Viewing page 2 of 3 pages. Previous 1 2 3 Next
Installation on a P4 Motherboard
Here is a photo showing the Kamakaze reconfigured for P4-478 use. The conversion process required a small phillips-head screwdriver and just 5~10 minutes.
At first glance, the plastic HS retention bracket appears identical to the ones that come on most P4-478 motherboards. Yet, the instruction sheet states:
You must replace the plastic retention bracket on the motherboard to the included with the included retention bracket in order to have a HS fit.
A close examination of the standard bracket vs. the supplied bracket revealed that the latter allows slightly more room on the inside perimeter, by shaving off a bit of plastic around the 4 corner locking tabs. The outer dimensions are the same. The HS is just big enough that it does not fit into the standard retention bracket, and it fits with a tiny bit of room to move in the supplied bracket. See the closeup comparison below. The bracket on the left is the one supplied with the Kamakaze.
This mean that in order to use the Kamakaze in a P4 system, the motherboard must be removed from the case or the dedicated HS retention bracket mounted before the motherboard is installed.
Swapping out the standard HS retention bracket on the test bench Intel motherboard was a bit of a pain, and the instruction sheet has nothing to say about the procedure. It consistes of popping out 4 plastic friction grommets, then releasing the bracket from the motherboard. The procedure is then reversed for the Kamakaze-supplied HS retention bracket.
In my sample, no grommets or screws or any way to secure the new retention bracket was supplied; this is probably an anomaly, given the nice packaging. (It is even possible that a bag of hardware fell unseen into the oblivion of the SPCR test room and was lost forever.) Luckily, the SPCR test bench is not without resources in odds and ends so the retention bracket was installed without problems.
It's a question I asked myself repeatedly while installing the HSF. The instruction sheet says:
Turn screws on both sides evenly, in clockwise to fasten the heat sink. Gradually tighten the screws until you feel the bars reach the top. Do not over tighten them, it will damage the screws and screw heads.
"Till the bars reach the top..." I really was not confident about that line. The photo below shows you why.
The bar referred to in the instruction sheet is about third of a centimeter from the top, and you can see it is quite bent already. There is a slight bow in the PCB of the botherboard beneath the HSF / CPU. Did it really need any further tightening? I did not think so. Pulling the bars all the way to the top seemed to be asking for trouble. Both the bars and the board would be bowed quite a lot and be under high tension. The bars on both sides were set to about the height shown above.
As a final check, at the end of a 100% CPU load run, the bar height was increased by several screw turns so that the distance "to the top" was halved compared to what is shown above. If the tension had been inadequate, there would be a drop in CPU temp within a couple of minutes, perhaps faster. As expected, there was no change in CPU temps.
The fan size allows the use of our standard heatsink testing method of a Panaflo 80mm low speed fan at 12, 7 and 5 volts. (This procedure allows heatsinks to be compared directly without the variable of different fan airflow / noise.) However, because of the high degree of integration of components in this HSF package, testing was also done with the provided fan.
The test was conducted on a P4 platform. There was inadequate time to do a complete socket-A test as well. It may appear as an addendum to this review in the near future. Here's an overview shot of the testing platform:
Key Components in P4 HS Test platform
Intel P4-2.53 Nominal power is 61.5W; may increase to ~75W if speed throttling doesn't stop it first. 71° C rated maximum junction temp. We'll pull the plug if any HS let's the temp go much above 65° C. NOTE: Recently upgraded from P4-1.8A CPU.
Intel D845PEBT2 motherboard - Intel 845PE Chipset; on-die thermal diode monitoring
Panaflo FBA08A12L1A 80mm DC fan and/or stock fan supplied with HS
Any VGA card (AGP)
256 MB DDRAM - PC2100 generic
Any hard drive (in Smart Drive from Silicon Acoustics)
DigiDoc5 w/ thermal sensors
Any Good PSU
Zalman Multi-Connector (ZM-MC1)
Arctic Silver 3 Thermal Compound - Arctic Silver is widely regarded as the best thermal compound available. The Arctic Silver people provide detailed instructions on how to get the best of their product, and these instructions are adhered to religiously. Except for one point: They caution that it...
"...will take a minimum of 72 hours, and as many as 200 hours to achieve maximum particle to particle thermal conduction and for the heatsink to CPU interface to reach maximum conductivity. (This period will be longer in a system without a fan on the heatsink.) The CPU's temperature will drop as much as 2C to 5C over this "break-in" period."
We just can't afford that kind of time, so measurements are usually taken anywhere from half an hour to a couple of hours after application. Slap our wrists and subtract 2C from all our temp readings if you wish.
CPUBurn is the stress program used to load the CPU to 100%. It heats up the CPU a bit more than just about any other utility tried so far.
The system was allowed to cool between tests for ~5 minutes with the HS fan at full speed.
|Help support this site, buy from one of our affiliate retailers!|