The purpose of vdroop?

[Stigma]

Heya,

Just wanted to drop the question if anyone knows what the purpose behoind vdroop is. From what I understand, vroop is not just a design flaw. It is actually supposed to happen according to Intel spec papers.

So if it is like this by design, then what purpose does it have? Features are usually made for a purpose.

-Stigma

[HaxR3]

Good point, what is the point of reducing voltage when the cpu needs it most?

[NickS]

Yeah. vDroop is bad for overclockers. If the CPU's voltage drops below a stable point, especially under load, it's gonna crash. At stock it's fine, it probably is there to lower load temps.

[wittekakker]

With my current motherboard (P5B) we have the ADP3198 chip for regulating the cpu Vcore. If we take a look at the datasheet, on page 30 we see Vdroop mentioned. Maybe the chip designers recommend a certain feedback configuration and the mainbord designers just follow those recommendations without knowing that there is a better solution for this. Or maybe there is another reason, I'm no professional

[[XC] riptide]

Vdroop is there to lower the voltage to reduce power input into the chip to 'acceptable levels' under load as per Intels design specsheets. So when a MB manufacturer gets the specs for a new chip they are obliged to follow intels instructions.
I'm sure it also serves the pupose of prolonging the life of power MOSFETS on MB's too. But I think Stigma you may be asking the question... 'If the chips design voltage is within a certain voltage-current configuration, why don;t they just have no vdroop and just that lower V to start off with?' that i don't know!

[Falkentyne]

That's a very good question.

I posted the answer to the OPs question almost two years ago yet my post went almost completely unnoticed. I believe I also linked to a PDF on Intel's webpage regarding voltage droop. It may still be searchable, depending on how far back the posts are archived.

The PDF might also explain why the original voltage isn't simply lowered without droop, but I don't remember anymore.

[roger_h99.nor]

Darn, I don't get it.
Intel supposingly instruct the mobo manufacturer to implement i certain droop in the vcore under load?

Always belived this was manufactory shortcut to save $$..

Regards,
r.-

[Holst]

I allways thought that vdroop was just the permitted droop under load and that ideally there should be no droop.

But thats obviously not how many boards are designed.

So im not sure either....

[accord99]

Intel's data sheets for all of their processors specifically mention the voltage drop with load. It's on pages 20-21 on the data sheet for the Core 2 Duos for example:

http://download.intel.com/design/pro...s/31327802.pdf

[Vapor]

From what I know on the subject, Intel designed their processors to still fully and reliably function with droop. This was to allow motherboard makers to use less expensive parts and less time doing R&D developing a circuit that features no droop (supposedly fairly difficult if you want it to last?).

As for why Intel requires droop to be implemented in a board? Got me....I do know the XBX does feature a reduced-droop setting in BIOS, so Intel does know it's not ideal for OCing.....

[mtl_hed]

Compensates for ESR as far as i know, makes boards cheaper. Not all boards have eleven SEPC's on their Vcore output

Tom

QFT



Also, a rhetorical question:

Which is harder to compensate for, vdroop or vcore ripple?

[sierra_bound]

It is a cost-cutting technology. As one VRM maker puts it, the droop "helps maintain output voltage regulation through load transients with fewer (or less costly) output capacitors."

[crotale]

Along with many of the points already mentioned, it's also a matter of a sizing question of the power regulation.
Building a vreg circuit with no droop when going from less than 10Amps at idle to instantly more than 100Amps at load requires a lot of components and more expensive parts.
This is particulary true for video cards.

[uOpt]

Really, the only question is why can't they just lower the Vcore under non-load, too?

[wwwww]

Why don't they just lower the default voltage?

[wittekakker]

Why don't they just lower the default voltage?

Safety margins I guess

[uOpt]

Safety margins I guess

But for what?

From what we overclockers know, the busy CPU needs the higher volts for stability and the idle CPU doesn't.

Why does Intel think that they cannot lower the volts on the idle CPU to the same level that they lower the Volts to when budy?

It's the exact opposite of what we know.

[wittekakker]

I was not responding on that. The message I replied to was about default voltage and why Intel puts it on X volt, or atleast that's what I think he ment.

[freecableguy]

A regulator's output voltage will decrease proportional to load (current). This is a great concept will dealing with multiple power supplies/generators as it allows for load sharing (beyong the scope of this topic). This is, however, NOT ideal when dealing with a single power supply and a single load (CPU). The job of a voltage regulator is to keep voltage as close to constant as possible over ALL ranges of current output. If CPU load where to increase drawing more amperage and the supply voltage were to decrease the situation would be one in which the power level (P=VI) would remain close to unchanged. Others have already hit the truth on the head though....in the end it comes down to two things, PCB real estate and $$$.

-FCG

[uOpt]

Exactly! Dropping the voltage decreases the current requirement on the VRM, Intel cant gaurantee all board designs...

It doesn't explain why you have to start out with higher voltage, though.

[roger_h99.nor]

Along with many of the points already mentioned, it's also a matter of a sizing question of the power regulation.
Building a vreg circuit with no droop when going from less than 10Amps at idle to instantly more than 100Amps at load requires a lot of components and more expensive parts.
This is particulary true for video cards.

First, sorry for kicking this thread to live again..but this issue has been bugging me for some weeks now

Just did a droopmod on a P5WDH, it involved lowering the resistance on a single resistor from 50K to about 20K. Why could'nt, ASUS in this case, just use at 20K resistor in the first place?

This suggests to me that this is not a cost-issue at all..

btw, droopmod did help my stable OC.

[Lome]

think about what happens when processor exits from high load
the exact opposite of droop

[[XC] riptide]

First, sorry for kicking this thread to live again..but this issue has been bugging me for some weeks now

Just did a droopmod on a P5WDH, it involved lowering the resistance on a single resistor from 50K to about 20K. Why could'nt, ASUS in this case, just use at 20K resistor in the first place?

This suggests to me that this is not a cost-issue at all..

btw, droopmod did help my stable OC.

Because statistically speaking ASUS would have a few more RMA's to deal with. Burned power regs anyone.

hey, how much was the stable OC affected by?

[RyderOCZ]

Should be plenty of explanation right here: http://www.thetechrepository.com/showthread.php?t=126

If you haven't already read it

[roger_h99.nor]

Because statistically speaking ASUS would have a few more RMA's to deal with. Burned power regs anyone.

hey, how much was the stable OC affected by?

This is my conclusion also.

Orthosstable oc went from ~ 3,8 to ~ 4,0. A litle disapointing compared to the gain achived on my late S478/Prescott system.

Thx for link RyderOCZ . haven't seen that one before! Nice reading.