Thanks for the help - this board just came back from the RMA service and I'm pretty sure it wasn't doing this before.
-j
Should a core voltage profile look like this?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Should a core voltage profile look like this?
Below is a chart showing Vcore vs. time for my Pentium M machine (760 2.0GHz, 533MHz FSB, i915GMm-HFS, 6600GT PCIe 16 video card, DDR2 RAM, and a Phantom 350 PSU.) I have the voltage set to 1.21(something) in the BIOS. At the far right side of the chart is the point where I changed my SETI client to run at all times in the background. The question for you is: Is that voltage drop normal? Ditto for the spikes. I thought this line was supposed to be FLAT.
Thanks for the help - this board just came back from the RMA service and I'm pretty sure it wasn't doing this before.
-j
Thanks for the help - this board just came back from the RMA service and I'm pretty sure it wasn't doing this before.
-j
Dont know if "should" is the right way to put it, but the Vcore fluctuates on most if not all mobos to some degree or another. Yours looks more stable than on my Abit NF7.
Are you having any stability issues? If not I would not worry too much.
Our overclocking cousins get quite het up about Vcore fluctuations and some go to extremes to flatten them out. Gets them another few Mhz out the system Try doing a google on "droop mod" plus your mobo name and see if one is available. The one on mine is done with a pencil to make the sensing circuit a bit more responsive.
Another thing I found made a small difference (though more than the droop mod) to mine was when I added heatsinks to the mosfets on the mobo's power regulation circuit. This system is being used passively though, so I dont know if it will work for you. Cant do much harm though if your carefull not to short out anything.
Are you having any stability issues? If not I would not worry too much.
Our overclocking cousins get quite het up about Vcore fluctuations and some go to extremes to flatten them out. Gets them another few Mhz out the system Try doing a google on "droop mod" plus your mobo name and see if one is available. The one on mine is done with a pencil to make the sensing circuit a bit more responsive.
Another thing I found made a small difference (though more than the droop mod) to mine was when I added heatsinks to the mosfets on the mobo's power regulation circuit. This system is being used passively though, so I dont know if it will work for you. Cant do much harm though if your carefull not to short out anything.
-
Devonavar
- SPCR Reviewer
- Posts: 1850
- Joined: Sun Sep 21, 2003 11:23 am
- Location: Vancouver, BC, Canada
It looks to me as though SpeedStep is changing the voltage dynamically, but the scale of change seems to small for that to be true.
I think it is changing between two different core voltages as needed, which makes it look very erratic, but is should be perfectly safe. There are basically only two voltages here, one around 1.215 and one around 1.195. The rest of the variation is probably artefacts from the measuring sensor.
As long as you're stable, I don't think there's much to worry about. The CPU can tolerate a fairly large range within VCore, and it will let you know if there's a problem by being unstable.
I'd be more worried if the pattern didn't look so regular (e.g., if there was a diagonal line rather than two horizontal ones) because it might indicate poor regulation which could potentially lead to unacceptable voltage peaks or valleys, but as it is I think it's fine.
I think it is changing between two different core voltages as needed, which makes it look very erratic, but is should be perfectly safe. There are basically only two voltages here, one around 1.215 and one around 1.195. The rest of the variation is probably artefacts from the measuring sensor.
As long as you're stable, I don't think there's much to worry about. The CPU can tolerate a fairly large range within VCore, and it will let you know if there's a problem by being unstable.
I'd be more worried if the pattern didn't look so regular (e.g., if there was a diagonal line rather than two horizontal ones) because it might indicate poor regulation which could potentially lead to unacceptable voltage peaks or valleys, but as it is I think it's fine.
Alright, well that's a relief. I'm not too familiar with this area.
I am having stability issues, the same ones as found here. I sent the motherboard back to AOpen (took a month & 1/2 and was NOT a good experience - I'll spare you the details) and they supposedly fixed it. I got it back, put the machine together and it's still showing the same symptoms. (FTR, the jumpers controlling the fsb speed were set to 400MHz when I got it. I suspect that was their 'fix'.)
Thanks for the help. If you (or anyone else) have any other good ideas I'd love to hear them.
-j
I am having stability issues, the same ones as found here. I sent the motherboard back to AOpen (took a month & 1/2 and was NOT a good experience - I'll spare you the details) and they supposedly fixed it. I got it back, put the machine together and it's still showing the same symptoms. (FTR, the jumpers controlling the fsb speed were set to 400MHz when I got it. I suspect that was their 'fix'.)
Thanks for the help. If you (or anyone else) have any other good ideas I'd love to hear them.
-j
Hmm.
My NF7 shows similar drops when I put it under load (prime 95). That was why I did the research and attempted to flatten out the Vcore. I haven't so far succeeded to get the graph flat, though I am going to continue trying when I have the time. Mine doesn't have speedstep so I'm not convinced that is what you are seeing.
My reason for doing this is slightly different. At standardish settings my mobo is stable, even with the sort of Vcore drops you are seeing on yours. My problem is that I am trying to do on the fly voltage changes via 8rdavcore. I posted here a few weeks ago looking for help, but haven't had any joy yet from other members.
http://forums.silentpcreview.com/viewt ... p?t=25879
It was searching the abit forums that I found mention of the droop mod, and thought that it might help. I certainly did get the machine to run more stable at the extremes (overclock and underclock) but did not manage to get 8rdavcore to behave.
I have a suspicion though that my memory may be playing a part in my problems, and it would be worth, I think investigating it as a factor in your system. Preferably by swapping it out for some known good modules that are well burnt in on a similar system. (any mates running similar?)
Would be usefull if others reading could do a quick check on their voltages and see if its the same?
Oh and I also used speedfan to check my incoming power to the mobo, the graphs dont show any correlation between the droops.
My NF7 shows similar drops when I put it under load (prime 95). That was why I did the research and attempted to flatten out the Vcore. I haven't so far succeeded to get the graph flat, though I am going to continue trying when I have the time. Mine doesn't have speedstep so I'm not convinced that is what you are seeing.
My reason for doing this is slightly different. At standardish settings my mobo is stable, even with the sort of Vcore drops you are seeing on yours. My problem is that I am trying to do on the fly voltage changes via 8rdavcore. I posted here a few weeks ago looking for help, but haven't had any joy yet from other members.
http://forums.silentpcreview.com/viewt ... p?t=25879
It was searching the abit forums that I found mention of the droop mod, and thought that it might help. I certainly did get the machine to run more stable at the extremes (overclock and underclock) but did not manage to get 8rdavcore to behave.
I have a suspicion though that my memory may be playing a part in my problems, and it would be worth, I think investigating it as a factor in your system. Preferably by swapping it out for some known good modules that are well burnt in on a similar system. (any mates running similar?)
Would be usefull if others reading could do a quick check on their voltages and see if its the same?
Oh and I also used speedfan to check my incoming power to the mobo, the graphs dont show any correlation between the droops.
When I look at the other voltages I see a pattern similar to Vcore's. It does look like more or less random noise in the sensor or the system. In my system it's the same pattern of spikes and droops in all the readings, no matter which rail.
I did a search for droop mod for this mb, but no luck. Looks like nobody's tried (or tried and had their mb burst into flames.) Wish I could help with your q's - but I have absolutely no experience whatsoever with what you're trying to do.
I'm wondering now if it could be mosfets overheating. The cpu heatsink I hacked together to get this mb working without it burning up overhangs the area between the CPU socket and the northbridge. I don't think there're mosfets there, but I'm not sure. SPCR had a good pic of this board here, but I can't tell what those components are (and I'm loathe to disassemble everything AGAIN.) Whatever IS under there isn't getting any air circulation, though. Maybe this has something to do with the random freezes under load?
-j
I did a search for droop mod for this mb, but no luck. Looks like nobody's tried (or tried and had their mb burst into flames.) Wish I could help with your q's - but I have absolutely no experience whatsoever with what you're trying to do.
I'm wondering now if it could be mosfets overheating. The cpu heatsink I hacked together to get this mb working without it burning up overhangs the area between the CPU socket and the northbridge. I don't think there're mosfets there, but I'm not sure. SPCR had a good pic of this board here, but I can't tell what those components are (and I'm loathe to disassemble everything AGAIN.) Whatever IS under there isn't getting any air circulation, though. Maybe this has something to do with the random freezes under load?
-j
the picture is a little hard to discren, but I can see a couple of components that could be them. If you look directly downwards from the coils(copper wrapped around blue donoughts with heatshrink at the base) there appears 2 of them. They look like little squares, sitting on shiny metal squares. There are two capacitors beside them and the blue slot to the right which i guess is where your floppy drive plugs in.
There is another component just above your processor fitting again next to a coil, this time a green one (the coil that is). "CPU FAN" is screen printed right next to it. Not sure that if it is one or not. There are several around the board like this.
I see another at the top right of the board again next to a green coil and three capacitors.
In my board, I have six all with caps and coils right next to them. But mine is Athlon based, and this gives the regulation for 3 lines
Not sure about pentium boards though. My guess is the differnet architectures have different power requirements. I am not that knowledgeable with this stuff, just was lucky that the OC boys like their Athlons, so could find plenty of BB posts.
I doubt though that its the mosfets at fault. They are clearly designed to run without heatsinks, though there may be benefits from heat sinking them. The manufacturers dont neccessarily build their products to be the best they can be, but rather to be good enough and cheap enough to build and make a profit on.
However if you are doing something that is outside the specs of the board, or processor it may help.
Watch though you dont send yourself chasing red herrings. I was convinced that sorting out my mosfets would resolve the problem.
As it happened it sorted out another issue, hissing coils. I also feel more comfortable running the board in a passive system, And it did allow more stability at higher clock speeds (big deal, I am not that interested in OCing)
Have another look at your memory, its in my experiance more likely to cause problems than mobo's do. You could try heat spreaders, or as i suggested before borrowing some good performing memory to elliminate it from your enquiries.
The other offender could be your PSU. Do you have a spare one you could try?
There is another component just above your processor fitting again next to a coil, this time a green one (the coil that is). "CPU FAN" is screen printed right next to it. Not sure that if it is one or not. There are several around the board like this.
I see another at the top right of the board again next to a green coil and three capacitors.
In my board, I have six all with caps and coils right next to them. But mine is Athlon based, and this gives the regulation for 3 lines
Not sure about pentium boards though. My guess is the differnet architectures have different power requirements. I am not that knowledgeable with this stuff, just was lucky that the OC boys like their Athlons, so could find plenty of BB posts.
I doubt though that its the mosfets at fault. They are clearly designed to run without heatsinks, though there may be benefits from heat sinking them. The manufacturers dont neccessarily build their products to be the best they can be, but rather to be good enough and cheap enough to build and make a profit on.
However if you are doing something that is outside the specs of the board, or processor it may help.
Watch though you dont send yourself chasing red herrings. I was convinced that sorting out my mosfets would resolve the problem.
As it happened it sorted out another issue, hissing coils. I also feel more comfortable running the board in a passive system, And it did allow more stability at higher clock speeds (big deal, I am not that interested in OCing)
Have another look at your memory, its in my experiance more likely to cause problems than mobo's do. You could try heat spreaders, or as i suggested before borrowing some good performing memory to elliminate it from your enquiries.
The other offender could be your PSU. Do you have a spare one you could try?
Oh yeah, I tried a search on droop modding for your board but found nothing really of use. I guess its the sort of thing the OC boys get horny over.
And yet again, if anyone else that reads this and uses speadfan can check theirs under similar conditions, I would be interested to know if its normal. I know that it is normal for my board to do it. But not if its a general thing.
And yet again, if anyone else that reads this and uses speadfan can check theirs under similar conditions, I would be interested to know if its normal. I know that it is normal for my board to do it. But not if its a general thing.
I'm currently in the process of ordering my dream machine, and am finding it very hard to wait, to try things out for myself.
I read this post with interest... and learned about droop modding (which sounded like something to add to my todo list until I worked out that it involves taking a soldering iron to your motherboard -shudder-)... and about some useful software products.
In my travels, I found this (Why Not To Trust Your Motherboard Voltage Reading) hidden away on Antec's website. (Irritatingly, they don't actually answer the question "Why"!)
Now Antec are discussing the difference between the 12V value reported by the motherboard, and the value measured by a multimeter. They don't talk about Vcore fluctuations, but I would have thought that if the motherboard can report something as basic as the 12V value wrong, it can surely get the Vcore value wrong too, can't it?
I think the interesting thing is the text:
I read this post with interest... and learned about droop modding (which sounded like something to add to my todo list until I worked out that it involves taking a soldering iron to your motherboard -shudder-)... and about some useful software products.
In my travels, I found this (Why Not To Trust Your Motherboard Voltage Reading) hidden away on Antec's website. (Irritatingly, they don't actually answer the question "Why"!)
Now Antec are discussing the difference between the 12V value reported by the motherboard, and the value measured by a multimeter. They don't talk about Vcore fluctuations, but I would have thought that if the motherboard can report something as basic as the 12V value wrong, it can surely get the Vcore value wrong too, can't it?
I think the interesting thing is the text:
Antec's article wrote: Note: the voltage reported by the motherboard actually fluctuated between 11.58, 11.64, and 11.70 Volts. We happened to snap the picture at the repeating low point. The voltage reported by the multimeter did not vary during the 15 minute photographic session.
I'm not sure about that test. Looks to me like they are testing the molex.
i may be wrong, but the mobo and the hard drives are on seperate rails, with seperate current limits. So IF! I understand it right, may not be affected by each other.
http://www.formfactors.org/developer%5 ... ic_br2.pdf
Of course I am probably wrong, anyone have a deeper than my cursary knowledge on this?
i may be wrong, but the mobo and the hard drives are on seperate rails, with seperate current limits. So IF! I understand it right, may not be affected by each other.
http://www.formfactors.org/developer%5 ... ic_br2.pdf
Of course I am probably wrong, anyone have a deeper than my cursary knowledge on this?
I just reread SPCR's power supply fundamentals section and this useful forum post and it seems to me that it's one big mess. My impression is that, if you have a single 12v rail, the same rail supplies both the hard drives (if it even uses it -- I think my IBM/Hitachi's don't, for instance) and the motherboard. It also occurs to me that testing the molex is what you want, actually, since that gives you the voltage as it's delivered to the device, not interpreted by some sensor. Then again, could be wrong.i may be wrong, but the mobo and the hard drives are on seperate rails, with seperate current limits.
Yeah, that Antec article is misnamed, isn't it
I went out and bought a Zalman 7000B last night, to replace the heatsink that seems to be oversize and blocking airflow. Shortly after I did that, I called AOpen to tell them I was still having prob's with the motherboard, and they're going to cross-ship a new one to me. I had trouble with AOpen's tech support department earlier (mis-addressed packages, done three different ways. It was interesting) but this is an encouraging sign.
Thanks for taking a look at the mb pic, justblair. I managed to find the components you were referring to. Guess those aren't mosfets between the cpu socket and northbridge. I've been chasing red herrings since June with this board, which is especially bothersome to me since it's by far the most expensive machine I've put together yet
-j
