An Introduction to Accurate Voltage Measurements - BIOS
(Page 2 of 4 )
BIOS and Software Readings
Each motherboard has a Super I/O chip that is almost always a Winbond IC, which uses a mathematical equation to calculate the temperatures and voltages through various positions. That Super I/O chip is linked with multiple sensors (both thermal and voltage). At first, those sensors monitor accordingly, doing their job quite well, but then they report their results on a continuous basis to the aforementioned I/O chip.
(128-pin PQFP Winbond W83697HF.)
This Super I/O chip gathers the feedback and then uses a mathematical equation to “convert” the results into a form that can be reported in more understandable logic. Ultimately, these results are taken over by the BIOS and after going through a binary to decimal conversion you should be able to see the reported feedback.
Of course, there are a few competent applications like MBM and Speedfan that are able to gather the results via the Super I/O chip and interpret the feedback themselves, thereby skipping BIOS. These utilities are useful, but we shouldn’t rely on them one hundred percent.
Let me explain. In short, the voltages (as well as temperatures) that are being reported (either via BIOS or software) can be inaccurate. Some go too far and actually are notorious for being far off. Unfortunately, this is true in many cases, but we cannot conclude that all of them are this bad.
One of the main causes for problems is variety. There are dozens of motherboard manufacturers, hundreds of motherboard models, and more than a few Super I/O chips (even though most are Winbond, Winbond makes tens of ICs). Due to this, it is really hard to calibrate each and every motherboard, BIOS, and software application. It makes sense, right? One of the slightest notches of calibration-error is enough to fool you.
What’s even more pathetic is the fact that almost everybody takes BIOS readings for granted. They think BIOS is, or at least should be, absolutely accurate. And we shouldn’t blame them. That’s how it should be. But that’s not the way it is.
Therefore, the most common sources of misreporting are because of incorrect calibration and/or inaccurate formulas that are used to report the readings. Your software’s auto-detection might fail and thus use a wrong formula. Second, those internal sensors can go wrong, too. Unfortunately the problem is when they are still in working condition and yet less sensitive/responsive; thus, reporting hard-approximate values which are close enough to seem normal (fooling the user), but not reliable at all.
When the average Joe notices 0V VCC while his computer is still up and running, he intuitively assumes that something’s wrong. In this case, it’s pretty straightforward. But the situation is much harder when it reports 1.30v instead of 1.40v, or 11.75v instead of the real 11.30v. It’s all about precision and the frequency with which they’re updated.
Next: Everlasting Vdroop Dilemma >>
More Hardware Guides Articles
More By Barzan "Tony" Antal
/ 12 
