Voltage Fluctuations

BradyT88

Distinguished
Feb 25, 2010
166
0
18,690
I have my PhII 925 OCed to 3.78GHz @ 1.55v and when I am running P95 my voltage fluctuates bewtween 1.2 and 1.52v. Everest CPU only stability test will cause these same fluctuations but they are less frequent. Is this normal?
album.php
 
If your computer is stable then it should be ok. This effect is called Vdroop. More information here:

http://www.nvnews.net/vbulletin/showthread.php?t=107986

Although it states Vdroop for Intel CPUs it is similar for AMD.

Also as you noticed your CPU drops to 1,100MHz when not in full load to preserve energy. Hence it also drops the Voltage as it does not need to run at 1.52V to maintain a 1.1GHz speed.
 
Solution
Hmm. try some other programs as i have seen benchmarks that supposedly load the CPU 100% but it turned out that it was 98% or something less and in that 2% the CPU-mobo had time to get the voltage down. It could also be an overheating issue? Thermal management? Do you experience any stability issues? If yes then that is what it probably is. If not, try monitoring with another program and definitely try another benchmarking program and monitor the behaviour.
 
Well I see these flunctuations in CPU-Z and HWMonitor as well as everest. When running P95 I get those flunctuations about 3 times more than I doing running everest stability test. I am going to experirment with 235x14.
 


Man that was a great read I have read alot of different overclocking guides but that was by far one of the easiest ones to understand i actually learned a couple new things Thanks for that post.
 



You are welcome :)
 
At 270 and below x14 I don't have any stability issues as long as the voltage is decent. I am 24 hours prime stable at 270x14 @ 1.55v and at 265x14 @ 1.5v. Although I have had these voltage fluctuations the whole time, just never thought much of it until I saw it all on a graph on everest earlier today, so I don't know if you can call those stable or not.
 



Hmm.. you got me on this one. If the motherboard is not faulty then perhaps someone with more experience on AMD systems could help out here. But if all is stable and you are happy then i would not worry about it. But out of curiosity i would also like to know if this is normal on an AMD.
 
According to some other forums I posted in, it sounds like this is pretty common with ASUS boards. I just put some memory heat sinks on my VRM chips as they believe my VRM's may be overheating and causing the issue.
 



Well something new to learn today. One way to tell about the VRMs... touch and feel. If they are getting hot then the heatsinks should help eliminate the problem. Tell me how you go after you put the heatsinks on. Does the voltage still fluctuate that much?
 
They were hot enough to burn skin. I put 4 little VGA memory sinks on and it seemed o cause the fluctuations to be less frequent. I am going to get some more sinks and maybe a small chipset fan or something to put over them.
 
That's not the best answer... That's not what is wrong. This is a lot more than a little vdroop. I'm dropping almost half of a volt. In fact, we still don't really know what it wrong yet. I am waiting on some heatsinks and fans that are in the mail to see if I can confirm that over heating VRM's is the real problem.
 
I got them installed and while they weren't quite enough. I was able to run P95 for about 15 minutes before I saw a drop and the drops were much more spaced out until about the 20 minutes mark, then they looked like the did before. So I would say it is pretty obvious that heat is the issue, but will I ever be able to completely eliminate the drops. I hope so.

Here's some photos:







The 5 sinks are the far right feel fine even when the drops were bad, but the middle ones were still really hot and the left ones would probably give you a nice burn. Hopefully my fans arrive tomorrow and fix that.
 
Hmm, are they all on the same MOSFETs? If so, you say that one row of MOSFETs never gets hot while the other 2 are very hot. That could indicate a problem with the design of the motherboard (and that is why other users experience the same). Instead of the voltage been spread over all the MOSFETs equally, one row does not get loaded properly, leaving the other 2 (hot ones) take over its weight, and therefore the end result. That would explain the whole situation. If that is indeed the case, there is nothing you can do to fix the load, only leverage the problem with the heatsinks and lots of air. Get those fans in there!
 
As you can see in the photo, the brown cubes, in the red rectangle, are the right row. They stay the coolest. The smaller black chips in the left blue rectangle, get the hottest by far and the bigger black chips in the middle green square are somewhere in between the two temps.

 
Oh, ok, that explains. Don't worry about the brown cubes on the right (they are called chokes by the way). They are not meant to be hot anyway (so my theory goes out of the window.). But yes the other two rows (MOSFET drivers far left) and MOSFETs in the middle (transistors) are going to get quite hot. It also looks like your mobo is using a 5 phase solution which is not ideal for overclocked systems. A common number for good quality motherboards that support overclocking is 8. The reason behind it is to keep the CPU constantly fed with enough power while relieving the rest of the power components. Obviously the less phases you have on your motherboard the more stress each one will get. The less rest they get, the hotter they get. The hotter, the less efficient. Less efficient=voltage drops + hotter, so it is kind of a vicious cycle.

A "phase" on your mobo=1xbrown box(choke) + 2xMOSFETS (middle row) + 2x drivers (leftmost row)

You will notice there is a phase for your PCIe+PCI cards and another one for the RAM. They should all be located near the corresponding components.