Vcore drop

[Grinja]

Hi,
I was CPU Z (I know software no good for measuring but just for) shows a Vcore of 1.392 (1.425 in BIOS). When I run Orthos CPU Z shows the voltage dropping to 1.368.

Should I be worried?

 

[JAG87]

nope

 

[Zap]

Normal? Yes. Worried? No.

It's called Vdroop, and some boards have it worse than others. It can affect overclocking.

 

[Modelworks]

Never trust the software.
I'm betting if you put a meter on the fets it would read normal.

 

[Mark R]

Yes. This is normal.

The voltage regulators for high-end CPUs are designed so that the voltage drops slightly under load, this allows the regulator to supply much cleaner power. A stricter regulator which doesn't droop under loader must overcompensate everytime the CPU changes load, causing much larger spikes and sags; the catch is that these spikes and sags are very short lived, so won't be caught by multimeters on on-board software monitoring, but they are just as damaging to the CPU).

 

[SparkyJJO]

Don't trust the software

CPU-Z reported my voltage fluctuating and whatnot, DMM proved that it wasn't at all. BIOS/other software reports my voltage at 1.325V and really it is about 1.38V.

Software voltage readings are not accurate.

 

[Super Nade]

Yeah..software is not too accurate. However, most boards show some droop when overclocked. The magnitude depends on how the Droop control is implemented in the Buck Regulator-PWM feedback loop. There should be a pin labelled "Droop" on your Buck-regulator datasheet. You might have to alter the RSense value if you want to minimize droop. The original setting is probably calibrated using a particular load-line...

 

[jmmtn4aj]

Originally posted by: JAG87
nope

Wow, your board has no vdroop?

 

[Modelworks]

I was really curious about this so I set up a test.
I just got a gigabyte p35-D3SL and started overlocking an E4500 on it.
To measure the voltages on vcore I used an oscilloscope connected to the fet output on the board.
I found something interesting.
The voltage in the bios , set to 1.38, was only 1.34 at the actual fet
When I ran the processor under load, both cores.
The voltage fluctuated between 1.3375 and 1.3425 .
The thing I found the most interesting is that after looking at the scope capture for an hour, it formed a perfect square wave output, with the 1.3425 being the peaks.

Not sure of why there is such a large diference in the bios setting of 1.38 and the actual output at 1.34. All other voltages, 12, 5, 3.3 stay well within spec.