Last updated 06 March 2008
04/03/2008 - Bigger heat-sink
The solution to the CPU heat problem was always going to be a bigger heat-sink. A quick search on eBay produced a whopper for £20 - A 150mm long 750AB profile rated at 0.32 K/W and ideal for point heat sources, but at 135mm x 125mm and weighing in at 2.7kg, it’s almost too big.
For the initial testing I mounted the heat-sink atop the original copper block, using the same bolting method as the previous sink (fortunately, the new sink has T slots in the base that I could use). For the test I ran the machine in the BIOS monitor page for an hour and a half (plenty of time to warm the heat-sink) with the machine drawing 85watts from the mains, the CPU temperature remained below 60’C (about +25’C) throughout the test. Not great, but good enough to fit it properly.
To my great relief the chassis, complete with the new heat-sink, fitted back into the case easily. The only snag being that the heat-sink was too high (it protruded out the side of the case by 5mm or so. I didn’t want to cut the heat-sink so I looked into losing the 5mm from the CPU space block.
Removing the P4 heat-sink bracket from the mother board allowed me to lower the sink down the the height of the larger filter capacitors (about 25mm from the surface of the board).
I machined a new 18mm copper block from stock then flattened and lapped it to the heat-sink. The ‘flat’ bottom of the sink had a distinct hump down the middle where it had distorted after it was extruded. The new block will fit far better than the original did during the test.
An second aluminium block is used to connect the north-bridge chip to the new heat-sink; The chip’s original tiny finned cooler was in a dead-space between the CPU and AGP card and with the large heat-sink in place it was getting too warm for my liking.
A small clip on the right hand side helps to keep the north-bridge shunt in contact with the heat-sink while on the left a bracket to the chassis supports the heat-sink when the PC is moved.
Extra holes and some metal pruning help to keep the air flow restrictions to a minimum above the heat-sink.
Testing the completed machine using the BIOS page and running continuous bench marks failed to get the CPU over 60’C (<30’C over ambient) the new heat sink is a remarkable success.
The solution to the CPU heat problem was always going to be a bigger heat-
For the initial testing I mounted the heat-
To my great relief the chassis, complete with the new heat-
Removing the P4 heat-
I machined a new 18mm copper block from stock then flattened and lapped it to the heat-
An second aluminium block is used to connect the north-
A small clip on the right hand side helps to keep the north-
Extra holes and some metal pruning help to keep the air flow restrictions to a minimum above the heat-
Testing the completed machine using the BIOS page and running continuous bench marks failed to get the CPU over 60’C (<30’C over ambient) the new heat sink is a remarkable success.
