View Single Post
  #16 (permalink)  
Old 27th November, 2002, 01:40 AM
CamoAlien's Avatar
CamoAlien CamoAlien is offline
Member
 
Join Date: February 2002
Location: Northern Minnesota
Posts: 72

Rudzer & nelamvr6 - Hi. Impressive MHz count you have there.

Now, you have a very valid point about the Vcore. By all means, it should be able to reach 2.0 volts at the 2.0 volt setting. However, might it be possible that your board actually is functioning correctly, and is not failing to provide the correct voltage? Let me elaborate, using the electrical approach.

Given: You have an overclocked Athlon XP (T-bred core) 2400+ @ 2236 MHz.
You are running a very high FSB (most likely irrelevant)
You want the 2.0 volt setting to work
At the speed that your CPU is running, it is drawing a massive amount of current, and is producing a very large amount of heat (which is why they make giant heatsinks and water blocks and peltier chips).

Now, say you set it to 2.00 volts. However, it gives you 1.92 volts. This is exactly 4% under what it should be. Sure, 0.08 volts may seem like alot, but it is only 4% under.

Now think about this: you have a bunch of computers running, and all of the lights in the room are on. You pop something in the microwave, and the lights get a bit dimmer. Why does this happen? Voltage drop, and it happens when you are drawing a large amount of power (or a large number of amperes). Normal voltage for your standard wall socket in North America should be between 120 and 130 volts. Drawing a large amount of current could easily cause it to drop to between 110 and 115 volts (in other words, a 13% drop in voltage).

Similarly, at this speed, your CPU is drawing a large amount of power, and is causing a voltage drop. Those VRM's are doing their best to supply it with 2.00 volts, but they are putting out 1.92. The industry specifications for most electronics consider a 5% deviation as a low tolerance, and you are 4% below, which is within spec.

The other thing is that it might not even be the VRM's. It might even be the traces on the PCB or even heat buildup within the CPU traces that cause a wee bit of resistance that would drop the voltage a little bit.

Add to it how inaccurate the voltage sensors are in the first place, and you can see why sometimes it may appear that you aren't getting what you are paying for.


Anyhow, if you are seriously concerned about the voltage issue, it may be worth a look with a Fluke digital multimeter. You could also check the Vcore lines with an oscilliscope to watch for excessive ripple, which is another source of poor voltage regulation. Then, if necessary, either solder a varistor or use a clip-on wire on the IC which determines the Vcore (can't quote the IC # for you, haven't kept up on this one) and manually adjust it.

Good luck.
__________________
Rei Ayanami - 2x 850e | SGI Zx10 | 3d Prophet II GTS 64m DVI | SGI 19" CRT | 1024mb RAM | IBM 36Z15 | Fujitsu MAP3147NP | Enermax FS-981xpss & eg651p-ve | 3Com 3c996b-t
Misato - XP-M 2600+ | K7S5A | 1024mb PC3200 | FX5950U | WD1600JB - See them all
Asuka - 2x 667b | Supermicro P3DRE | FX5200 | SGI 19" CRT | 384 RDRAM | 3Com 3c996b-t | PX716-A | HP 9200s

Last edited by CamoAlien; 27th November, 2002 at 01:50 AM.
Reply With Quote