I'm putting together a system based on a KT7A and a T'Bird 1.2G. Have been having issues w/ an idle temp. of 48 C. (kinda' high?).
Anyway, last night I spent a long, long time methodically testing a variety of fan orientations on an Alpha PAL3065. The findings were interesting and rather illuminating(?).
1. A copper shim will raise CPU temp. readouts by 2 C., with and without a translucent coating of Arctic Silver on the HS (CPU coated similarly, of course). This confirms similar findings on other sites.
2. There is no significant difference in having the HS fan blowing towards/away from the CPU (1< C. in favour of blowing away--this is specific to an Alpha PAL6035, which isn't designed to work w/out a shroud).
(At this point the HS was reseated w/out the shim.)
3. There was no difference in the effectiveness of a Y.S. Tech. fan over a Sanyo Denki in either orientation (both obtained from Plycon.com). All orientations (towards/away from) resulted in 45 C. CPU temp/28 C. system temp. after ten minutes w/ case open. Closing the case resulted in a 1 C. rise consistently after five minutes.
Further to 3:
Of note is that the Sanyo Denki appears to be as effective as Y.S. fan, although the latter provides 8.1 cfm more than the Sanyo Denki (27.2 vs 19.1 cfm). Conversely, the Sanyo Denki is 7dBA quieter (28 vs 35 dBA)--the choice is obvious if noise control is an issue.
The results are specific to the following specs.
* Abit KT7A (WW BIOS, Fail-safe defaults) ; T'Bird 1.2G ; 384 Mushkin v. 1.5;
* Alpha PAL6035 (lapped) & Arctic Silver (translucent coating)
* Hercules Prophet II Ultra w/ stock fan
* Adaptec 19160: Cheetah 18XL 9.1G mounted in Coolermaster 4002 HD cooler (top bay); TEAC floppy
* Lian Li PC-60 (all-aluminum) case: 2 ADDA 80 mm. intake; 1 ADDA 80 mm. exhaust)
* Enermax EG451P-VE power supply (92 mm. somewhat directly over CPU; 80 mm. exhaust)
Postscript: What is perplexing is that the config. is geared towards cooling efficiency, but the CPU temp. remains consistently on the high side at idle. The flatness of the lapped HS was confirmed w/ two different straight edges, and in all instances the HS paste was applied as recommended.
This suggests that the KT7A's CPU temp. sensor (at least mine), or the board as a whole, is out of whack. This may explain the similar results between the Sanyo Denki and Y.S. Tech fans. Any suggestions; observations; insights; did I miss something?
Anyway, last night I spent a long, long time methodically testing a variety of fan orientations on an Alpha PAL3065. The findings were interesting and rather illuminating(?).
1. A copper shim will raise CPU temp. readouts by 2 C., with and without a translucent coating of Arctic Silver on the HS (CPU coated similarly, of course). This confirms similar findings on other sites.
2. There is no significant difference in having the HS fan blowing towards/away from the CPU (1< C. in favour of blowing away--this is specific to an Alpha PAL6035, which isn't designed to work w/out a shroud).
(At this point the HS was reseated w/out the shim.)
3. There was no difference in the effectiveness of a Y.S. Tech. fan over a Sanyo Denki in either orientation (both obtained from Plycon.com). All orientations (towards/away from) resulted in 45 C. CPU temp/28 C. system temp. after ten minutes w/ case open. Closing the case resulted in a 1 C. rise consistently after five minutes.
Further to 3:
Of note is that the Sanyo Denki appears to be as effective as Y.S. fan, although the latter provides 8.1 cfm more than the Sanyo Denki (27.2 vs 19.1 cfm). Conversely, the Sanyo Denki is 7dBA quieter (28 vs 35 dBA)--the choice is obvious if noise control is an issue.
The results are specific to the following specs.
* Abit KT7A (WW BIOS, Fail-safe defaults) ; T'Bird 1.2G ; 384 Mushkin v. 1.5;
* Alpha PAL6035 (lapped) & Arctic Silver (translucent coating)
* Hercules Prophet II Ultra w/ stock fan
* Adaptec 19160: Cheetah 18XL 9.1G mounted in Coolermaster 4002 HD cooler (top bay); TEAC floppy
* Lian Li PC-60 (all-aluminum) case: 2 ADDA 80 mm. intake; 1 ADDA 80 mm. exhaust)
* Enermax EG451P-VE power supply (92 mm. somewhat directly over CPU; 80 mm. exhaust)
Postscript: What is perplexing is that the config. is geared towards cooling efficiency, but the CPU temp. remains consistently on the high side at idle. The flatness of the lapped HS was confirmed w/ two different straight edges, and in all instances the HS paste was applied as recommended.
This suggests that the KT7A's CPU temp. sensor (at least mine), or the board as a whole, is out of whack. This may explain the similar results between the Sanyo Denki and Y.S. Tech fans. Any suggestions; observations; insights; did I miss something?
Facebook
Twitter
Instagram