David Craig
The 20" iMac G5 fan noise issue continues to grate on my nerves, as I have recently been working with large images and taxing the processor. I got so irritated one night that I decided to remove the fans to examine them in an attempt to understand the root cause of the problem.
I actually dissembled the mechanical components of the "CPU fan" (the noisiest of the two).I refer to the "CPU fan" as the one near the top center of the chassis and ducted to the G5 processor heat sync.
The "system fan" is the one just to the left of that when viewed with the back removed. (The hard drive fan is located in the lower intake grill, is fairly quiet and of a different design. It was not included in my dissection and is not bothersome to me).
While I am not a mechanical engineer nor cooling fan designer, and I still don't know exactly what's going on, I have a theory and some facts that I can share with you, based on my scientific curiosity about the problem.
The fans are actually easy to remove. They are held by small torx-type screws. It's crazy that Apple has us replacing entire mid-plane assemblies in an attempt to isolate this problem.
The fans are manufactured by Delta Electronics. They are similar in construction to each other. They both bear model numbers BFB0812H, apparently a Delta Model number.
The model numbers are not listed on Delta's web page, but similar model numbers are. I'm guessing that these are custom variations of a Delta Model that where manufactured specifically for Apple to fit the physical requirements of the iMac.
The Apple part numbers are: "system fan" 603-5521 rated 12v DC @ 0.50A. "CPU fan": 603-5459 rated 12v DC @ 0.36A.
The dissection:
The "CPU fan" was disassembled. The goal was to do this nondestructively so I could still use my iMac! This was delicate work and I was very careful.
The fan is a DC brushless type. The fan blade has a metal shaft of about 0.116" diameter that runs through the center of the assembly and has a grove on its far end. The shaft passes through a coil spring that places pressure against the inside of the hub and the top ball-bearing assembly. The shaft passes through the iron poles, central tube, and the bottom bearing. The retaining clip fits the grove at the end (as it emerges from the bottom bearing) and keeps the fan blade from popping out.
A bar magnet is formed into a circle to exactly fit the hollow inside circumference of the fan blade hub. The material looks like the flexible type of magnetic material you would find on the back of a refrigerator magnet, only thicker - about 1/8" thick and maybe 1/4" wide. This rotates with the fan over the four iron poles.
There are four stationary, laminated iron stators (or poles, or whatever the terminology is) that the hub fits over, each forming 90 a degree arc of the circle and separated by a small gap. Each pole is wound with a coil of magnet wire. Inside the poles is the central plastic tube molded into the fan housing that holds the bearings and through which the fan shaft passes.
There is a small, round circuit board with a hole that fits over the central tube, and under the poles. It's about the same diameter as the circle formed by the poles. It appears to be glued or molded into the plastic fan housing - I didn't dare try remove this for I think that would have been destructive. However, I could see on the exposed (phenolic side) of the board a single, small semiconductor. I believe this to be a hall-effect device used for sensing the fan RPMs. It is located just below the circumference of the rotation of the permanent magnet in the hub. There are four wires brought out from the circuit board, terminating in a connector that plugs into the iMac mid-plane board. I'm guessing these connections are power to the motor and an output from the rpm sensor.
Peering under the small space between the edge of the circuit board and the surface of the housing, I could identify some low-profile surface-mount components near its edge, including at least one 8-pin IC.
The two small ball bearing assemblies seem to operate very smoothly and are made of polished steel. I removed the bearing assemblies. I can rotate the hollow center shaft by inserting the end of a sharpened pencil while holding the bearings. They are much like the larger ball bearing assemblies you might find in a hardware store. There's a slight residue of some lubricating film that could be seen on the surface of the bearing assembly. There is some small but noticeable lateral play between the central hollow shaft of the bearing assembly and the outer bearing housing that presumably contains the ball (or roller?) bearings. The bearing assemblies have an outer diameter of 0.315" and thickness of 0.157". I couldn't measure the inside diameter of the hollow, central shaft, but it's obviously slightly larger than the 0.116" diameter fan shaft.
Although someone on the Apple discussion boards has claimed to cure their fan noise problems by "lubricating the fans", I resisted the temptation to immerse the bearings in some lubricant like silicone or such, fearing that I would just contaminate the existing lubrication and ruin it. The bearings seemed quite smooth already - I don't think additional lubrication would have helped in my case.
Observations:
The fan seems to be well designed. It has active electronics and sensing circuitry. It is not a simple DC motor that you might buy at Radio Shack. There seems to be no rubbing between the plastic fan blade and housing. There seems to be sufficient clearance between the magnet inside the hub and the four iron poles such that there's no rubbing or binding. Using compressed air to spin the fan blade seems not to generate the grating noise, although - I can feel a vibration as I hold the housing. I believe that what I'm feeling is the from the following effect:
There is a very strong magnetic attraction between the permanent magnet and the iron poles. When the fan is assembled, slowly rotating the fan by hand shows that there is a definite and strong affinity for the fan to stop in one of the four 90 degree positions. This is apparently where the north and south poles of the circumferential bar magnet come together - and where the center mass of the iron poles close this magnetic circuit.
My theory:
The noise is not caused by mechanical rubbing of the fan blade and housing or other components. Instead, I think that as the fan rotates at high speeds, there is a slight lateral wobble caused by the play I noted in the bearings.
The wobble is due to the magnet passing by the four iron poles and being strongly pulled toward that pole (or perhaps being repelled as the coils are actively pulsed by the driving circuitry). In any case, I believe it comes down to the lateral play in the ball bearings.
I believe the noise is emanating from the bearings themselves as they move laterally ever so slightly, but at high speeds. Perhaps the noise is then being amplified or resonating within the plastic fan housing as the vibrations are transmitted through the structure.
If this theory is correct, one might expect the noise to have a strong frequency component at four times the rotational speed of the fan. I've yet to actually test this.
By the way, I carefully reassembled my fan and the computer with no ill effects. The noise level is as it was. The whole operation was done carefully and was nondestructive.
A solution?:
Regardless of the actual cause - bad bearings or whatever, the fan noise is real and it is especially irritating.
Apple should take Delta Electronics to task on this issue or simply accept blame themselves for choosing these fans. Although the ball bearings seem like a nice design feature, perhaps a higher quality bearing with less lateral play would solve the problem, or perhaps sleeve type bearings would work better - if these are practical in this application. As I mentioned, I'm not a mechanical engineer. This is all just based on my scientific curiosity.
In any case, Apple should fix this problem once and for all!They should do it by offering a kit to their customers containing redesigned fans with the simple tools and instructions that would be required for replacing the "system fan" and "CPU fan". There's no need for people to be replacing their mid-planes. Having replaced one myself, I can tell you that it would be quicker and easier to replace the fans themselves.
Apple should do this because it's the right thing to do. They haven't been giving us any answers. I love Macintosh computers, but if this were a PC I'd have tossed it out the window by now. Loyal Apple customers such as myself should not have to go to such lengths as reverse-engineering a fan. These fans ruin an otherwise wonderful, beautiful computer and sour the whole Macintosh experience. I would think Apple would learn this lesson from past experience, like the "wind tunnel" Mac. Good grief!
The 20" iMac G5 fan noise issue continues to grate on my nerves, as I have recently been working with large images and taxing the processor. I got so irritated one night that I decided to remove the fans to examine them in an attempt to understand the root cause of the problem.
I actually dissembled the mechanical components of the "CPU fan" (the noisiest of the two).I refer to the "CPU fan" as the one near the top center of the chassis and ducted to the G5 processor heat sync.
The "system fan" is the one just to the left of that when viewed with the back removed. (The hard drive fan is located in the lower intake grill, is fairly quiet and of a different design. It was not included in my dissection and is not bothersome to me).
While I am not a mechanical engineer nor cooling fan designer, and I still don't know exactly what's going on, I have a theory and some facts that I can share with you, based on my scientific curiosity about the problem.
The fans are actually easy to remove. They are held by small torx-type screws. It's crazy that Apple has us replacing entire mid-plane assemblies in an attempt to isolate this problem.
The fans are manufactured by Delta Electronics. They are similar in construction to each other. They both bear model numbers BFB0812H, apparently a Delta Model number.
The model numbers are not listed on Delta's web page, but similar model numbers are. I'm guessing that these are custom variations of a Delta Model that where manufactured specifically for Apple to fit the physical requirements of the iMac.
The Apple part numbers are: "system fan" 603-5521 rated 12v DC @ 0.50A. "CPU fan": 603-5459 rated 12v DC @ 0.36A.
The dissection:
The "CPU fan" was disassembled. The goal was to do this nondestructively so I could still use my iMac! This was delicate work and I was very careful.
The fan is a DC brushless type. The fan blade has a metal shaft of about 0.116" diameter that runs through the center of the assembly and has a grove on its far end. The shaft passes through a coil spring that places pressure against the inside of the hub and the top ball-bearing assembly. The shaft passes through the iron poles, central tube, and the bottom bearing. The retaining clip fits the grove at the end (as it emerges from the bottom bearing) and keeps the fan blade from popping out.
A bar magnet is formed into a circle to exactly fit the hollow inside circumference of the fan blade hub. The material looks like the flexible type of magnetic material you would find on the back of a refrigerator magnet, only thicker - about 1/8" thick and maybe 1/4" wide. This rotates with the fan over the four iron poles.
There are four stationary, laminated iron stators (or poles, or whatever the terminology is) that the hub fits over, each forming 90 a degree arc of the circle and separated by a small gap. Each pole is wound with a coil of magnet wire. Inside the poles is the central plastic tube molded into the fan housing that holds the bearings and through which the fan shaft passes.
There is a small, round circuit board with a hole that fits over the central tube, and under the poles. It's about the same diameter as the circle formed by the poles. It appears to be glued or molded into the plastic fan housing - I didn't dare try remove this for I think that would have been destructive. However, I could see on the exposed (phenolic side) of the board a single, small semiconductor. I believe this to be a hall-effect device used for sensing the fan RPMs. It is located just below the circumference of the rotation of the permanent magnet in the hub. There are four wires brought out from the circuit board, terminating in a connector that plugs into the iMac mid-plane board. I'm guessing these connections are power to the motor and an output from the rpm sensor.
Peering under the small space between the edge of the circuit board and the surface of the housing, I could identify some low-profile surface-mount components near its edge, including at least one 8-pin IC.
The two small ball bearing assemblies seem to operate very smoothly and are made of polished steel. I removed the bearing assemblies. I can rotate the hollow center shaft by inserting the end of a sharpened pencil while holding the bearings. They are much like the larger ball bearing assemblies you might find in a hardware store. There's a slight residue of some lubricating film that could be seen on the surface of the bearing assembly. There is some small but noticeable lateral play between the central hollow shaft of the bearing assembly and the outer bearing housing that presumably contains the ball (or roller?) bearings. The bearing assemblies have an outer diameter of 0.315" and thickness of 0.157". I couldn't measure the inside diameter of the hollow, central shaft, but it's obviously slightly larger than the 0.116" diameter fan shaft.
Although someone on the Apple discussion boards has claimed to cure their fan noise problems by "lubricating the fans", I resisted the temptation to immerse the bearings in some lubricant like silicone or such, fearing that I would just contaminate the existing lubrication and ruin it. The bearings seemed quite smooth already - I don't think additional lubrication would have helped in my case.
Observations:
The fan seems to be well designed. It has active electronics and sensing circuitry. It is not a simple DC motor that you might buy at Radio Shack. There seems to be no rubbing between the plastic fan blade and housing. There seems to be sufficient clearance between the magnet inside the hub and the four iron poles such that there's no rubbing or binding. Using compressed air to spin the fan blade seems not to generate the grating noise, although - I can feel a vibration as I hold the housing. I believe that what I'm feeling is the from the following effect:
There is a very strong magnetic attraction between the permanent magnet and the iron poles. When the fan is assembled, slowly rotating the fan by hand shows that there is a definite and strong affinity for the fan to stop in one of the four 90 degree positions. This is apparently where the north and south poles of the circumferential bar magnet come together - and where the center mass of the iron poles close this magnetic circuit.
My theory:
The noise is not caused by mechanical rubbing of the fan blade and housing or other components. Instead, I think that as the fan rotates at high speeds, there is a slight lateral wobble caused by the play I noted in the bearings.
The wobble is due to the magnet passing by the four iron poles and being strongly pulled toward that pole (or perhaps being repelled as the coils are actively pulsed by the driving circuitry). In any case, I believe it comes down to the lateral play in the ball bearings.
I believe the noise is emanating from the bearings themselves as they move laterally ever so slightly, but at high speeds. Perhaps the noise is then being amplified or resonating within the plastic fan housing as the vibrations are transmitted through the structure.
If this theory is correct, one might expect the noise to have a strong frequency component at four times the rotational speed of the fan. I've yet to actually test this.
By the way, I carefully reassembled my fan and the computer with no ill effects. The noise level is as it was. The whole operation was done carefully and was nondestructive.
A solution?:
Regardless of the actual cause - bad bearings or whatever, the fan noise is real and it is especially irritating.
Apple should take Delta Electronics to task on this issue or simply accept blame themselves for choosing these fans. Although the ball bearings seem like a nice design feature, perhaps a higher quality bearing with less lateral play would solve the problem, or perhaps sleeve type bearings would work better - if these are practical in this application. As I mentioned, I'm not a mechanical engineer. This is all just based on my scientific curiosity.
In any case, Apple should fix this problem once and for all!They should do it by offering a kit to their customers containing redesigned fans with the simple tools and instructions that would be required for replacing the "system fan" and "CPU fan". There's no need for people to be replacing their mid-planes. Having replaced one myself, I can tell you that it would be quicker and easier to replace the fans themselves.
Apple should do this because it's the right thing to do. They haven't been giving us any answers. I love Macintosh computers, but if this were a PC I'd have tossed it out the window by now. Loyal Apple customers such as myself should not have to go to such lengths as reverse-engineering a fan. These fans ruin an otherwise wonderful, beautiful computer and sour the whole Macintosh experience. I would think Apple would learn this lesson from past experience, like the "wind tunnel" Mac. Good grief!