Here's my DIY recipe to lessen the noise done by a dual 867 G4 MDD ! Coz even with Apple replacement kit, things didn't sound good enough 2 me. I've been perfecting it and testing it thoroughly 4 weeks, and now feel just like I've reached the desired sound level! First of all, a little reminder: when one boots a MDD under OS X, it's not that noisy at the beginning. With OS 9, it's already a bit noisier, coz the firmware controlling the central 120mm fan is not as efficient that under OSX, even using the last upgrade we were offered 6 months ago or so. Then, under OS X, depending on the work, high or low CPU request, as well as the ambient temperature, the fans get turning quicker and quicker, producing a quite displeasing incremental sound level. My goal was to reduce the overall beast sound level, but also to make it more constant by some means. Many solutions exist, often quite expensive (Verax kit costs around 250 euros) or a modification in the computer, which voids the warranty. I found inspiration in a XLR8yourmac article found on MacBidouille [french version of hardmac.com for those who wouldn't know yet :-) NdT}: http://www.xlr8yourmac.com/systems/mdd_cooling_mods. But even there I wasn't glad with the results, so I tried to go further. To sum up the process : - central fan is unplugged from its variable power supply located on the motherboard and plugged back to the second power supply (of constant power) used by the optical drives, thru a 3 pins -> 4 pins Molex adapter with a 5V tension. Between the two of them, I placed a potentiometer in order to regulate this fan rotation speed according to circumstances. One can also do without this latter, the fan noise stays reasonable, but above all it's CONSTANT! (see photo 1) - adding a second fan (NoiseBlocker S2 80mm) directly on the CPUs heat sinker , linked to the second power supply of the ATA-100 bay via a 12 V, 3 pins -> 4 pins Molex adapter (this fan is so silent, one can go frankly with it !). To fix it to the radiator, I just did what they have suggested in the article : I used 4 "slics" rubber fixations, which don't need to be screwed and therefore limit vibrations. I tested the two possible positions, the one where the air flow is TOWARDS the CPUs and the opposite one, the latter giving the best results : the fan sticker has to be face to the top, not to the bottom, when it is fixed to the heat sinker. The solution given by the article, where you get the 2x 60mm fans from the former G4 power supply is valuable, but 4 slightly worse results because of their noise : they get really noisy when tension is over 7 V , thus not so much air is boosted than with S2 with12V which one could almost not hear ! (see photo 2) - adding a new 80mm fan behind the case, so that hot air is repelled from the G4. I, for once, used a NoName fan that I had in stock, which I linked to the same power supply than the S2 but this time with a 7V Molex adapter, so that it's silent. Even using a S2, it's useless to use more power, since above that level, the air output is so high that only the air flow going through the case holes causes much noise. I fixed that fan with classical cable holders passed thru the case holes, after recycling small sticking rubber pieces from one of the two fans of the former power supply ; I fixed them at each of the 4 corners of this 80mm fan, to reduce vibrations, hence noise. If one doesn't want to trickle the Molex to get the tension from 12V to 5 or 7V (it's easy enough though : one just has to invert the yellow and red cables for 5V using tweezers to extract them from white plastic), there are adapters to do that, but a potentiometer, such as the one delivered with the "BOX" version of the S2, will do (see photo 3). Results : - the noise due to my G4 is now quite alleviated : low and above all, constant. I managed to find the perfect central fan speed so that it both is silent enough and lets enough air in. - long hours spent using DnetC for calculations (when you're an user you know how much it might cause a machine to heat up !) and measuring thru Loudometer (what an incredible soft !) give me a maximal temperature of 58° C. And in idle mode (computer turned on many hours without doing anything), 51° C max. I have to mention that ambient temperature is 28° C (I use the current summer heat to improve my settings !). So NO RISK at all to see the CPU burn or even malfunction. Therefore I've had ABSOLUTELY NO CRASH ever since I've tried those setting adjustments. Drawbacks : - there's room left for only one HD in the bay linked to the IDE ATA-100 bus, because of the fan on the radiator - it's a bit tricky not to be overwhelmed by all those cables in the machine, especially to avoid torsions when closing it or contacts with the heat sinker that still gets quite hot. - one has to be a DIY man, especially to fix the fan behind the machine and get the cables thru the door, as well as some threads with the molex, but yet I think many beginners will be able to do so. Advantages : - LOW and CONSTANT noise despite CPU request and ambient temperature - a very hot HD can be used in the ATA-100 bay, as the fan on top of the radiator boosts its air directly on it. - This whole assembly is totally reversible, the G4 bearing no scar. All that belongs to the G4 itself remains UNTOUCHED, so no problem with the warranty in case of a breakdown ! - it's really cheap ! when considering every single piece that has to be bought : 1) 2x NoiseBlocker S2 BOX = 30 euros 2) 1x slics fixations = 3 euros 3) adapters : 4 to 10 euros, and useless if you use the ones provided with S2 BOX. |
