Its a drop of CPU voltage when the CPU is taking a lot of power. Happens when the voltage regulator on the motherboard is not well designed. Usually for socket 775 processors a 4 phase or better power regulator is recommended.

For boards suffering from vdroop, sometimes overclockers voltmod them for higher voltage to compensate for the voltage drop. Setting a higher voltage in bios is also possible, but often results are unsatisfactory. But increasing the voltage has its problems as well, as when the processor is idle, the voltage will be quite high (no voltage drop), which may damage the processor.

---

Added on June 24, 2007, 4:17 pmAnd yes, it also affects other boards, not just Asus. Asus is often mentioned because they used to often use really small mosfets in the voltage regulator which were prone to overheating (which causes mosfets to conduct less current)

This post has been edited by lohwenli: Jun 24 2007, 04:17 PM

Not a chance. Even with a PSU with adjustable rails there will still be vdroop if the mobo voltage regulator is not well designed.


Cool article. Downside of pencil mods and other hard mods is they don't solve the vdroop problem entirely or introduce problems of their own. But anyway only very hardcore overclockers will care...

It IS a problem, and has to be taken into account during design, as its pretty much unavoidable. Although its important to minimise it, its not cost effective to use a very complex design, so manufacturers make do with simpler, cheaper designs that work and compensate by adjusting the voltage to a value where vdroop is still acceptable. The problem starts during overclocking, as higher power consumption in overclocking will mean more current, which will cause the vdroop to be more severe. There will be minimal vdroop when the processor is at idle, but severe vdroop when the processor is fully loaded, which is bad, as overclockers may increase voltage to high levels to compensate for vdroop, and the absence of vdroop in idle will mean the voltage may be dangerously high (processor damage possible).

You better double check that, the formula should be V=IR, where V is the resulting voltage drop, I is the current taken by the processor, and R is the equivalent series resistance in the voltage regulator circuit from the sensing point to the processor.

P=VI is to calculate power usage, specifically conversion from electricity to heat (which applies in almost all electrical devices). Although your description does explain a downside of vdroop, it does not explain accurately why it happens. Yes, a poorly designed board suffering severe vdroop will actually make the problem of overheating MOSFETs worse and worse, as the voltage drop (using V=IR) across overheated MOSFETs will result in a power loss and they will overheat more. And the more the MOSFETs are overheated the worse the voltage drop-a vicious cycle. Contrary to your explaination, current (I) can only be increase up to a certain point until the MOSFETs are working at full capacity-which would be limited by overheating (a safety feature of MOSFETs).

This post has been edited by lohwenli: Jun 26 2007, 10:15 AM

There's no sure way of finding out unless you are very familiar with the voltage regulator circuit of that particular motherboard. At best, you can guess the peak current of the mosfets during the max duty cycle of the voltage regulator by estimating from the power usage of the processor. If the number you come up with is higher than the rated current for the mosfets, then the mosfets are overloaded. However, finding out the exact current used by the processor is a major problem, even the best estimates can be off by quite a bit unless you have proper equipment and are prepared to hard-mod the motherboard.


Sucess rate of 1 & 2 highly depends on how well the voltage regulator circuit was designed in the first place. If the manufacturers decided to cut costs by using fewer or lower quality caps, then you'll improve on it. However, if the circuit was already well-designed to start with, you may just screw everything up, with potential problems like PSU damage (due to surge current at startup), voltage fluctuations (due to the different circuit harmonics),

3 (AKA hard-modding) on the other hand, has been done by overclockers with some success, but like you said, need to be very familiar with the circuit. I did it once before, seriously its not for the faint of heart.