Computer won't POST/boot

[angelkiller]

Originally Posted by 95Viper

I am not really up on the i7s and such, will be soon though(gathering parts).

Are your memory settings(timings, voltage, etc.) in the bios matching your ram's(like 8-8-8-24).
Maybe a bump in voltage on the ram. Just asking... isn't that ram rated at like 1.65V and you are running it at 1.52V?

If the ram is checking out ok, try swapping around and see if it is maybe a bad memory slot or check for foreign objects and dust in the memory slots.

Just for grins and giggles, are your standoffs, screws, and phenolic washers properly done.
No shorts or such from an off alignment and the like?

Like I said, just asking 'cause I ain't up on the memory tweaks and all for the i7\x58 platform... yet.

Oh yeah, that error code 0x0000001A is usually memory associated.

You see, I've had this problem from day 1. So I never got around to changing my memory settings from defaults. Defaults are 1066 MHz, 1.5x v, 8-8-8-xx. I figured since it can do 1600 MHz 8-8-8-xx at 1.65v, then it can do 1066, same timings at 1.5v. I can try to bump up the voltage... but right now this thing barely lets me in the bios.

My ram passed memtest a few months ago. I cant get this thing to boot anything so I can't run it now. So bad ram is a possibility.

I took the mobo out of the case and guess what? Found an unused standoff. (no hole in the mobo where the standoff was) Took that out. Booted up. And then it shut down. I figure that if it's a short, it would either not turn on, or turn off immediately. The fact that it runs ok

Regarding the weird memory recognition issue, I was able to get this. It shows yet another memory config. The sticks may be in a different order, but they're in the same slots. ??

Originally Posted by westom

Getting it into Windows to confirm a supply is good can come later. Currently it will not even go into Windows. So before it even tries to get to Windows or crashes, what are the voltages. If power is the reason why it does not get into Windows, then power is will appear defective before or after it tries.

Provide numbers for those six wires as described earlier. Especially, what happens on any one red, orange, and yellow wire as the system attempts to boot. Strange voltage numbers may explain why some memory is seen and another not. Either some memory is intermittent or motherboard semiconductors act intermittently.

Computer is being extremely uncooperative. Thing won't even boot anymore. Might need to wait longer when I unplug the PSU in between shutdowns. When it does boot, I can't get more than 3 min out of it. And it's been pretty consistent, which is unusual. Here's what I have so far, will edit if the computer boots.

yellow - 12.24v
red - 5.11v
purple - 3.28v (PSU was plugged in, computer off)

No fluxuations outside of a few .01v that I can see with my MM.



I should also add that I contacted Silverstone Support a few days ago and told them I'm still having issues.

Originally Posted by Silverstone Support

I double check with the RMA dept and they said they send the same unit because the power supply passed the test and a report was sent with the unit

Broken english is vague, but I'm going to assume that my PSU was tested with whatever they test PSUs with. They sent the same unit back to me, confirmed by serial number. So Silverstone is saying the PSU is good.

[westom]

Originally Posted by angelkiller

Here's what I have so far, will edit if the computer boots.
yellow - 12.24v
red - 5.11v
purple - 3.28v (PSU was plugged in, computer off)

Those numbers are good considering other important numbers were not provided. Those numbers suggest 3 volt load is heavier loaded compared to other two voltages. A better reply would be later once the system actually is loaded. So far, power system looks good.

However those numbers should not exist 'when plugged in but computer off'. Those numbers should only exist when computer is powered on - after power switch is pressed.

Missing are numbers from the green, gray, and purple wires. All numbers are important. Otherwise some information in three provided numbers is lost. Those other three numbers should be taken both before and when the power switch is pressed.

[95Viper]

Been, reading around and some have resolved the problem with a bios update, some reverted to older bios, and some by shorting the 1st bios chip to fool the system to use the backup bios.
But, I would not recommend the bios route 'til you can get it running stable.

Others, it seems the majority, have done the options below.

Try some of these:

1. backing off the cooling bracket a bit.
2. re-seat the CPU
3. removing the cpu and do a very close examination of the socket for bent or mis-aligning pins and any discoloration on socket or cpu.

[Bundy]

It it reasonable to say that you get more success with less RAM? It seems so from your posts. This would mean you are having a voltage issue either on the memory controller or the RAM.

I occasionally get the same problem but i have an x48 chipset and the memory controller in in the NB. To fix mine, I have to boot up with 1 or 2 sticks, then bump up NB volts before I populate the remaining RAM slots.

For you, I'm not sure but can you try booting with only 1 stick or RAM? If successful, can you raise the RAM and CPU volts a little before populating the remaining slots?

[angelkiller]

Originally Posted by westom

Those numbers are good considering other important numbers were not provided. Those numbers suggest 3 volt load is heavier loaded compared to other two voltages. A better reply would be later once the system actually is loaded. So far, power system looks good.

However those numbers should not exist 'when plugged in but computer off'. Those numbers should only exist when computer is powered on - after power switch is pressed.

Missing are numbers from the green, gray, and purple wires. All numbers are important. Otherwise some information in three provided numbers is lost. Those other three numbers should be taken both before and when the power switch is pressed.

Only the purple wire had voltage when the system was off. But according to the ATX spec, the purple wire is 5v standby. Is it a problem that it's at 3.3v?

Ok here is the full list of numbers. Some were taken last night, some this morning. Exact same hardware config, some bios settings may be slightly different, can't be 100% sure they're the same.

          Off     On
purple    3.27    5.06
red               5.11
orange            3.33
yellow    0.44    12.24
green     2.94    0.07
gray              5.14

I'm no expert, but everything looks to be normal. I don't see anything unusual. Computer shut down after about 8minutes. It lasted longer because it was unplugged all night. At least it's following some kind of pattern. (yay?) Here are some bios shots I was able to get before the shutdown.

MIT Main Page
Advnced Memory Settings
Advanced Frequency Settings
Advanced Voltage Settings
Advanced Voltage Settings (the rest of it)
Power Management
PC Health


In the interest of providing as much info as possible, I should say that sometimes the bios says something along the lines of "bios corrupt, recovering from backup". I figured that this was caused by the numerous power on and random shutdowns the system goes through. This only happens about once every 10+ boots. Not often at all, but I've seen it happen about three or four times since I've been troubleshooting this computer. No idea what this means in terms of possible problems. Maybe the first bios chip is corrupt? Assuming it is, would it cause the issues I'm having now?

@ 95viper and Bundy: I haven't forgotten your posts, I just wanted a consistent hardware setup so I could get those numbers for westom.

[Arrakis+9]

Originally Posted by angelkiller

In the interest of providing as much info as possible, I should say that sometimes the bios says something along the lines of "bios corrupt, recovering from backup". I figured that this was caused by the numerous power on and random shutdowns the system goes through. This only happens about once every 10+ boots. Not often at all, but I've seen it happen about three or four times since I've been troubleshooting this computer. No idea what this means in terms of possible problems. Maybe the first bios chip is corrupt? Assuming it is, would it cause the issues I'm having now?

it could, its not unusual and you cant rule anything out on a "hunch" in trouble shooting.

[angelkiller]

Originally Posted by 95Viper

Been, reading around and some have resolved the problem with a bios update, some reverted to older bios, and some by shorting the 1st bios chip to fool the system to use the backup bios.
But, I would not recommend the bios route 'til you can get it running stable.

Others, it seems the majority, have done the options below.

Try some of these:

1. backing off the cooling bracket a bit.
2. re-seat the CPU
3. removing the cpu and do a very close examination of the socket for bent or mis-aligning pins and any discoloration on socket or cpu.

I updated the BIOS sometime last week. No noticeable differences. I'm a bit weary of flashing an older bios because I've never seen a newer bios cause issues. Especially since this one has been out for over a month. (would have been pulled if it caused problems) However, I'm not going to eliminate anything... it's a possibility.

Shorting the first bios chip is something I'd be willing to do. Any links?

What do you mean by "backing off the cooling bracket a bit"? Do you mean loosen the waterblock on the CPU?

I re seated the CPU yesterday. I didn't look for bent pins and such, but I'll do that in a sec.

Originally Posted by Bundy

It it reasonable to say that you get more success with less RAM? It seems so from your posts. This would mean you are having a voltage issue either on the memory controller or the RAM.

I occasionally get the same problem but i have an x48 chipset and the memory controller in in the NB. To fix mine, I have to boot up with 1 or 2 sticks, then bump up NB volts before I populate the remaining RAM slots.

For you, I'm not sure but can you try booting with only 1 stick or RAM? If successful, can you raise the RAM and CPU volts a little before populating the remaining slots?

There has been no correlation between number of ram sticks and booting. No correlation between which slot is populated and booting. I haven't tried every possibility, but I'll do that just to be sure.

I think I have a real problem if I have to up voltages to get the system to boot at stock clocks. (and UNDER stock clocks for the ram) I will up the memory's voltage to 1.65 and try every combination of a single stick of ram.

[Arrakis+9]

While your at it, take a shot of canned air to the socket & the underside of the cpu for good measure

[DRDNA]

I bet its the RAM.
Even if its hard to pin point the stick or sticks I feel its still the Ram
I have seen ram that will work fine individually but would not work worth crap in pairs..

Also maybe you have not set the ram voltage properly and thus your issues. But like i said I feel this is ram related to the utmost. GL

[angelkiller]

Originally Posted by Arrakis+9

While your at it, take a shot of canned air to the socket & the underside of the cpu for good measure

No canned air.

Originally Posted by DRDNA

I bet its the RAM.
Even if its hard to pin point the stick or sticks I feel its still the Ram
I have seen ram that will work fine individually but would not work worth crap in pairs..

Also maybe you have not set the ram voltage properly and thus your issues. But like i said I feel this is ram related to the utmost. GL

I'm not 100% sure it's the ram just yet. These same sticks did pass memtest error free a while back. (doesn't prove anything, but I think it does say something though) I do think the issue is either A) The psu is stable, but something in it causes it to randomly reset. When it works, it works, but sometimes it'll randomly not work. B) Mobo is incompatible with the RAM or has one (or more) bad RAM slots. C) Bad stick(s) of RAM. But I don't have any more evidence to narrow it further than that.

True, I have not set the ram voltage to 1.65, but then again, the ram is only running at 1066MHz as opposed to 1600MHz. (2/3 of rated speed) I would think that it would be able to run 1066 on 1.5v. But to confirm, I'm going to set the ram speed higher when I test all the sticks/slots.

Here's the CPU and the socket. Cheap camera is cheap.

Imprint left from waterblock
Imprint on the Apogee XT
Socket #1
Socket #2
Socket #3
The CPU
The back of the CPU
The back of the CPU (cpu rotated 180 degrees)

Pics don't do justice. Everything looks good from my eyes. No bent pins, no burn marks.

[brandonwh64]

Set your ram at 1066mhz 1.65v CAS 9 and see if it will boot fine

[angelkiller]

Originally Posted by brandonwh64

Set your ram at 1066mhz 1.65v CAS 9 and see if it will boot fine

Well, I did that and it let me get into windows. But two minutes at the windows desktop (while I tried to get a screenie of cpuz) it turned off.

But I did get into Windows, which is the first time that's happened in a few days. Random occurance? Or did these settings actually make a difference? The computer's behavior is too erratic to tell imo.



Edit: Here's the image I got from CPU-Z after I used those settings.

[brandonwh64]

Originally Posted by angelkiller

Well, I did that and it let me get into windows. But two minutes at the windows desktop (while I tried to get a screenie of cpuz) it turned off.

But I did get into Windows, which is the first time that's happened in a few days. Random occurance? Or did these settings actually make a difference? The computer's behavior is too erratic to tell imo.

http://img.techpowerup.org/110208/an...omputer-14.png

Edit: Here's the image I got from CPU-Z after I used those settings.

Thats no coincidence.... RAM IS BAD or one stick is.

[angelkiller]

Originally Posted by brandonwh64

Thats no coincidence.... RAM IS BAD or one stick is.

I thought if the RAM is bad then the computer wouldn't boot at all or would blue screen all over the place.

Can bad RAM explain the shutdowns? Like is random shutdowns a sign of bad ram? What about the "I have to unplug it and wait a few minutes before it'll power on again" thing?

I just booted the system with only one stick of ram. It didn't post. No bios screen or anything. It stayed on for about a minute then shut off. We might be getting somewhere. Lemme try a different stick real quick.

[angelkiller]

Memtest only sees 4GB also... but then it lists out all three dimms.

Computer turned off within a minute after taking this pic.


Edit: Tried it again, got to three minutes something.

[westom]

Originally Posted by angelkiller

Off On
purple 3.27 5.06
red 5.11
orange 3.33
yellow 0.44 12.24
green 2.94 0.07
gray 5.14

You have a problem with the 5V Standby power when power is off. Apparently something normally powered by the red 5 V is drawing too much power from the purple 5V wire when red 5V is off. That device is typically used to detect an event that causes a power controller to power on the supply.

Gray and other wires had a voltage. Even zero volts is a voltage.

Gray wire confirms that the red orange, and yellow wire voltages are remaining sufficient once the system is powered.

Green wire is ordering the power supply to power on. But its 'stay off' voltage (although sufficient) does not have typically sufficient noise margin. However, that would be explained by the clearly way too low purple wire voltage (3.27). Confirms you did not make a measuring error. Confirms a problem exists on the 5V Standby circuit.

Finally a concept for everyone. They say on CSI, "Follow the evidence". Most of the preceding posts are examples of how to not diagnosis a problem. Also called shotgunning or wild speculation. Following the evidence means first obtaining facts and numbers to identify a problem. Problem (at least the first one) is clear: excessively loaded 5V Standby power supply. Next step is to discover what is causing excessive loading. And that also means numbers. No part need be replaced or swapped until a defect is identified. The difference between "Follow the evidence" and a bad diagnostic procedure called "Shotgunning".

How to identify the defective part. First, and this is what everyone should know. Never disconnect or install anything until an AC power cord is removed from the receptacle. And static electric charges are discharged. That purple wire voltage (and another reasons for getting numbers) says why parts can be damaged if a power cord is not first removed. Monitoring the purples wire when a power cord is disconnected shows how long to wait before doing any changes.

Among the many ways of identifying excessive loading is to measure that 5V Standby current with the meter (probably too difficult). Or removing various parts to list all parts that cause 5 volts to increase to at least 4.9 volts.

Once we find or replace the defective part, then we start all over again with voltage measurements searching for any other defects or to identify the entire power 'system' as good. We don't do anything else until the power 'system' is first known good - without doubt.

[angelkiller]

Originally Posted by westom

You have a problem with the 5V Standby power when power is off. Apparently something normally powered by the red 5 V is drawing too much power from the purple 5V wire when red 5V is off. That device is typically used to detect an event that causes a power controller to power on the supply.

Gray and other wires had a voltage. Even zero volts is a voltage.

Gray wire confirms that the red orange, and yellow wire voltages are remaining sufficient once the system is powered.

Green wire is ordering the power supply to power on. But its 'stay off' voltage (although sufficient) does not have typically sufficient noise margin. However, that would be explained by the clearly way too low purple wire voltage (3.27). Confirms you did not make a measuring error. Confirms a problem exists on the 5V Standby circuit.

Finally a concept for everyone. They say on CSI, "Follow the evidence". Most of the preceding posts are examples of how to not diagnosis a problem. Also called shotgunning or wild speculation. Following the evidence means first obtaining facts and numbers to identify a problem. Problem (at least the first one) is clear: excessively loaded 5V Standby power supply. Next step is to discover what is causing excessive loading. And that also means numbers. No part need be replaced or swapped until a defect is identified. The difference between "Follow the evidence" and a bad diagnostic procedure called "Shotgunning".

How to identify the defective part. First, and this is what everyone should know. Never disconnect or install anything until an AC power cord is removed from the receptacle. And static electric charges are discharged. That purple wire voltage (and another reasons for getting numbers) says why parts can be damaged if a power cord is not first removed. Monitoring the purples wire when a power cord is disconnected shows how long to wait before doing any changes.

Among the many ways of identifying excessive loading is to measure that 5V Standby current with the meter (probably too difficult). Or removing various parts to list all parts that cause 5 volts to increase to at least 4.9 volts.

Once we find or replace the defective part, then we start all over again with voltage measurements searching for any other defects or to identify the entire power 'system' as good. We don't do anything else until the power 'system' is first known good - without doubt.

Your logic is flawless.

PSU has been unpugged for 3+ hours. Purple wire measures zero. I plug it in, still zero. Flip the PSU switch and the purple wire goes to 5.11v. Normal operation. I turn on the computer and start running memtest again. Purple wire slowly drops to 5.06v over a couple of minutes. Memtest runs for about twelve minutes (error free) then the computer shuts down. As soon as the computer shuts down the purple wire falls to 3.24v. I flip the psu switch off. Purple wire goes to 0.01v. I flip the psu switch back on, back to 5.11v. I turn the computer on, it only lasts for about a minute before it shuts down again. Purple wire again falls to 3.24v. After about ten minutes of not doing anything, the purple wire has stabilized at 3.28v. Computer will not turn on in this state.

Unplugging my usb card reader caused the voltage to go from 3.28v to 3.30v. Mouse had no effect. Flash drive, ethernet cable, headphones, PS2 keyboard all had no effect. Unplugging the ATX cable caused the voltage to rise to 3.31v. Unplugging my hard drive, SSD and optical drive and case fans had no effect. Accidentially nudging the ATX cable (coming out of the PSU; it's modular) caused the voltage to fluxuate between 3.31v and 3.33v. Unplugging the 8pin power connector as well as the two 6pin PCI-E cables had no effect. Just to be sure, I unplugged the ATX cable from the PSU, still 3.31v. At this point, everything is disconnected from the PSU.



Can we make any conclusions?

[westom]

Originally Posted by angelkiller

As soon as the computer shuts down the purple wire falls to 3.24v. I flip the psu switch off. Purple wire goes to 0.01v. I flip the psu switch back on, back to 5.11v. I turn the computer on, it only lasts for about a minute before it shuts down again. Purple wire again falls to 3.24v. After about ten minutes of not doing anything, the purple wire has stabilized at 3.28v. Computer will not turn on in this state.

My initial thought was the 5V Standby was overloaded. After reading this and viewing the numbers, I am wondering if the 3.3 volts is somehow getting shorted to the 5V Standby.

Well, you post suggests that all removable peripherals do not cause this problem. That simply leaves two suspects. Peripherals on the motherboard (ie NIC, keyboard interface, other devices that must have 5V Standby power to awaken the system). Well, you might try to follow the 5V Standby traces looking for a mechanical short. But I doubt you will find any. This defective might be in one of the semiconductors. As it gets warm, it shorts.

Applying a cold pack to that semiconductor might even identify the defective semiconductor. Being colder means it does not short or takes longer to short. Curious experiments. Apparently something on the motherboard is pulling 5V Standby to 3.3 volts. Likely: something is shorting 5V Standby to 3.3 volt rail. Or, less likely, that short is happening inside the PSU. Occurring long after the power supply vendor stop testing.

Anything you might do just to confirm a 'getting warm' power supply is not failing might be useful (ie use the supply elsewhere and monitor its voltage). Two suspects remain. Unfortunately the one more likely to short 5V Standby to 3.3 volts in on the motherboard. And if, per chance, it is the supply, you now have facts to get that supply replaced under warranty.

BTW, one reason for doing this. Yours is a unique problem never seen before. We both get to learn.

[angelkiller]

Originally Posted by westom

Anything you might do just to confirm a 'getting warm' power supply is not failing might be useful (ie use the supply elsewhere and monitor its voltage). Two suspects remain. Unfortunately the one more likely to short 5V Standby to 3.3 volts in on the motherboard. And if, per chance, it is the supply, you now have facts to get that supply replaced under warranty.

BTW, one reason for doing this. Yours is a unique problem never seen before. We both get to learn.

What do you mean by 'elsewhere'? Another computer? Or in a warm/cool environment.

[westom]

Originally Posted by angelkiller

What do you mean by 'elsewhere'? Another computer? Or in a warm/cool environment.

'Elsewhere' might be load of another computer. Or might be loading the 3.3v and 5V Standby with 2 and 10 watt resistors from Radio Shack. (Another computer is easier.)

'Elsewhere' is suggested because finding a short on a motherboard (especially if it only occurs with heat) without advanced test equipment is difficult. 'Elsewhere' is to only eliminate the less likely suspect from a suspect list of two.

Another technique we use (but might require too much time) is to use temperature. For example, refrigerate the computer. Then touch a soldering iron to each IC after power up. Does the resulting heat cause a failure in less than 30 seconds (and not when motherboard is cooler)? Heat is a diagnostic tool to isolate a failure that has now been clearly identified.

Which chips are suspect? View the BIOS. Any peripheral that can force a computer restart is on the suspect list. Most functions are in the South Bridge chip. A large chip that connects to PCI bus, USB, Ethernet, and other peripherals. (North bridge connects to memory.)

If placing a bet, odds overwhelmingly indict the motherboard. 3.3 volts and 5V Standby are two separate power supplies inside the PSU. Would be independent of each other inside a PSU. And are common to motherboard chips.

[rangor]

angelkiller: if you end up RMA'ing the motherboard, please update this thread to see how it works out. With all the deducing going on, this has turned out better than some mystery novels I've read.

[angelkiller]

Originally Posted by westom

'Elsewhere' might be load of another computer. Or might be loading the 3.3v and 5V Standby with 2 and 10 watt resistors from Radio Shack. (Another computer is easier.)

'Elsewhere' is suggested because finding a short on a motherboard (especially if it only occurs with heat) without advanced test equipment is difficult. 'Elsewhere' is to only eliminate the less likely suspect from a suspect list of two.

Another technique we use (but might require too much time) is to use temperature. For example, refrigerate the computer. Then touch a soldering iron to each IC after power up. Does the resulting heat cause a failure in less than 30 seconds (and not when motherboard is cooler)? Heat is a diagnostic tool to isolate a failure that has now been clearly identified.

Which chips are suspect? View the BIOS. Any peripheral that can force a computer restart is on the suspect list. Most functions are in the South Bridge chip. A large chip that connects to PCI bus, USB, Ethernet, and other peripherals. (North bridge connects to memory.)

If placing a bet, odds overwhelmingly indict the motherboard. 3.3 volts and 5V Standby are two separate power supplies inside the PSU. Would be independent of each other inside a PSU. And are common to motherboard chips.

I don't have any other computers with me right now and there's no way this thing will fit in my mini-fridge.

Still have one question. We saw that the PSU still outputs 3.3v on the 5vSB rail after unplugging everything. Isn't this a clear indication that the PSU is at fault? If the mobo was suspect, when I unplugged it from the PSU, wouldn't the 5vSB rail return to normal?

Originally Posted by rangor

angelkiller: if you end up RMA'ing the motherboard, please update this thread to see how it works out. With all the deducing going on, this has turned out better than some mystery novels I've read.

Will do. I'm sure this is much more entertaining to read than to experience.

[westom]

Originally Posted by angelkiller

Still have one question. We saw that the PSU still outputs 3.3v on the 5vSB rail after unplugging everything.

I did not see that. I saw that the 5VSB remains at 5 for a period of time. Then suddenly dropped to 3.3 volts. But always with the motherboard connected.

5VSB is a power supply inside the PSU separate from all other voltages. It occupies a common box. But operates independently. Some semiconductors on a motherboard are powered both by 5VSB and 3.3 volts. Therein lies a reason for a short between two voltages after a time period. I did not see where 5VSB was 3.3 volts when the PSU was disconnected from the motherboard.

Motherboard and PSU are two suspects. But 5VSB should not share any circuits with the 3.3 VDC inside the supply. Maybe something is intermittently shorting. But statistically, a more likely reason for 3.3 shorted to 5 VSB are semiconductors (or PC traces) on a motherboard.

Which was why a cold pack and soldering iron testing was suggested. Temperature being an excellent diagnostic tool (combined with the meter readings) to isolate a problem to a particular semiconductor or circuit.

As I noted before, yours is a unique (and interesting) failure. I never saw this type failure before.

[angelkiller]

Finally some good news. Sent the PSU back to Silverstone again and they finally sent me a new PSU. I got a ST1000-NV which is a more recent revision of my vanilla ST1000. I put it in and it worked. Everything worked. And it's been working ever since.

It was the PSU. Case closed. Thanks to everyone for all the help throughout this process. It's appreciated.



I also want to mention that the analysis in this thread was pretty spot on. I was pretty sure it was the PSU because unplugging it and waiting, consistently allowed me to boot the computer again. To be honest, just about everything else had been ruled out. Ram was good, no graphical artifacts, no way bad hard drives could cause this, still had the problem when computer was outside of the case so no shorts, and the computer could pass stress tests so no overheating. Nothing else but the PSU. All the problems I had could be caused by a bad PSU and my PSU was the only thing I couldn't say was 100% working. It got replaced and the problems stopped. Problem solved.

[OneMoar]

silverstone are junk