- Joined
- Sep 29, 2011
- Location
- North Central Idaho
I've been shoveling the coal to an early run PhII 555 x2 3.1GHz for 2 years now. Cooler was an ARCTIC COOLING Freezer 64 Pro 92mm. I was one of the lucky ones: It popped open to 4 cores and ran fine at 3.8GHz (4.1 for about 20min on P95). I was overjoyed and set to work transcoding my DVD library for the media center. But, late last year I had to drop the mult to X18.5 (3.7GHz) and again in May this year to X18, then last month to X17.5 (or 3.5GHz). I worked the water out of it. It ran pretty much @ 50..53C when I was at the console.
I decided to have a go at mitigating the re-diffusion process that was going on on the die and purchased an Antec Kuhler h2o 920. I selected this unit (as I understand all Antec and Corsair cooling kits are made by the same OEM) cuz it looked identical to the favorably considered >> Corsair H100 except it had a three row radiator as opposed to the H100s two row radiator. <<
ed: Totally wrong. Comparison was w/ an H70 which indeed looks to have a three row radiator. What cinched the decision was that J&R had it on sale at $89+ w/ free shipping. BTW J&R must have deep stock on Antec coolers as I noticed the two row single fan unit on sale today.
Moving forward: I ran into problems right away with installation of the integrated WB/pump/controller module connection to the 'wire side' motherboard fastening assembly/bracket. The fastening bracket was virtually impossible to keep registered with the 'spigots' of the captive nuts fully seated in the motherboard through holes. Flashback: I should have removed the motherboard! I solved the problem, or so I thought, by placing a piece of foam rubber on my desk that would keep the fastening assembly in place by pushing up through the access opening to the cpu under side area in my case.
Engaging the WB assembly clip/bracket screws was difficult at best. Very easy to start fasteners cross threaded. Luckily the anti rotation facility provided by the injection molded retainer bracket was puny enough to not allow any damage to the clip screws or captive nuts below. Observation: Retaining screw pitch is way to high. Question: Why didn't they just throw in a hand full of screws that fit all the stiffener brackets for target installations? This should have been far less costly than molding a custom retainer part and making 4 stainless captive nuts.
Anyway, it all went together and temps went down to the mid 40s and I was able to bump the speed back up to 3.6GHz. Very happy, I was able to stave off an expensive cpu upgrade with a lesser upgrade that would migrate forward in future upgrades. Then the boot dropped.
Two weeks ago, I came in at my usual 0500hrs to turn up the network and my workstation. Workstation was dead, nada, zip, null. I was dumbfounded, the machine had been running perfectly and in it's new configuration for some +/- two weeks. With no feedback from the machine at all: I manually reset the bios (M4A78T-E), forced an old Sony monitor to stay alive w/ no input in VGA mode and still nothing. All I had left was isolation, inspection, then substitution. I first started disconnecting everything from the board but for one ram stick, no joy.
Inspection revealed that one of the captive nut spigots was not registered in the motherboard through hole. It was immediately clear that this has produced pressure on the motherboard that would try to bend it into a cup shape inside the area defined by the through holes. To test, I loosened the WB assembly mounting screws two turns at a time and tried to start the system. At the point where the mounting screws released 'all spring' in the WB assembly mounting clip, the system started.A last ditch attempt at saving things was tried by putting a drop of RTV Silicone at each of the corners of the bottom bracket to hold it in place and let it cure. No joy. As soon as "any" tension was developed in clamping the WB assembly down, the system would quit. I ordered a new board, M5A97-EVO. I realize fully that I have no one to blame for this incident but myself and accept full responsibility. The board was toast. One of the traces from some layer or through hole socket mounting plating had barfed.
While waiting for the new board, I studied the WB assembly mounting system and thought about how to make it better, for me, keep in mind. Thinking about the injection molded backing bracket with captive nut retainers, led me to understand that even when properly installed, the strength/stiffness of the connecting members between nut locations was not sufficient to support the motherboard against 'cupping' under load of the top WB assembly clip tension. Not looking for a scapegoat here just returning to my engineering roots. I tried fitting the original heat sink mounting bracket screws through the plastic insert towers/retainers that were fitted to the WB assembly clip and discovered that that would require a one way modification. I was prepared to purchase the correct length metric screws at the local hardware but not risk modifying those small plastic retainers. Then I realized that I had a spare steel bracket from the dead M4A78T-E and drill reamed out the captive nuts in it to slip fit the spigots of the captive nuts on the Astec supplied bracket. I glued the plastic bracket to the steel bracket with 4 drops of RTV Silicone and when the new board arrived, poof! I now have an adequately stiff backing plate for the cpu area and have not ruined anything but an old motherboard. yeowie!
A couple of 64GB Corsair M4s arrived w/ the new M5A97-EVO and waiting for the 8120 to arrive with a BD-r to transcode my Blue-Rays. I'll post those results on AMD-cpu area.
I decided to have a go at mitigating the re-diffusion process that was going on on the die and purchased an Antec Kuhler h2o 920. I selected this unit (as I understand all Antec and Corsair cooling kits are made by the same OEM) cuz it looked identical to the favorably considered >> Corsair H100 except it had a three row radiator as opposed to the H100s two row radiator. <<
Moving forward: I ran into problems right away with installation of the integrated WB/pump/controller module connection to the 'wire side' motherboard fastening assembly/bracket. The fastening bracket was virtually impossible to keep registered with the 'spigots' of the captive nuts fully seated in the motherboard through holes. Flashback: I should have removed the motherboard! I solved the problem, or so I thought, by placing a piece of foam rubber on my desk that would keep the fastening assembly in place by pushing up through the access opening to the cpu under side area in my case.
Engaging the WB assembly clip/bracket screws was difficult at best. Very easy to start fasteners cross threaded. Luckily the anti rotation facility provided by the injection molded retainer bracket was puny enough to not allow any damage to the clip screws or captive nuts below. Observation: Retaining screw pitch is way to high. Question: Why didn't they just throw in a hand full of screws that fit all the stiffener brackets for target installations? This should have been far less costly than molding a custom retainer part and making 4 stainless captive nuts.
Anyway, it all went together and temps went down to the mid 40s and I was able to bump the speed back up to 3.6GHz. Very happy, I was able to stave off an expensive cpu upgrade with a lesser upgrade that would migrate forward in future upgrades. Then the boot dropped.
Two weeks ago, I came in at my usual 0500hrs to turn up the network and my workstation. Workstation was dead, nada, zip, null. I was dumbfounded, the machine had been running perfectly and in it's new configuration for some +/- two weeks. With no feedback from the machine at all: I manually reset the bios (M4A78T-E), forced an old Sony monitor to stay alive w/ no input in VGA mode and still nothing. All I had left was isolation, inspection, then substitution. I first started disconnecting everything from the board but for one ram stick, no joy.
Inspection revealed that one of the captive nut spigots was not registered in the motherboard through hole. It was immediately clear that this has produced pressure on the motherboard that would try to bend it into a cup shape inside the area defined by the through holes. To test, I loosened the WB assembly mounting screws two turns at a time and tried to start the system. At the point where the mounting screws released 'all spring' in the WB assembly mounting clip, the system started.A last ditch attempt at saving things was tried by putting a drop of RTV Silicone at each of the corners of the bottom bracket to hold it in place and let it cure. No joy. As soon as "any" tension was developed in clamping the WB assembly down, the system would quit. I ordered a new board, M5A97-EVO. I realize fully that I have no one to blame for this incident but myself and accept full responsibility. The board was toast. One of the traces from some layer or through hole socket mounting plating had barfed.
While waiting for the new board, I studied the WB assembly mounting system and thought about how to make it better, for me, keep in mind. Thinking about the injection molded backing bracket with captive nut retainers, led me to understand that even when properly installed, the strength/stiffness of the connecting members between nut locations was not sufficient to support the motherboard against 'cupping' under load of the top WB assembly clip tension. Not looking for a scapegoat here just returning to my engineering roots. I tried fitting the original heat sink mounting bracket screws through the plastic insert towers/retainers that were fitted to the WB assembly clip and discovered that that would require a one way modification. I was prepared to purchase the correct length metric screws at the local hardware but not risk modifying those small plastic retainers. Then I realized that I had a spare steel bracket from the dead M4A78T-E and drill reamed out the captive nuts in it to slip fit the spigots of the captive nuts on the Astec supplied bracket. I glued the plastic bracket to the steel bracket with 4 drops of RTV Silicone and when the new board arrived, poof! I now have an adequately stiff backing plate for the cpu area and have not ruined anything but an old motherboard. yeowie!
A couple of 64GB Corsair M4s arrived w/ the new M5A97-EVO and waiting for the 8120 to arrive with a BD-r to transcode my Blue-Rays. I'll post those results on AMD-cpu area.
Last edited: