans. By cutting and restoring the power to the fans several times a second, the fan spins more slowly, practically eliminating the whooshes and whirs of high-speed turbulent air flow. Nice.

Unfortunately, each time the power is restored to the fan, the fan assembly jumps just a bit, producing a small but audible click or tick. Like a pebble in one's shoe, this steady ticking, repeated several times a second all day long, eventually becomes quite irritating.

So, when I began looking at noise-reduction options for my PC, my initial goal was to eliminate these sharp ticking sounds, but without increasing the other fan noises beyond their already low levels.

Plus, as also described in Part One, I'd also discovered that the airflow in my PC's case largely bypassed the hard drive, which consequently ran warmer than I wanted, averaging about 20 degrees F/12 degrees C warmer than the air inside the system case. This was well within spec, but it seemed to me that there was no real reason for the drive to be more than a few degrees warmer than the rest of the case. (Needless extra heat would eventually shorten the life of the drive.) So, I also wanted my cooling solution to eliminate the dead-air spot near the hard drive and thus keep the hard-drive temperature closer to the system's internal air temperature.

Small Initial Missteps
Initially, my tinkering was personal and very casual--it wasn't part of a controlled experiment, and wasn't intended as a topic for publication. Nevertheless, I did take some standard precautions before doing anything: As always, I made a full backup (actually a software "image" of the hard drive) so no files were at risk. And I was monitoring the temperatures of the CPU, motherboard and hard drive via MBM5. I even had the software set up to trigger an automated orderly shutdown of my system if any temperatures got too high. So I wasn't worried about toasting my system or losing data.

But beyond that, my initial approach was very casual: I assumed the ticking noise was caused by a flaw in the fans themselves, so I went online and after a brief search, bought a pair of fans (both under $10) that claimed to operate at a quiet 20dB each. No, it wasn't a very scientific process--yet.

I took out the original case fan--a simple screwdriver operation involving just four screws--and replaced it with one of the new fans. I mounted the second fan in an otherwise-empty spot in the front corner of the case, aimed upward toward the bottom of the hard drive. (There had been no fan there originally, although the system was designed to accommodate one.) Both fans simply plugged into existing fan-power connectors on the motherboard.

This initial, crude experiment was a partial success. My hard drive temperatures dropped to within a degree or two of the case air temperature, so I had successfully eliminated the dead-air spot inside the case. But the ticking noise remained when the fans were at low speed. Clearly, it wasn't the fans that were the problem there, but the way they were being driven. That was "aha!" moment No. 1.

"Aha!" moment No. 2 came when I realized that the two fans together didn't seem much noisier than the one original fan had been. How could that be?

That's when I finally realized that something interesting was going on, and that there was more to PC cooling than met the ear. I then belatedly began the research that turned into Parts One and Two of this series.

Once I was up to speed on the sources of fan noise--including odd ticks and clicks--and learned why two fans weren't necessarily twice as loud as one fan, I was ready for a more serious attempt at controlling noise in my PC.

First Full Success
For all the reasons discussed in Parts One and Two, I decided to opt for voltage-reduced, slow-spinning fans. I ordered a $9 Zalman ZM-F1 fan that includes an integral resistor; and a $3 Zalman "Multiconnector" so I could feed the previously installed hard drive fan 5v instead of the 12v supplied by the motherboard.

The Zalman fan is an exact replacement for any standard case fan: Unplug the old fan's connector from the motherboard, remove four screws, and take the fan out of the case. Put the Zalman fan in, insert the same four screws, plug it in exactly as the other fan had been (fan connectors are keyed, so you can't plug them in wrong), and you're done. Total time: Maybe five minutes.

The multiconnector is even simpler to use--a no-tools, fingers-only fix. Plug the multiconnector into any unused power supply connector (it's keyed so you can't plug it in incorrectly), unplug the fan's power connector from the motherboard, and plug it into the multiconnector. Total time: Maybe a minute.

When I restarted the system, I entered the BIOS setup, and disabled BIOS control of the fans, so the voltage-reduced fans would be fed steady power, without the BIOS trying to control their speed.

It all worked beautifully: Both fans now spun slowly, quietly and steadily, without clicks; the hard drive temperatures remained almost 20F/12C cooler than they had been before; and the system temperatures overall were essentially the same as they'd been with the slow-but-clicking BIOS fan control. The cost of the Zalman fan, the multiconnector, and the new hard-drive fan totaled around $20. It's about the best 20 bucks I've ever spent on hardware!

Other Fans, Other Noises
I could have stopped there--a simple, two-fan $20 fix that removed my system's most annoying noise problem and solved a previously hidden hard-drive thermal problem. But I was intrigued: If it was so simple and inexpensive to solve PC noise and cooling problems, what else might be possible?

The next-loudest noise source in my system was the factory-supplied fan and heat sink used to cool the CPU. It's a standard Intel-issue design using a heavy copper base plate with 52 thin copper fins, topped by a high-speed, and oddly sized 65mm fan. Although this design gives very good thermal performance, it's fairly noisy: The close-spaced fins produce a lot of air turbulence; and the fan itself produces a noticeable whine.

There are many, many excellent heat sink/fan combinations available, so please don't take my choice, below, as a specific recommendation. Rather, it's simply an example of one successful design. I'm sure I could have gotten similar results from other brands and models.

But given my success with the Zalman case fan and multiconnector, my inclination was to give a Zalman CPU heat sink and fan a try. I chose their CNPS7000A-Cu, a very large, all-copper heat sink with petal-shaped fins surrounding an oversized--92mm--centrally-mounted, low-speed fan. The unit also comes with a Fan Mate speed controller, which lets you select the exact speed and noise levels you want.

What attracted me to this particular model was Zalman's claim of nearly silent operation: just 20dB at the fan's slowest speed. (Even at full speed, the unit is rated at a very quiet 25dB.) That enormous fan is the key: it doesn't have to spin fast to move a lot of air through the oversized heat sink, which in turn can dissipate heat even with the slower, quieter, less-turbulent airflow. It looked like an excellent design, so I bought one; I paid $37 for the complete kit.

I admit I was a little nervous in removing the original heat sink, but all went smoothly. The Zalman kit contained literally everything needed to install the new unit, and the instructions also were very clear. Zalman even has animated Flash tutorials on their site to show you, step by step, how to install their fans and heat sinks. I worked very slowly and carefully; the whole thing took maybe an hour, although if I had to do it again, I could probably do it all in under 15 minutes. It's really not hard at all.

Once the new fan and heat sink were installed I used the Fanmate to try out different speeds. Over the next day or so, I learned that I could use the lowest, slowest fan speed and still get CPU temperatures in the same low range as previously provided by the factory-installed fan and heat sink--but with just a fraction of the original noise. The only sound produced by the Zalman unit at low speed is a barely-detectable white-noise whoosh. It's a truly excellent fan and heat-sink combination.

Not a Mosquito, But...
There is a certain irony to reducing fan noises in your PC: As you control each major noise source, you may hear other noises that previously were masked by louder sounds from other components!

That's what happened to me: With the case and CPU fans all now producing very low-level, inoffensive white noise, I noticed a remaining whine that I traced to the fan on the video card--an nVidia unit with a tiny, 40mm high-speed fan whirring away on a small aluminum heat sink.

Did I have to replace this? No. But I had the sound-control "bug" by now, and wanted to remove all non-white-noise from the fans in the system.

Once again, although several companies make quiet video card coolers, I stuck with Zalman, as their other cooling products had worked fine for me. Their $22 ZM80C-HP video card cooler is a passive design using an oversized heat sink and a "heat pipe;" a sealed pipe that contains a low-boiling-point working fluid, analogous to the working fluid in your refrigerator or air conditioner. But unlike those devices, a heat pipe has no mechanical moving parts: The fluid boils at the hot end of the pipe (near the video chip, in this case) and condenses at the cool end (at the heat sink); the working fluid's "phase change" (liquid to gas and back) transports far more heat than a non-phase-change device would.

For high-end video cards that may produce unusual amounts of heat, Zalman recommends an optional fan for the video-card cooler to increase its cooling efficiency even further. As you might guess, the Zalman-specified fan is hugely oversized, slow-spinning and very quiet. And yes, I bought one. (Incidentally, the video cooler fan includes a multiconnector, described earlier, which normally costs a separate $3 or so. The multiconnector can drive two fans at 5 volts and two at 12 volts from a single power supply connection.)

The video cooler totally eliminated the whine from the original tiny fan, and added only a soft white-noise whoosh to the overall noise from my PC.

Total Costs And Benefits
Let's recap: At this point I'd replaced the case fan with a resistor-equipped low-speed unit; and added a new fan, driven at 5 volts, to cool the hard drive. This initial change, in itself, solved the original minor thermal problem my system had, and completely eliminated the ticking noise that sent me on my quest for quiet in the first place. The total cost was about $20, and it would have been fine to stop at this point.

But for another $37, I eliminated almost all the noise from the original CPU fan and heat-sink assembly, using instead the oversized Zalman CPU fan and heat-sink kit. A final, and admittedly optional, $22 video cooler eliminated the whine of my video card fan; topped off with an also optional $15 low-speed fan to get maximum thermal performance out of the new video-card cooler.

The total cost for the above five pieces of hardware was $97.

What that bought me was extreme quiet--literally "whisper quiet," in the range of 30dB or less. In fact, when I was done, I asked my wife to come into the office to gauge her reaction. I turned on some standard, unmodified PCs so she'd have a reference sound, and then turned them off and showed her the newly-quieted PC I'd been working on. It had been on the whole time she was in the room, and was literally at her feet; but she hadn't noticed that it was running--it was that quiet.

I'm very happy with the results. The PC isn't silent--that wasn't my goal--but it is wonderfully quiet, producing just a very gentle, inoffensive background whoosh.

What About The Heat?
Quiet doesn't matter if your system cooks itself to death, of course, so I wanted to verify that I still had sufficient cooling for all circumstances, and not just when the system was loafing along.

So I downloaded the simple (and free) BurnMax CPU stress-testing tool from BenchHQ. BurnMax will consume CPU cycles at whatever priority you specify, even driving your CPU at 100% output for as long as you want. This amounts to a worst-case test for CPU cooling, as sustained 100% operation will make a CPU as hot as possible. (You can find many other CPU stress- and benchmark tests, if you wish. A Google search lists dozens, for example.

Before I'd made any modifications to my system, I'd seen the CPU temperature climb to a moderate 128 degrees F/53 degrees C during lengthy CPU-intensive activities; well below the 158 degrees F/70 degrees C maximum temperature specified as safe by Intel for a P4, and indicative of the good thermal performance of the original noisy CPU heat sink and fan.

But with a worst-case heating test, and with the CPU now being cooled with quiet, low-speed fans installed, I assumed I'd see temperatures higher than I'd previously seen during real-life conditions, but I didn't know how much higher. (And, as described in parts One and Two, I was running software that would shut down my system to prevent thermal damage, in the event of imminent overheating.)

As I started the tests, the initial temperatures were 87 degrees F/31 degrees C for both motherboard sensors and for the hard drive; and 93 degrees F/34 degrees C for the CPU. I started Task Manager and BurnMax. With BurnMax set to consume all cycles in real-time priority, Task Manager showed the P4 CPU was indeed maxed out at a steady 100%. The temperatures started to climb almost immediately.

After 10 minutes, the rate of temperature increase had noticeably slowed: The CPU was at 123 degrees F/55 degrees C, and the two motherboard sensors were at 107 degrees F/42 degrees C (near the CPU) and 98 degrees F/37 degrees C at a location a short distance away. The hard drive--idle during this CPU test--remained at 87 degrees F/31 degrees C.

After about 20 minutes, the temperatures seemed to stabilize, with no further change, but I left everything cooking (with the CPU at 100% output and all fans spinning at their slowest settings) for another 10 minutes, just to be sure. After a full 30 minutes of 100% output, the CPU was at 125 degrees F/52 degrees C, the motherboard sensors at 109 degrees F/43 degrees C and 98 degrees F/37 degrees C; and the hard drive still at 87 degrees F/31 degrees C.

So my newly quieted system was actually better-cooled than it had been in its factory configuration: Even at steady, full, 100% output for an extended period, the system ran cooler than it previously had under some normal, everyday operations that were far less stressful than the artificial test.

So: The system was now quiet and cool--truly the best of both.

More, If You Want It
The only thing that gives me pause in my new setup is that my office isn't air-conditioned. All the above tests were performed during winter, with the room's heat controlled to about 68 degrees F/20 degrees C or so. In the summer, the afternoon temperatures in my office can sometimes exceed 90 degrees F/32 degrees C.

Adding 20 degrees F/12 degrees C to the temperatures I've currently measured still would be OK for the CPU and hard drive, but I don't know if the system's thermal performance will be that linear. So it's something I'll have to watch.

Adding thermostatic control might help: letting the fans spin up as the case temperature climbs is an obvious way to increase cooling capacity at the cost of a little extra noise during the warmest times. But (1) I don't want to go back to the constant clicking noises that started this whole experiment; and (2) I'd rather not add a lot of complexity to the setup.

One option I'm thinking about is replacing the current power supply with one designed for thermostatic cooling. For example, Seasonic makes power supplies with numerous cooling-oriented features, including an enormous 120mm slow-speed fan that's thermostatically controlled. At normal temperatures, the fan is extremely quiet, but will speed up as needed to keep the case well ventilated. And with a fan that size in my PC, I could probably remove one or more of the other fans without affecting overall cooling. Once again, I might be able to end up with even better cooling with still less noise.

And, as an interesting aside: The Seasonic power supplies also are designed with high-efficiency components that generate less heat in the first place, and that has the added benefit of saving electrical power. (Less energy is lost as heat.) Seasonic claims that one of their power supplies can pay for itself in energy savings alone in a year or so, and I see no reason to dispute the claim. Cooler, quieter, and less expensive--now there's a trifecta!

If I go that route, I'll report on it in a future article. For now, please join in my discussion forum on the Listening Post. You've seen how you don't have to spend much--possibly as little as $10 or so--to make your PC noticeably quieter. And you've seen how a few tens of dollars can make your PC quite literally whisper quiet--emitting a tiny fraction of the noise it probably does now. And you've seen what I've done, so far, to my PC.

But what have you done, or what do you plan to do to your PC? What tools do you use to keep track of what's going on inside your PC? What fans, heat sinks, or other devices have you used to keep your PC cool--and quiet? And just how important is noise control to you? Join in the discussion.

To find out more about Fred Langa, please visit his page on the Listening Post.

We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Comment  | 
Email This  | 
Print  | 
RSS
More Insights
Copyright © 2020 UBM Electronics, A UBM company, All rights reserved. Privacy Policy | Terms of Service