Accurate temp readings with Asus A7V333?

[CrazySaint]

I'm considering purchasing the Asus A7V333 board, and I was wondering if the temp readings used for the C.O.P. feature could be accessed via software (i.e., motherboard monitor) or maybe some accessory. I know you can buy temp sensors, but I've heard that they aren't very accurate and I figured the readings used for the C.O.P. would be more accurate.

 

[CrazySaint]

bump

 

[mechBgon]

There's some ambiguity about it. On Asus' site, they sort of skirt the issue. Read their blurb and it's clear that the COP function uses the thermal diode. A sudden change in signal from the thermal diode will trigger a shutdown.

But does the BIOS report the temperature from the thermal diode, or does it report it from a separate thermistor elsewhere? THAT is the question. As you see in Asus' article, they say that the COP function is software-independent. Asus makes their own ASIC (application-specific integrated circuit) and could have it monitoring the CPU's internal diode, but still could use a socket thermistor that the BIOS and other programs can see (including Asus Probe). I now own an A7V333-R myself (haven't built it yet), and there is one surface-mounted device inside the socket which is different from its neighbors. I guess the way to determine whether or not the board gets readings from the die, aside from asking Asus themselves, would be to look at the speed of temperature changes when the CPU goes from idle to full power.

Once I get my board running, I'll install Asus Probe and try doing a full swing from idle+software cooling to full load+no software cooling, and watch the response time. If it takes just a few seconds for the temperature to begin climbing, it's probably the on-die diode. If it takes 10 seconds or more, it's probably a socket thermistor.

 

[Mikewarrior2]

MechBGon,

Hope you can figure this out ... Hpoefully the A7v333 isn't like the SOltek DRV5, which has safety functions of the internal diode, but gives temps via socket-thermsitor.

One has to wonder why the hell they did that .


Mike

 

[mechBgon]

I know! You'd think it was taboo to use the AthlonXP internal diode to measure temperatures, the way the motherboard manufacturers avoid it!

I should have results by Friday. The hottest-running application I have is 3D modelling software, which causes socket-thermistor readings to peak about 2C higher than SETI@home does, so that should be a good way to apply a very sudden heat load and watch the response time in Asus Probe. As I said, I can make the swing more dramatic by switching off software CPU cooling (a feature of Asus Probe) right before starting a rendering. I wonder what the shortest reading intervals are... guess I'll know soon.

One of the other things I want to try out is the MyLogo feature. Got my 16-color cheetah logo all ready and waiting. This is a feature that Asus doesn't cover in their manual, but it sounds like it's done using the Asus BIOS-update utility, which also allows customization of the voice messages ("HEY IDIOT! YOU FORGOT TO GIVE ME ANY RAM!" will be one of mine ).

 

[Mikewarrior2]

MechBGon,

Awesome, looking forward to whatever you can find out.



Mike

 

[CrazySaint]

I'll be looking forward to your findings as well. )

 

[mechBgon]

Ok, I dropped my A7V333R into my home system in place of my K7S5A. You can say what you like about Win98's stability, but it certainly is flexible... it blue-screened on the first startup, but it made it through the second one and I was able to plunge into a quick-and-dirty test of the thermal question (after installing 4-in-1 drivers, etc).

Asus Probe started off by claiming that my +12V voltage is 13.8V, which didn't give me a lot of confidence in it. It's not getting that reading from the BIOS, at any rate. Asus Probe showed my Win98 system idling at 58C, and I had specified 1-second update frequency. I switched on AP's software cooling by checking the checkbox, but the temperature sat at 58C.

Next I started SETI just to see what would happen. Over the course of two minutes or so, the CPU temperature rose from 58C to 60C. I stopped SETI and it fell to 57C over about two minutes' time.

Next I fired up trueSpace 4.3 and pulled up a scene with an hourglass. I made five extra copies of the hourglass and spread them around. The CPU has to work very hard to compute the reflections and refractions of glass, so I knew that would keep it busy for a while. I started the render session and watched the temperature. Over the course of about four minutes, it rose and peaked at 61C. I watched for about five minutes and then cut off the rendering and timed the fall in temperature. It hit:

60C after 35 seconds
59C after 1:17
58C after 2:03
57C after 2:43

From the lag in the temperature response, and the very small spread of temperatures, my initial guess is that the system is not getting its reading from the core diode, but rather from a socket thermistor.

I just thought of a method of testing that theory, which is to get the motherboard out of the case, locate the underside of the socket, and hold something cold against it while watching the temperature readings. If they drop quickly in response to a cold object on the underside of the motherboard, that would be evidence in support of socket-thermistor readings. Stay tuned (and wish me luck)!

EDIT: and before I try this stunt, let me say that MyLogo ROCKS! It is very easy to do.

1) get the image you want, and use a quality image editor to reduce its size to 640x480 pixels, then convert it to 16 colors. I recommend making it into a GIF with 4-bit color (which is 16 colors), using PaintShop Pro or something similar (you can download a trial version if you need to). After getting it into 16-color mode to your satisfaction, choose Save As... and save it as a BMP file.

2) download the BIOS from Asus, even if you don't need to update to a new version. Save it on your hard drive.

3) After installing the Asus Update utility from the CD that comes with the motherboard, go to Start > Programs and launch the Asus MyLogo utility. Tell it where to find the BIOS you downloaded to your hard drive, and then the image you've made. It will replace the stock A7V333 logo with your own logo.

4) Now you have a customized BIOS file, and you just need to run Asus Update from Start > Programs and install your custom BIOS.

I have my Cheetah avatar at startup now, very appropriate since I will have a Seagate Cheetah X15-36LP in there (assuming I don't destroy my motherboard in the cause of science first )

 

[mechBgon]

Ok, the evidence is conclusive: both the BIOS and Asus Probe are getting their readings from a socket thermistor on my A7V333R. I have it out of the case (thanks to Enermax and their super-long ATX cables) and I went to the BIOS and then held an ice pack wrapped in a t-shirt against the underside of the motherboard where the CPU socket is. The temperature instantly began to drop... FAST. The BIOS updates the temp and RPM readings about twice per second and it dropped and rebounded extremely quickly in response to the ice pack. Yep, kind of like we wish Asus Probe would do. :|

I then booted into Windows and did the same thing while watching Asus Probe. No doubt about it... this board may read the AthlonXP's internal diode for overheat protection purposes as Asus claims, but the end user is only going to get vague temperature trends from the socket thermistor, unless perhaps Motherboard Monitor will access the CPU diode.

I may not be able to answer any further questions until pretty late tomorrow, because I will be doing some Dr. Frankenstein action on my two systems (home and work) and they'll both be getting fresh installations of Windows. I hope this was a help, although it wasn't the most scientific approach.

 

[Mikewarrior2]

Now I'm really wondering where the hell the diode support is... hello to everyone who yelled at me that it was coming??? it still aint here!!!

The two primary "cpu-overheat protection" motherboards both do use the internal diode, but show socket-thermistor temps? Man, something is either fishy about the internal diode temps, or they are too high and marketing doesn't like it .



Mike

 

[mechBgon]

What's interesting is that I am getting temperatures in the high 50's/low 60's on an AthlonXP 1700+ (at stock speed) with an Alpha PAL8045 using a Panaflo L1A. On a motherboard with a socket thermistor, this would usually yield much lower "apparent" temperatures in the 40's. Did Asus try to calibrate the socket-thermistor readings look high enough that they might pass for CPU-diode readings? By golly, I don't care if the temperature reads 80C, I want to know the TRUTH about my core temperature.

I took my A7V333 to work, where it will spend the next four years or so. We got another one that my supervisor wanted me to build for his nephew (I got to pick the parts). I flashed the BIOS using the Asus Update utility and I think the BIOS was corrupted by the floppy drive that we re-used from his old system. The system refused to POST from there on. Sometimes the onboard voice would say there was a CPU error. After trying everything else I could think of, I booted my A7V333 to a DOS prompt, started the AFLASH.EXE update utility, pulled my BIOS chip out :Q and put the dodgy one in, then flashed. Unfortunately, this still failed to result in a board that would POST. Talk about egg on my face... guess I'll call Asus tomorrow.

 

[CrazySaint]

mechBgon,
Thanks for posting the info! Sucks that they apparantly use thermistors, though...(and that your BIOS is messed up, hope you get it fixed). It really makes no sense that they wouldn't use the thermal diode to report temps, though we can at least hope that they did, in fact, calibrate the thermistor with the thermal diode (or that Motherboard Monitor can access the temp diode readings). One last hope might be the Asus iPanel, which fits into a front 5.25" drive bay and has an LED that gives cpu temp, boot error, and fan RPM readings. I wonder if there's any chance that it might read off the thermal diode, though it seems unlikely...

 

[CrazySaint]

I just read a review of the iPanel, and apparantly its temp readings are just downright buggy, with temps seeming to change randomly.

 

[mechBgon]

First of all, my own A7V333R is still working (thank goodness), it's the supervisor's nephew's board that isn't. We've RMA'ed that one. I lent him my ECS K7S5A for the meantime.

So I installed Win2000 on my board along with my new hard drive, and I do have an iPanel Deluxe. When I ordered it, I thought it would be giving real-time core temperature readouts

but even if it were taking core temperature readings, the fastest response time for the iPanel is 8 seconds (this is a BIOS option).

Something is making my mouse "stick" every little while. :| After installing all the proper drivers, I'm starting to suspect maybe the iPanel is causing this problem, by intermittently accessing the motherboard for fan/temp/voltage info. Next week I'll unplug the iPanel and see if the mouse stops "sticking" and releasing periodically. I have been told by a couple of knowledgeable people that the iPanel is frowned upon by Asus' tech support people... I had already ordered mine before I found that out, but anyone thinking about an iPanel might want to wait until Tuesday and I'll report whether it's the cause of the mouse problem. If the iPanel is the problem... anyone wanna buy an iPanel Deluxe?

Another possibility is that the board is boggled by my Enermax thermal-controlled case fans, which give really weird RPM readings that change constantly (like 56000 rpm!). I'll try pulling the plug on them too.

I also have a few other pieces of info that may be useful:

1) The FireWire connector on an Antec PerformancePlus case is split into individual one-wire plugs so they can be plugged into the motherboard in the correct order, no matter what pinout is used on the board. However, Asus used a FireWire plug with small pins, and the Antec plugs don't really stay put on the Asus fitting. Don't make plans to use the front FireWire port on this combination of board and case.

2) Alpha PAL8045 heatsink does indeed fit beautifully. The DIMM slots are rather close to the front edge of the heatsink (less than 1cm) so it's not going to get as much air from that side of the sink.

3) the silkscreened info on the board is incorrect in at least one area, and that is the Power LED. The Power LED plug needs to go one pin further to the back than what the silkscreening says.

4) Maybe most of you know this already from reading the reveiw at LostCircuits, but there are undocumented jumpers which control the voltage to the DIMMs and the default voltage is rather high. Read the review for more info. In a nutshell, you should take the jumpers off of the pins to restore normal voltages.

5) the board does indeed talk. The messages can be customized too. Instead of "System completed Power-On Self Test!", you could have "The Force is strong with this one!" or whatever you want to record and program the BIOS with. I haven't tried it. The default messages are spoken by a female voice, slightly more personal-sounding than a typical voicemail system. In my case, the speakers on my monitor were plugged into the output jack on the board's rear panel, but there's a jumper that can send the voice to a system speaker instead.

6) Mikewarrior2, you may be interested to know that at the top of the board there is a set of three solder contacts marked "Thermarise," and the silkscreening indicates that there are two options: Athlon/Duron and Palomino if I recall correctly. Hmmm!

7) Asus has a new 1007 BIOS as of yesterday. It adds two items, according to their site:

Add Item [Thermal shutdown temperature].
Add Option [ Fast Resume ] to control IDE restore at S3.


8) I haven't seen any evidence of QFan varying the fan speed.

 

[Mikewarrior2]

MechBgon,

Are there still Jumpters there? Cause there were jumpers on A7V266Es ... but no diode reading was ever implemented by Asus for that board... curious because Asus uses their own custom ASIC chip.



Mike

 

[RagingGuardian]

I just installed an A7V333 with a Duron 1G overclocked to 1.2G and it says my temps are at 53C:Q The same cpu never peaked over 42C in my Tiger MPX.

 

[mechBgon]

Mikewarrior2, nope, they have solder points, but no pins, where it says "Thermarise."

RagingGuardian, my AthlonXP 1700+ also shows at much higher temperatures in the A7V333 than in my K7S5A. I think Asus has tried to calibrate their socket thermistor so that it reads closer to the actual core temperature of the CPU. Why they did that, I'm not sure... could be they are trying to make it look like they're getting actual core readings when they're not, or it could be they're just trying to be more realistic, where other board makers just let the thermistor give its own temperature instead.

If I could do it all again I'd probably get an A7S333 and a faster CPU, and add a FireWire/USB2.0 card when I actually needed it. Or maybe an nForce board. Maybe my outlook will be more positive when my mouse doesn't stick and the other A7V333 is back from RMA. :|

 

[Mikewarrior2]

RagingGuardian,

That's why cross-motherboard temp comparisons absolutely do not work. Not in one way or another, with any reasonable accuracy .


MechBGon,

Asus has been compensating up their readings for a while now... since hte A7V. And yeah, i think it is compensated to more closely reflect DIE temp of the CPU. Unfortunately, this has lead to hundreds of people claiming that Asus temps are too high, when in actuality, other motherboards are too low .



Mike