VRM, Phase Control settings, and stress on parts

[BonzaiDuck]

My ASUS board is about two generations old, but it definitely falls into the new wave of mobo products using solid-state components and more features in BIOS.

There are features for "VRM Duty Cycle" and "VRM Phase Control." Duty cycle offers a tradeoff between "thermal balance" and "current balance." Phase Control, like LLC, has levels topping out with "Extreme."

Both these features urge user not to disable thermal monitoring.

With the "extreme" choice for these items, what hardware components are stressed and how? . . . . If anybody knows? I'm guessing that they add heat to motherboard components, but the mobo maker had chosen to build them into the board and the BIOS.

Then again, maybe not. I just want more perspective on what it is I'm doing to get stability for higher clocks and minimum VCORE.

Any thoughts?

Moved from CPUs to Motherboards
-ViRGE

 

[Deders]

VRM are Voltage Regulator Modules, they are most likely under heatsinks near your CPU.

Like the name suggests they regulate the voltage to the CPU to give it a stable current.

A more refined signal will allow higher overclocks, but at the cost of more heat as the VRM's will be working harder.

If you have good airflow in your case it's probably not something you need to worry about too much, what are your current CPU temps when stressed?

 

[BonzaiDuck]

VRM are Voltage Regulator Modules, they are most likely under heatsinks near your CPU.

Like the name suggests they regulate the voltage to the CPU to give it a stable current.

A more refined signal will allow higher overclocks, but at the cost of more heat as the VRM's will be working harder.

If you have good airflow in your case it's probably not something you need to worry about too much, what are your current CPU temps when stressed?

Well, I feel pretty good about that. With the Noctua NH-D14 and the case fitted with 2x 200mm fans, a 140mm CPU fan and 120x38mm exhaust, I decided to tune the system for a multiplier increase from 46 to 47 -- so running at 4.7 Ghz.

With Prime95 testing, my calculated average of the four cores of the i7-2600K reaches about 72C at a room-ambient between 75 and 79F. Under the same conditions, LinX shows around 78 to 79C for the average. I run LinX for affinity to "cores" 0,2,4,6 with hyperthreading enabled.

My average would be lower, but I have one core sensor which I believe is out of whack. I know there is an error range of +/- 5C on these things. The other three cores are pretty close together in temperatures reported.

The ASUS software apparently uses some estimation algorithm to convert the core values to something like "TCASE." So it shows at most 72C for the LinX tests.

I haven't done any stunning "ducting" mods to cool the motherboard further, but I have an NZXT 200mm case fan with 166CFM rating that blows air all over the motherboard.

UPDATE: Per another thread in the cooling forum, I replaced the Noctua fans on my NH-D14 with a single Akasa Viper-140-R. What I thought had been a core average of around 78C is now a measured average of 74C under LinX testing.

 

[john3850]

Except for my first petium I have always added one or two small fans that were tie backed to the 4 wb mounting studs.
You may gain some extra cooling from hsf but WBs seem to trap the Vrm heat even more.
The extra Vrm and cpu heat is what causes noise and lowers your oc.
To solve this op raise the vcore even more which leads to more heat.
Next the 212 hsf is the holy Grail for cooling.

 

[BonzaiDuck]

Except for my first petium I have always added one or two small fans that were tie backed to the 4 wb mounting studs.
You may gain some extra cooling from hsf but WBs seem to trap the Vrm heat even more.
The extra Vrm and cpu heat is what causes noise and lowers your oc.
To solve this op raise the vcore even more which leads to more heat.
Next the 212 hsf is the holy Grail for cooling.

We've touched base before on the Sandys -- back in '11. I'm supposed to be savvy about techno-acronyms. But since I can make the excuse that I'm old, I'll just ask the question:

What's a WB?

 

[john3850]

Sorry wb = water block

 

[BonzaiDuck]

Sorry wb = water block

I should've figured that out, but I wouldn't be corresponding with a lot of WC gurus to collect the details of my system or share about how to tweak it. That's because I've avoided watercooling for a long time, despite an interest in Rube-Goldberg water-exotics.

And I think that's changing. . . I think I'm planning another system . . .

But you can do that stuff with the fans zip-tied to the waterblock mounts. You could also take a lot of time to cut a motherboard-duct-plate, maybe integrate some Mag-Lev 40mm fans to it, and force air through narrow spaces over components on the way to a quick exhaust.

There may not be much of an increase in noise from the fans, and of course the other approach would deaden that noise anyway.

 

[john3850]

The extra Vrm and cpu heat is what causes {noise} and lowers your oc.
What I meant by noise is electrical noise .

Electrical noise: Causes, effects, solutions Paperback
Electrical noise, in its various forms, can adversely affect any product using electronic circuitry. Its potential to cause damage or dysfunction is increasing today as electronic circuits become more and more complex. Today's computers and microprocessor-based systems operate at higher speeds and provide more features with reduced size and weight through the use of complex solid state components, both analog and digital. These are inherently fragile and susceptible to damage and/or malfunction from electrical noise.

I also believe that your best chances at a high stable oc example 49k-53k starts to diminish from wear and stress to mb way before the cpu starts to degrade.

 

[BonzaiDuck]

The extra Vrm and cpu heat is what causes {noise} and lowers your oc.
What I meant by noise is electrical noise .

Electrical noise: Causes, effects, solutions Paperback
Electrical noise, in its various forms, can adversely affect any product using electronic circuitry. Its potential to cause damage or dysfunction is increasing today as electronic circuits become more and more complex. Today's computers and microprocessor-based systems operate at higher speeds and provide more features with reduced size and weight through the use of complex solid state components, both analog and digital. These are inherently fragile and susceptible to damage and/or malfunction from electrical noise.

I also believe that your best chances at a high stable oc example 49k-53k starts to diminish from wear and stress to mb way before the cpu starts to degrade.

Yeah -- I understand the difference between electrical noise and acoustical noise, but the fans you mentioned are for cooling mobo parts, and fans can create acoustical noise.

I agree about "what goes South first." Before they started using more solid-state components, I'd see it happened to my motherboards. I even saw the voltage "drift" problem on an eVGA LGA-775 780i board. I had to add a couple notches to the vcore setting just to keep the reported value where it had to be.

I still have some of the older boards (775) -- a 680i Striker which is now just the WHS-2011 box, the eVGA and a Gigabyte P45 board. I reset them from any overclock settings at least a couple years ago.

I suppose only time will tell on my Z68-V-Pro. The last mystery-dilemma I tried to resolve was posted all over here, but it was fairly straightforward -- running too many RAM modules above spec. It wasn't a problem that was "getting worse," but pretty much staying the same from the moment it began to happen -- which in turn was within a week of the changing the default RAM settings. Everything else seems to function just as it did two years ago, and I replaced the RAM with a two-stick kit spec'd to run at the settings that had caused the trouble.

 

[bryanl]

VRM are Voltage Regulator Modules, they are most likely under heatsinks near your CPU.

Like the name suggests they regulate the voltage to the CPU to give it a stable current.

How does a VRM stabilize the current, except to keep it below maximum?