GA-EX58-UD4P SYS-FAN3 not reporting fan speed

[tdhg566]

The SYS_FAN3 connector on my GA-EX58-UD4P mb provides fan power, but doesn't report speed. In fact, there's no "record" of it on the PC Health BIOS page. Only four fan speeds are listed there, but the mb has 5 connectors. Speedfan sees five fan speed buckets, but one of them, SYS_FAN3, is zero. I can hook up fans to this connector, just can't monitor their health.

Would be nice to collect the speed information

?suggestions

 

[badge]

Maybe the MB is designed that way. Hooking up fans with only neg. and pos. connection are not speed monitoring capable. Fans require a third connector to monitor speed. Fans with Neg/Pos only will not measure speed.

 

[tdhg566]

Thank you for the suggestion. All fan connectors on the mb are 3 pin. This connector is listed in the mb diagram in the manual. The BIOS PC Health screen labels for the fans don't match the names on the mb, but the mb does match the manual.

My thought? (hope I'm wrong): This is a relatively new mb, and I think the BIOS is wrong and is mis-matching the fans with the buckets, and someone just forgot to collect the data on SYS_FAN3.

But that's just my thought.....

 

[bilbat]

Try the beta SpeedFan here:
http://almico.com/sfbetaprogram.php
you'll need to register for access...

 

[tdhg566]

Thanks. Did, with no change. The beta reports the same information (5 fans) that the production version of Speedfan does.

Perhaps I wasn't clear. I think Speedfan is reporting the information correctly. At least it shows 5 fans. IMHO it's the BIOS that's only showing 4 fans on the PC Health screen, and only collecting speed data on 4 fans.

 

[bilbat]

Well, this one sure bit me in the ass! Hunted down an old reference to a tech spec sheet for the ITE chip that keeps track of this stuff to see how many channels it had. Got a NoExMemProt fault, figured it for for yet another 'false positive' (been getting a few of those lately, enough to annoy me) so I rebooted, jumped into the BIOS and disabled protection; went back, and POWEE - must have been an actual buffer overrun - blue screened Win7x64 (haven't seen that before!), corrupted my BIOS (needing a CMOS reset & re-load to correct), corrupted a pair of in-process torrents, wiped out a video transcoding job, and corrupted my RAID1 pair of RE3s, requiring a hot rebuild... Needless to say, that's the end of running 'unprotected' - I'm putting a rubber on the end of my cable modem connection! I'll be back after hunting a little more circumspectly for that spec sheet...

 

[tdhg566]

Wow. It sure did bite you! Certainly appreciate your research, but wouldn't wish for this kind of outcome just because I'm missing a fan speed bucket. Thanks!

 

[bilbat]

Ack - certainly wasn't your fault! And, I learned a number of things: there is a reason for no-execute memory protect; I'm damned glad I use the CMOS parameter <F11 - F12> storage system; and, the Intel RAID recovery stuff is impeccable! Might have been random - was transcoding a batch of old Twilight Zone episodes, one might have leaked my system into another dimension...

OK - finally found the little %$#&! Turns out that the "IT87" designation that shows up in all my monitoring stuff refers to a popular family of LPC (LowPinCount) multi-function monitoring chips: IT8705F, IT8712F, IT8718F (which appears to be the one in most GB MOBOs), etc. I have pulled down several specs, located here:
http://www.ite.com.tw/EN/products_kind.aspx?CategoryID=3&ID=5
Every one I've read so far supports five fan feedback channels; up until the X58s, GB has only chosen to access four from their MOBOs; still haven't found which IT87xx is used in the X58 boards, but at least now I know what I'm looking for!

These sorts of things are fun - like being Sherlock Holmes for a day...

 

[tdhg566]

Fascinating indeed! I love this kind of stuff too, but don't have your experience or resources to do the investigation. At this point in my life, I just want it to work LOL. But I do enjoy reading the accounts of your efforts.

 

[jwdaigle]

Hi guys -

I too am searching for info on how to run speedfan with this otherwise awesome board. When I run speedfan 4.38 (havent tried the beta yet), it appears that the speeds being reported are really wacko. Im kind of new to speedfan, so am cluless on how to map things correctly.

From what I gather, people post a config file that defines correct mappings for each mobo (due to the large number of chips and design decisions made on what to hook up to what).

If this is plain trial and error, if someone will enlighten me what to try and what to expect, I am more than willing to help.

Until then, I will wait in the hopes that one of you has already figured it out

Thanks!

Joe

 

[badge]

With Speedfan click Config and disable the 'wacko' readings. Like the readings with a flame notation.

 

[jwdaigle]

The voltages are way way off also - as in I think I would be seeing smoke if they were true. +12V reports as 1.38, +3.3V reports as 0.19, -12V reports as -18.39 etc.

Im sure I can disable the mismapped readings, but was hoping to use speedfan to read these settings.

 

[bilbat]

My latest guess is that the IT8720F (the LPCIO used in the new x58 MOBOs), which was just added in SpeedFan version 4.38, might still have a few rough edges. The linux guys are struggling a bit with it, too - there appear to be some peculiarities from the older, well supported 8718's. The Intel LPC bus spec (which is old, like '02) is here:
http://www.intel.com/design/chipsets/industry/25128901.pdf
and I'm still trying to track down full spec sheets for the parts themselves. I suggest an e-mail to Alfredo Comparetti, the 'speedfan guy' at:
mailto:alfredo@almico.com
he's a really great guy, and very responsive to questions/suggestions; I've talked with him a few times over nVidia's bald refusal to share any info regarding the supposedly 'open' ESA specification...

 

[jwdaigle]

OK, will do. Just sent an email to Alfredo with whatever diagnostic information I could muster up.

Will post back if anything gets resolved.

Thanks for the pointers -

joe