Voltage vs. Temperature

M

MastaLock

Limp Gawd
Joined
Feb 6, 2007
Messages
139
I've been overclocking the rig in my signature and I have a question. With the Tuniq Tower I am running pretty low temps (51C max I've seen) and fairly high voltage (1.4875ish).

My question is, is there a limit the to voltage I am pump into my chip (1.6v from my mobo) as long as I keep the temps below 65C? I know generally the lower voltage the better, but if I was able to run it at 1.6, would it necessarily allow me to overclock more?

Current Voltages
Cpu:1.4875 (Only shows up as 1.44 in CPU-Z though...)
NB: 1.56
Memory: 2.178 @ 4-4-4-12 2T

If there is anything I am overlooking to boost it up some more feel free to comment on it. Thanks for the help.
 
Yes, I understand that more volts = less CPU life, but will it fail before 2 years?


Well, I hope not (lol). Thats all it really needs to last me.

So anyways, back to the original question, is there a limit to the voltage I can apply and the CPU won't overclock anymore?
 
Going from 1.50 volts to 1.60 volts creates a lot of extra heat but doesn't really give you a lot of extra MHz.

For 24/7 I don't see much point in going much over 1.50 volts but even at 1.60 volts, you probably won't hurt it. If you do hurt it, that's what the warranty is for. :D

Just remember to keep an eye on the temps.

Don't do nothing too crazy like this idiot!
http://www.hardforum.com/showpost.php?p=1030718057&postcount=56
 
So basically, topping out at 3.2Ghz @ 1.5v means that I probably won't be hitting 3.5Ghz right?
 
Why not just run it at 1.6v and find out..... 1.6v @ 65c running 24/7 will probably lose 50% of your cpu life...... youd need a new cpu 5 or 6 years from now lol..... if not then RMA.....
 
Vcc max is 1.55 Volts Table 4 page 18

http://download.intel.com/design/processor/datashts/31327803.pdf

Its your stuff, do what you want. I dont go over 1.45 but thats just me. If I had a processor tree in the back yard I would go higher.

Note also you are talking about really ramping up the cpu voltage which is mainly provided by the power MosFETS of cpu voltage requlator circuitry which normally are arranged immedately around the cpu area. With that Tuniq you have eliminated the "blowby" from the stock down blowing heatsink that is used to help cool the MosFETS decreasing their reliability and more to the point, increasing the chances of unstable operation/voltage due to high operating temps and cranking up the CPU voltage only will make it more likely to have issues.

Looking at a picture of your board I see no heatsinks on the FETs .

Consider doing this, see my post in this thread, basically put ramsinks on the FETs.

http://www.hardforum.com/showthread.php?t=1164676
 
Nice, thanks for the technical response. The FETs are still getting some airflow from my 25cm side fan.

So I have another question, is the heat in the FETs the main source of my 'V-droop' that it looks like I'm experiencing?
 
MastaLock said:
Nice, thanks for the technical response. The FETs are still getting some airflow from my 25cm side fan.

So I have another question, is the heat in the FETs the main source of my 'V-droop' that it looks like I'm experiencing?
Click to expand...

Excellent ( 25cm fan) :D

nope, its a function of the quality of the entire Voltage Regulator system, as you draw more current from the VR circuit due to overclocking the voltage drop across the components that are feeding the power to the cpu increases a tiny bit and they "steal" a little of the voltage. E=I*R so as I goes up, so does E. So lets make up some silly numbers to clarify, lets say that when a FET is delivering 2amps and its internal resistance is .5 ohms the voltage drop (loss, turned into heat) in/across the FET would be 1 volt ( 2 x.5) now if you pour on the OC and the cpu demands 4 amps drop would be 2 volts (4 x .5) and the VRD has to try to compensate for double the voltage loss which it cant do exactly. So voltage loss is occuring in the components and some attempt is made to adjust for it but its not perfect. Of course these numbers are way silly and oversimplified but you get the idea. (its really wild due to the high frequency switching, calculus to determine real loss, pulse width changing, area under the curve for power, so the compensation required is not linear and difficult to adjust for with all the stuff going on).

Now you can implement all kinds of feedback to monitor the actual voltage delivered and "tweak" up the voltage to overcome the losses but its expensive to design and implement and not perfect. Everything is a tradeoff. Its also why some of us get excited over stuff like 12 phase CPU voltage regulator circuits (hmm definately need more neked women around here)

Oh lots of people say PWM when they really mean the FET's. I am on a mini-quest to get the termonology fixed.

VRD or VRM or Cpu voltage regualtor - entire circuit feeding cpu power. Main parts are PWM controller, FET switchers, filter circuit.
PWM - controller chip, the brains of the VRD/VRM, does not deliver power.
MosFETS or FETs- workhorse - do the swithing of power, controlled by the PWM, to feed the correct voltage to the cpu.
Inductors and Caps - filter circuits after the FETS to filter our noise and ripple and smooth out the power to the cpu.

When people do Vdroop mods they are messing with the senseing/feedback circuit that controls what the PWM chip thinks is being delived to the cpu to "bump" it up some. I am not a fan, done properly it can overcome some factory toleance stackups etc. and do some good, done badly it can scew up stuff most people have no idea is going on or test for.

Best idea is to get a board that will allow you to bump the the cpu voltage in the bios in a lot of small increments so you can just "go to the next setting" to compensate and also get a board with a good VRD to start with. Some droop is unavoidable unless you put a laboratory quality supply on the board.
 
You must log in or register to reply here.
Back
Top