Power Mac Noise - Fan Mods

This page describes the solutions that I have tried. There are other people who are taking different approaches to the problem. I encourage you to review them all and see what best fits your needs. Note that some of these solutions are bound to violate your warranty with Apple. None of these have been proven in any way. Use the information on this site at your own risk.

Before you do anything else, record and analyze your Power Mac noise. Know what you need to fix before you fix it!

My original Power Mac was very quiet by the time I made all of the modifications. Even though I wouldn't necessary recommend anyone make any of the same changes, it's instructive to show how to approach the problem. Once you have done the spectrum analysis, you can focus on fixes which will help your problem area. The following table shows which of the fixes I tried applies:

Frequency of largest noise peak	Solution
120Hz	Put a damping material between the power supply fans and the case.
0-200Hz	Your golden! Excluding the 120Hz spike, noise here is good.
600-800Hz	Check your firmware if you are running OS9. Otherwise, replacing the case fan as I did will help a lot, but I have not found a satisfactory replacement at this point.
1300-1500Hz	Replace your 60mm Delta power supply fans.

Power Supply Fan Replacement

For the actual installation, I simply followed the excellent instructions that were posted on Accelerate Your Mac! Here's the link:

First Fan Mod in a Power Mac G4 DDR RAM Tower

The instructions themselves are excellent. I did run into several problems that I think people should consider:

The screw in the rear of the case that holds the power supply refused to come out. It stripped on my first attempt. Although the instructions recommended the use of a Torx 10 screwdriver, I noticed that there is quite a bit of play there. It's almost as if the screw was more like a Torx 12. All of the screws in the rear had a bit of play. The Torx screws on the side fit perfectly. Someone else suggested that the screws on the rear are actually 2.5mm hex keys, although I haven't tried yet.
I ended up buying a screw extractor kit to remove the offending screw. The screw was obviously a total lost, although no damage occurred to the case.
After I replaced the fans, I had a very hard time sliding the power supply back into place. After 15-20 minutes of experimentation and heavy sweating, I finally eased it back in by using a screwdriver to force the front of the power supply down. I suspect that my particular case has a mis-alignment that caused the power supply to be so hard to install in the case. I also suspect that is why the screw stripped when I tried to remove it. It was probably tightened unusually hard at the factory to force the power supply into alignment.

Unfortunately, I never saved the noise analysis of the Power Mac before I made the change. I do remember that the noise peak at 1520Hz was 6dB higher than the one at 660Hz. After the power supply replacement, the noise peak at 1520hz was 10dB lower than the one at 660Hz. That means that replacing the power supply fans reduced the noise at 1520Hz by 16dB. Pretty darn good.

Power Supply Fan Insert

After the power supply replacement, I felt that the compuer was usable, but still somewhat distracting. It's not the volume, but the frequency or pitch. I ended up playing around some more with the original fans and noticed that they vibrated quite heavily in the power supply. Furthermore, the whining wasn't apparent until I actually installed the power supply into the case! That got me thinking that maybe it was the vibration that was causing the annoying noise. I had run the fans individually in my hand and even though the original Delta fans were loud, they had a nice low frequency whooshing sound that sounded nothing like the Power Mac as a whole.

Here's a couple of cheesy diagrams that may make it clearer.

Side cutout view	Front cutout view

Based on the idea that vibrations were causing problems, I simply inserted a piece of damping material between the power supply fans and the case. Ideally, it would be nice to use some of that rubber material that is used elsewhere in the Power Mac. Since I had no idea where to get that, I just cut up an old Palm leather case and used a slice of the leather for damping. Note that the thickness of the material makes it harder to slide the power supply back in. I originally tried putting more leather towards the middle of the power supply, but the power supply would not slide in properly.

Subjectively, I think the addition of the leather padding dropped the noise by another notch. IMHO, the drop was almost as much as replacing the fans themselves. In fact, for those people who are wary of violating their warranty, it might be easier to just try the padding without the fan modification. Opening up the power supply is not rocket science, but it's a lot harder and riskier than putting in a leather insert.

Here are the before and and after spectrums.

Power Mac spectrum (with new fans)	Power Mac spectrum (with news fans and leather insert)

	Before		After		Change in Relative Decibels
	Decibels	Relative Decibels	Decibels	Relative Decibels
120Hz	-44 dB	-4 dB	-67 dB	-19 dB	-15 dB
660Hz	-40 dB	0 dB	-48 dB	0 dB	0 dB
1520Hz	-50 dB	-10 dB	-57 dB	-11 dB	-1 dB

Case Fan Replacement

At the base of the Power Mac case is a very large 120mm Delta fan. This sucker is huge and at 150+ CFM, has incredible cooling capacity. I literally could not find any other 120mm fan which moved as much air as this Delta fan. In fact, the only reason I replaced the case fan is that I ordered it before trying the power fan insert. After putting in the power fan insert, I was reasonably happy with my Power Mac. It was certainly usable on a daily basis.

So, the best candidate that I could find to replace the 120mm Delta fan was the Panaflo 120mmx38mm 114 CFM - U Series. The Panaflo fans are well known for quiet fans. There are other brands that are even quieter, but their specs were not even in the same league as the Delta. Some are as low as 50CFM!

Before I installed the new Panaflo case I ran a 8-hour torture test that consisted of performing two Divx encodings over and over again. This maxed out CPU usage on both processors. When I checked the computer in the morning, I noticed the computer didn't sound any louder. Since the 120mm Delta case fan is variable speed, that implied the fans were not even breaking a sweat. I figured that the Panaflo's probably had enough cooling capacity as long as I didn't stuff my Power Mac to the gills with hot SCSI disks and PCI cards.

I don't have any instructions or pictures, but here are some notes about the installation:

The 120mm case fan is not screwed in any way. It slides into simple metal tabs like the power supply. IMHO, this is a bad design since any looseness in the fitting will exaggerate the fan vibration.
The Panaflo is the same size as the Delta, but the edges are thicker. It's the edges which slide into the metal tabs. I had to use a Dremel tool to trim the Panaflo down to size. This is easy to mess up, but miraculously it ended up fitting pretty well for me.
The Panaflo comes with bare wires by default. Make sure to order to the 3 Pin Fan Connector Kit . Actually, you won't need the 3-pin connector, just the metal pig tails that you scrimp around the bare wire. Put those in the Delta fans connector in the same way as with the power supply fans. Refer to those instructions at First Fan Mod in a Power Mac G4 DDR RAM Tower.
The Panaflo became unbalanced and died within 24 hours! I will explain more later.

Here are the before and after spectrums:

Power Mac spectrum (with new fans, leather)	Power Mac spectrum (with news fans, leather, and Panaflo case fan)

	Before		After		Change in Relative Decibels
	Decibels	Relative Decibels	Decibels	Relative Decibels
120Hz	-67 dB	-19 dB	-44 dB	0 dB	??
660Hz	-48 dB	0 dB	Unmeasurable		??
1520Hz	-57 dB	-11 dB	Unmeasurable		??

Ok, I will be the first to admit that these results don't entirely make sense to me. On one hand, the disappearance of the peak at 660Hz makes sense as I noticed a significant drop in the case fan noise. The Power Mac was downright quiet at this point. At times, I had to strain to hear it over the background noise from a refrigerator 15 feet away. On the other hand, I am not sure why the 120Hz peak suddenly surged back up unless it was caused by the Panaflo fan itself. Finally nothing would quite explain why the 1520Hz peak simply disappeared. It should have stayed the same. The best explanation that I have is that the microphone or Minidisc recorder boosted the gain when the noise level dropped. That may have skewed the results.

In any case, all of this became moot because by the second day I noticed persistent stuttering noises from the case fan. Eventually, it started occurring every few minutes for several seconds at a time. It was pretty quiet, but somewhat disturbing. I examined the fan while it ran and realized that the stuttering was in fact vibration. The fan appeared to become unbalanced on a regular basis. It was running strong enough that I had to apply strong pressure to stop it. Therefore, I don't think it was affecting its cooling capacity. As I poked and prodded at it in a vain attempt to fix it, the fan suddenly stopped working. Without so much as a wimper, it simply died.

Since the case fan is directly response for cooling the CPU and hard drive(s), I didn't want to take the risk of trying another 120mm case fan. On the other hand, I was so enamored with the results that I decided I couldn't go back. Instead, I decided to exchange my computer in the hopes that I would get a quieter one. So far, I think it was worth it.

Power Supply Fan Insert - Revisited

If you read the history of my Power Mac, you will know that I ended up exchanging my original Power Mac. Although the new one is quieter, I wasn't entirely satisifed. So, without making any other changes, I simply added the power supply fan insert. Here are the results:

Power Mac 2 spectrum	Power Mac 2 spectrum (with leather insert)

	Before		After		Change in Relative Decibels
	Decibels	Relative Decibels	Decibels	Relative Decibels
120Hz	-35 dB	0 dB	-44 dB	-2 dB	-12 dB
720Hz	-45 dB	-10 dB	-42 dB	0 dB	0 dB
1285Hz	-44 dB	-9 dB	-45 dB	-3 dB	-2 dB

So this pretty much confirms the results of the first power fan insert. The noise in the 120Hz range drops significantly. A little drop occurs at the high frequency. I am normalizing all of the decibel levels to that of the case fan. Maybe that isn't entirely scientific, but my personal opinion is that it works pretty well. I have done many recordings now and the noise levels do vary. Part of the reason is that I don't have perfectly a controlled environment, and my microphone is not ideal. Nevertheless, if you look at the big picture, it's clear that the power supply fan insert reduces noise at 120Hz.

Back to Main