http://www.cofan-usa.com

43144 Christy Street • Fremont, CA 94538
Tel (800) 766-6097 • Fax (510) 490-7931
email: fans@cofan-usa.com

Kool-Cast^tm KC-266 for Celeron^tm

^{by John Bogush}

  It has been well proven that the heatsink and fan that Intel includes with the retail Celeron is adequate for overclocking at least up to 450MHz for most of the processors in the Celeron line.  It is on occasion that perhaps a stubborn Celeron 300a may be coaxed to that magical 450MHz with just a bit more cooling than the retail unit is capable.  The OEM Celeron leaves an individual with a choice of several different after market cooling solutions for their shiny new Celeron.  It is my opinion that most of the first available after market Celeron coolers were just quickly thrown together so the manufacturers had a product in place to snag the consumer's hard earned dollar.  Then in steps COFAN USA with their newest cooler called the KC-266 for the Celeron processor.

  The KC-266 first caught my attention when I received a Glacier 4500C for review and testing.  The Glacier 4500C is a triple fan Celeron cooler from which I discovered after removing the fans that the heatsink was actually from COFAN USA (more on the Glacier unit later).  Several other folks had also mentioned the KC-266 to me though I was not familiar with the product.     I shot an email over to COFAN USA about being interested in reviewing the KC-266 and I promptly received a reply requesting me to give COFAN USA general manager Dave Demos a phone call.  After giving Dave a call the next day we had a pleasant conversation with me stating my interests and intent with the KC-266 and him taking the time to explain the general engineering and development principals behind it.  Dave explained the KC-266 heatsink was of a die-cast aluminum with a milled processor contact surface and a very high quality coreless fan.  He also stated that the mounting process was very quick and simple requiring no tools other than the users own hands.    He promised a couple of samples would be at my door in a few days so I patiently waited.

  Mr. UPS arrived three days later with a package containing the KC-266 samples so I headed for the computer lab (spare bedroom) and got started.  Upon opening the package I found the KC-266 to be composed of a die-cast heatsink with a single 50mmx50mmx12mm fan mounted at the heatsinks center and a combination plastic and metal wire retaining mechanism.  The fan has a standard motherboard connector that is compatible with our beloved P2B and BH6/BX6 x motherboards.     I found the die-cast aluminum 413 heatsink to be of excellent quality free from any pitting and edge flashing found on so many die-cast products.  The processor contact patch was indeed milled relatively smooth with very slight grooves which appear to be from the machining process.  I could not "snag" the grooves with my finger nail and the quality of the surface far exceeded that of the ridge laden extruded aluminum heatsinks which are most common.  The ridges on the extruded heatsinks always run parallel to the direction that the heatsink exits the extrusion die and can exceed several thousandths of and inch causing poor thermal energy transfer making a considerable amount or thermal compound necessary.  The mounting scheme is very simple as stated by Dave and it only took me a few seconds to mount the KC-266 to a Celeron 300a.  One quick note here, writing on the boxes that the KC-266s were in that I received state that no thermal interface material is necessary.  I did not like that idea because by now I was well aware of the fact that the Celeron slugs themselves are not truly flat, the four corners of the slug surface are slightly raised causing an air space between the processor and any heatsink.  I went ahead and began testing the KC-266 without thermal compound as directed on the packaging and soon found that it was a "no go" as at 450MHz and 2.0 volts the 300a quickly heated up causing stability problems.  I decided that I would do all testing with the standard silicone/zinc thermal compound I would normally use from this point on.

  My reference system booted up and loaded Windows98 at 450MHz with no problems noted.  I went ahead and tried 464MHz and 504MHz  both at a core voltage setting of 2.0 volts with again no problems noted.    I decided it was time for some stress testing so at 504MHz and 2.0 volts I started Half-Life and began to loop the Blowout demo.    During the demo at the appropriate time I recorded the processor temperature by measuring it with my Fluke noncontact thermal temperature sensor on the rear of the processor.  This technique differs from my usual heatsink temperature recordings because I believe it is a better indication of performance to measure the processor itself rather than the cooling device.

   All temperatures are recorded in Celsius.  The data was collected with the 300a running under the same load and the same ambient air temperature with each heatsink.  Surprisingly even to myself the KC-266 exceeded the performance of even the Global WIN FAB04 single fan which was up till now the best performing single fan cooler I have tested.  The fin design of the KC-266 is a duct system which directs the air flow from the fan in a deliberate manner.     The area under the fan is even sloped to help make the air flow as efficient as possible.  Also note that the KC-266's fan is a 12mm thick unit which is 2mm greater than the standard 10mm fan found on most heatsinks

  Before I finished this review I did email Dave and ask him about the subject of the Celeron's uneven slug surface and COFAN USA's recommendation of using no thermal interface material.  Dave responded to me by telephone and said that the early engineering blanks that were provided by Intel did indeed have a very flat and even surface which as we all now know obviously differ from the final product.  He said that COFAN USA would make a thermal interface material available for use with the KC-266 and mentioned a product by Chromerics called T7-10 which he described as a two sided adhesive thermal pad which will conform to the surface or the Celeron at operating temperature creating efficient thermal transfer between the processor and the KC-266.  It was a pleasant surprise to see COFAN USA respond to this reviewer's feedback with such a positive response and I believe it is COFAN USA's intent to deliver the best performing product possible.

  All in all the KC-266 did not disappoint me.  It is the best performing single fan heatsink I have tested thus far and I would definitely recommend it's consideration by anyone looking for a quiet single fan Celeron cooling solution.

A final note:

  I mentioned the Glacier 4500C which I had previously tested.  The Glacier 4500C did perform well and even gave my trusty Global WIN FAB24 a run for it's money but at around $35 it does not give you what you pay for in my opinion.  As evident in my full version heatsink comparison chart it does not out perform the single fan KC-266 by much and the KC-266 reportedly sells for around only $15.  I asked Dave Demos of COFAN USA about the Glacier 4500C and he said that they had only recently become aware of the modification to their product resulting in the Glacier unit and that the extra two would fans impede the airflow of the center fan resulting in poor performance as to what would be expected from an expensive multi fan unit.

KC-266 DESCRIPTION 

Thermal Characteristic 

Heat Sink 

Integrated Fan 

Miscellaneous