measuring Vcore

[BigAlNZ]

Hi Guys,

I have been playing around with measuring Vcore on some old motherboards.

Quite simply I have been turning the mobo on and putting negative on a mounting screw ground and positive on one of the caps along the side of the cpu.

Is this correct?

Now its all very well being able to measure Vcore but I need to know what it should be? I assume for a given socket Vcore is a set value, or does Vcore change depending on the CPU?

Ie Dual Core and Quad core can both fit a 775 socket, but can they have different Vcores? (ignoring overclockers here for a mo)

I have had a look at a few intel data sheets which seem to specify Vcore, and sometimes a voltage range for Vcore.

When a mobo has bad caps, can it be shown on a DMM with a Vcore that is way outside values?

Is Vcore and Vcc for a CPU the same thing?

Cheers

-Al

[Per Hansson]

Re: measuring Vcore

Vcore should be measured on the Coils next to the CPU, these are the closest things you can generally reach from the topside of the board...

The most accurate is to measure directly on the underside of the CPU socket, wikipedia for example has pinouts, and pinouts.ru too

But that's really overkill...
In the olden days CPU's where just binned at the factory for the speed which they would run reliably at, all CPU's sharing the same vcore

So for example you had Barton based Athlon XP CPU's and they ran at 1.65v
And from 1666Mhz > 2100Mhz

But nowdays it's more complex, CPU's are binned both for speed in Mhz but also at how low a voltage they can run
So nowdays the CPU's differ in vcore, they have a range they accept but every CPU will be different

My current Q9450 for example has a VID of 1.25v, but it could be anything between 0.85v - 1.3625v
The lower the reported vcore voltage is the cooler the CPU will run and generally the more it will overclock...
http://www.cpu-world.com/CPUs/Core_2...569Q9450).html

Core Temp can be used to check the VID of a CPU; http://www.alcpu.com/CoreTemp/

As for the actual Vcore voltage you should see that is set by the BIOS
Or on older boards by switches on the board itself

Vcore flucuasions can be measured by a DMM, but it is normal to see fluctuations, when the CPU draws 100% power vcore will sag, this is normal
You will not know if you have bad caps or not, for that you need a ESR meter

[shadow]

Re: measuring Vcore

I can not talk about all processors here. However of the processors that I have had experience with. It is the CPU that governs the VCore voltage. Different CPU's fitting the same socket run on slightly different voltages.

In my experience, even two 'identical' CPU's can have different Vcore voltages. When I say identical, I mean the same socket and processor speed. The steeping of the CPU's were different. The two CPU's were manufactured at different times and hence they are not exactly identical CPU's but extremely similar CPU's.

I am not sure why this is the case, I just know that it is.

[gdement]

Re: measuring Vcore

Intel started not specifying an exact Vcore during the late P4 days. So-called "multi-VID" chips started coming out, meaning you don't know what VID is programmed on a given chip, even if you know the S-spec code.


You can test without a CPU just to see if it's a CPU killer. But on a good board the reading will either be 0v or maybe a minimum value until you actually plug a CPU in to it.
At least on older boards, the CPU pulls various pins either to ground or +(5v?), and the VRM uses those signals to set it's voltage. Without a processor, the board pulls all of them one way (ground probably) which corresponds with the minimum Vcore setting.

When using caps (or anything else) to measure Vcore, ideally you should check for continuity with the proper pins on the CPU socket so you know it's actually on Vcore. Sometimes unexpected stuff can be near. I remember an ABit board which actually had AGP caps *above* the Vcore caps. Nuts, but I checked it several times and I'm sure that's what they were.

[gg1978]

Re: measuring Vcore

VRM signals float high, and the CPU pulls some of them down to ground to signal what voltage it wants.. So, if all pins are high ie "1111111", that means there is no CPU in socket, and VRM will be turned off. VRM 8.1 started out with a 4 bit VRM id, while VRM 11, the latest one, basically has a 7 bit VID..

I forget exactly which VRM spec allows Vcore to change dynamically with CPU loading/clockspeed.. Probably past VRM 9.0 i'm guessing.. As Per Hannsen says, the output inductors are a good place to probe, as you're less likely to short something out. The actual Vcore will be very slightly less than what you measure at the inductors, even though the trace that goes to the CPU is massive, but then again, the current and the delta current are very high as well..

Try putting your video card in the other PCIe slot.. I've seen some nvidia boards with a bum pcie slot, but putting the card in the other one would allow the board to POST. Or try a PCI video card..

[BigAlNZ]

Re: measuring Vcore

Vcore should be measured on the Coils next to the CPU, these are the closest things you can generally reach from the topside of the board...

The most accurate is to measure directly on the underside of the CPU socket, wikipedia for example has pinouts, and pinouts.ru too

But that's really overkill...
In the olden days CPU's where just binned at the factory for the speed which they would run reliably at, all CPU's sharing the same vcore

So for example you had Barton based Athlon XP CPU's and they ran at 1.65v
And from 1666Mhz > 2100Mhz

But nowdays it's more complex, CPU's are binned both for speed in Mhz but also at how low a voltage they can run
So nowdays the CPU's differ in vcore, they have a range they accept but every CPU will be different

My current Q9450 for example has a VID of 1.25v, but it could be anything between 0.85v - 1.3625v
The lower the reported vcore voltage is the cooler the CPU will run and generally the more it will overclock...
http://www.cpu-world.com/CPUs/Core_2...569Q9450).html

Core Temp can be used to check the VID of a CPU; http://www.alcpu.com/CoreTemp/

As for the actual Vcore voltage you should see that is set by the BIOS
Or on older boards by switches on the board itself

Vcore flucuasions can be measured by a DMM, but it is normal to see fluctuations, when the CPU draws 100% power vcore will sag, this is normal
You will not know if you have bad caps or not, for that you need a ESR meter

Thanks. Thats a pretty good reply.

What about measuring Vdimm - is that more straight forward?

[Per Hansson]

Re: measuring Vcore

Yea, I generally measure Vdimm on the memory socket itself, it is possible to get a probe touching the lead even when you have memory in the socket of most sockets I've seen

It can also be measured on the mosfet that delivers current to the memory of course...
As for the voltage this too is set by the BIOS, but if you run at "Auto" the voltage should be whatever is programmed into the SPD, or if there is no voltage info in the SPD just what the JEDEC spec says for that type of memory

Though variances occur here too due to "performance memory" where manufacturers run the memory at voltages outside the JEDEC spec to attain higher speeds...

[sam67]

Re: measuring Vcore

Is there any where that shows you a bit more in depth i.e photo's of examples as to where to check with dvm as i would like to learn some of these measurements ..Any help appreciated.

Cheers

[Headcase]

Re: measuring Vcore

How is the Vcore on a 'dead' motherboard tested? I have two machines (possibly) in need of recapping, and one of them doesn't power on. I don't have spare CPU's, they're both Intel but different socket types. I tried testing the dead machine at the coils, and got 0.3V with the chip in, and nothing without the chip installed. How do I tell if this motherboard will kill a CPU, or if it has?

Thanks,
Headcase

[ChaosLegionnaire]

Re: measuring Vcore

due to variances in things like dust particle count and level of impurities present in the silicon wafer during manufacturing, each and every cpu manufactured is "unique".

this "uniqueness" most often manifests itself during overclocking. when no two cpus from different batches can overclock to the exact same speed with the exact same vcore unless u're really lucky.

due to that, each cpu has different voltage requirements to stay stable at their specified binned speed. also, from what i heard about how intel bins their cpus, they usually test whats the max stable speed a cpu will be at a given voltage then mark it down 2 speed bins and then sell it. thus if u have a 2.4ghz cpu, its actually been tested to be stable at 2 speed grades higher at 2.8 ghz.

hope this sheds some light on the question raised about vcore differences among cpus of the same speed.

[Headcase]

Re: measuring Vcore

I think I should restate, maybe I rambled too much and didn't describe my problem succinctly, sorry about that.

The revised version: ()

I have a machine that won't power on, and it looks like the board needs recapping. I'd like to test the vcore first to see if it's going to kill the CPU or not. Is this possible to do, and if so, how?

[Per Hansson]

Re: measuring Vcore

The voltage is set by the CPU.
But there are CPU test cards you can buy on eBay, they have diodes lighting up telling you if there is a faut, short etc...

On the other hand there are also CPU's costing just a few dollars you can buy on eBay