Help Needed Diagnosing Startup Issue after Power Disconnect

E

Epos7

Gawd
Joined
Aug 31, 2015
Messages
892
I first noticed this problem a few weeks ago, and initially thought it was an overheating issue, but I don't believe that to be the case.

After a power surge, I returned to my computer to find it powered on, but with no image on the display, and with the graphics card fan spinning at 100%. I shut the system down, and attempted to restart. The computer powered back on, but the graphics card fan returned to 100% speed, and I still had no output to the display. I powered the computer down, returned ten minutes later, and it started normally.

Since then, I have noticed the system exhibits this behavior any time it is disconnected from power. The first time it is powered on after being reconnected, the graphics card fan spins up to 100% and will not output to the display. The only way to start the system after it is disconnected from power is to turn it on, turn it off, wait five to ten minutes, then turn it on again.

This build is only about a month old, and since it wasn't disconnected from power before the power surge, I can't say if the issue appeared after the power surge, or was always present. Either way, a power conditioner is in my future. I still need to determine what is causing the startup issues after power disconnection.

Windows 10
ASRock X99E-ITX/ac
Core i7 5820K
Noctua C14
Corsair LPX DDR4 8GB x 2
Silverstone SX500-LG PSU
ASUS GTX 960 Mini
Samsung 850 EVO

I have scoured the internets and haven't been able to find anyone who had a similar issue. I'm guessing it's either a motherboard or video card issue. If the system remains connected to power, it boots and shuts down normally. It doesn't show any instability under stress testing (Prime95). Absolutely no issues unless the power goes out, or I unplug the computer.

Any insight would be greatly appreciated! I have tried contacting ASRock to see if it could be a BIOS issue, but did not hear back.
 
Have you reset the BIOS to optimized defaults yet? Make sure there isn't something in there messing with which video to start or something. Try that first I guess.
 
I don't have an option to load optimized defaults, but I did reset defaults, and that made no difference.

I should also note that resetting the CMOS causes the same issue:

Shut down computer > press reset CMOS button > release reset CMOS button > computer starts on its own (gfx fan spins up to 100% and no display)

Also, if I modify the BIOS settings, save and exit, upon exit the gfx fan spins to 100% then the computer shuts down on its own, and I have to wait 5-10 minutes to start.
 
No luck trying to narrow down the issue, and no response still from ASRock. I think the next step, unfortunately, is to just start replacing parts. I think I'm going to start with the GFX card.
 
ASRock got back to me and believe it is a motherboard issue. They have offered to issue an RMA to fix the problem.

On the other hand, we had another power surge yesterday, and now the computer functions normally. :confused:

I'm at a loss.
 
Any specific features or specs I should look for in a power conditioner? Not a ton of information to be found on the subject, and a lot of it is geared toward audio/video setups. From what I gather, voltage regulation isn't very important with modern power supplies, but surge protection and noise filtration are. This is where things get murky as there are different ways to accomplish both and I think some are better than others.
 
Back to having the same issue.

The more I read about power conditioners, the more it seems the cheap ones are no more than a glorified surge protector. It also sounds like all modern PSUs have line conditioning circuitry built in. It seems like a power conditioner might only really be necessary if you live in a region that suffers from brown-outs, which is pretty rare in most first world countries now. Unfortunately there isn't a whole lot of information available, and most power conditioners are marketed toward the audio industry, so there is even less real information.

This is an issue that only occurs when we have a blackout (rare) or I unplug the computer (also pretty rare). Is it worth replacing a bunch of components to solve? I'll happily do so if it hints at a deeper issue, but if this is the only thing I have to deal with, I could save myself a whole bunch of time by just living with it.
 
Epos7 said:
This is an issue that only occurs when we have a blackout (rare) or I unplug the computer (also pretty rare). Is it worth replacing a bunch of components to solve?
Click to expand...
So much to address / correct. This will start with your power cycling issues.

Normal is for a defective power system to boot and operate a computer. Often this defect becomes more obvious in a warm room (ie 100 degree F - a normal temperature for any properly functioning computer). Eventually the computer becomes intermittent at 70 degrees. Heat is not a reason for failure. Heat is a diagnostic tool to find hardware defects.

Defined was a surge. If so, then other destroyed appliances must also be listed. What else was damaged? Or was it a blackout. A surge is a high voltage (for example 1000 volts). A blackout is something completely different (zero volts). Unfortunately many internet discussions do not even know the difference. Then recommend a surge protector to cure everything. To obtain a useful reply means the differences is critical.

That defect would be obvious when a computer is new (if using a meter and some directions).

Central part of a computer's power 'system' is its controller. Inputs include that front panel power button. A controller decides when a PSU will power on and when a CPU is permitted to execute. At one point, a PSU powered on but the CPU (apparently) was not permitted to operate. So fans were never told by the CPU to slow down.

That is speculation tempered by observation. To say anything accurate means a digital meter, some requested instructions, and minutes of labor. Posting resulting 3 digit numbers means those who really know how a computer works can say what is happening, what has failed, and maybe why the failure exist.

You currently have two choices. Just keep replacing good parts until something works. Or identify a defect immediately (to only replace the defective part) by obtaining some numbers. Your choice.

No magic box solution exists. Plenty of urban myths need correction. But for now, let's address a current problem. Averting future ones comes later.
 
Thanks for the reply.

Situation was a blackout, not a surge. That was a mistake in my original post. Second time posting about the issue, first was here:

http://hardforum.com/showthread.php?t=1879838&highlight=

I didn't identify it as a power surge at the time, but started to believe that was the case after some of the replies I received.

I have a meter, if you can tell me what I need to measure or point me toward some directions it would be very helpful.
 
Restore every connection as when the computer worked. AC power cord connected to a receptacle. Computer not on. Set a digital meter to 20 VDC. Attach its black probe to the chassis (bare metal; not paint).

Locate a purple wire (pin 9) from PSU to where it attaches to the motherboard. Use a red probe to touch that wire inside a nylon connector that attaches to motherboard. If necessary, make that connection using a needle or paper clip. It should read somewhere around 5 volts. Record that number to three digits.

That purple wire number should be done for both conditions - when the computer is off and later does not properly power up. And when computer is off and does power up normally.

Next, do same with a green wire (pin 16). Then press computer's Power On button. Monitor how meter changes and what it eventually settles to. First number should be something well above 2.6. Second number should be something near to zero, Actual numbers and time to change (behavior) are relevant. Again, this must be done when power up is abnormal and when normal.

Repeat same to a gray wire (pin 8) for both abnormal startup and normal startup. Note a higher starting voltage, a lower final voltage, and its behavior. Report those three digit numbers and behavior.

Setup computer to execute as much software as possible. IOW it should be outputting sound loudly, while searching the disk, while playing complex graphics (ie a move or game), while powering a USB device, while accessing the internet, etc. Having it access many peripherals simultaneously is important. If it cannot power up, then monitor any one red (pin 4,21-23), orange (pin 1,2,12 or 13), and yellow (pin 10 or 11) wire for what each does as and after its power button is pressed.

Report all three digit numbers from those six wires. Next reply will identify or exonerate suspects.

BTW, if wires are not colored, then a PSU may not be ATX Standard. See www.smpspowersupply.com for color and pinouts.
 
I have competed the measurements, here are the results:

Power on, no load:
9: 5.08V 16: 97.5mV 8: 4.66V 23: 5.07V 13: 3.34V 11: 12.2V

Power on, system under load. Running Witcher 3, watching YouTube video, playing audio CD in Windows Media Player, copy files from USB drive to local disk, copy files from NAS drive to local disk, running Prime95. Task manager reports CPU utilization 99%, memory 54%, disk 50%, network 90%:
9: 5.04V 16: 106mV 8: 4.65V 23: 5.04V 13: 3.31V 11: 12.13V

Power off, before normal startup:
9: 5.09V 16: 4.62V 8: 19mV 23: 20mV 13: 1.46V 11: 0.969V

During normal power on:
9: drops from 5.09V to 5.07V 16: drops from 4.62V to 99mV 8: rises from 19mV to 4.66V 23: rises from 25mV to 5.07V 13: rises from 1.53V to 3.34V 11: rises from 1.04V to 12.2V

During failed power on:
9: drops from 5.09V to 5.08V 16: drops from 4.62V to 97mV 8: rises from 19mV to 145mV 23: rises from 25mV to 5.08V 13: rises from 1.46V to 3.34V 11: rises from .978V to 12.2V

Obviously there is something going on with pin 8 during the failed power on. Your theory about the CPU not being permitted to operate would seem correct, as during a failed power on the CPU fans do not spin.

I should also note that I was only able to reproduce the problem by unplugging the power cable after the computer had been booted into Windows for some time, then plugging it back in. If Windows had only been booted for a short time, or if I pulled the power cable during boot, I could not reproduce the issue this time around.
 
Last edited:
Those numbers are not taken as instructions requested. For example, a purple wire (+5VSB) is measured only with power off and power cord is connected.

Green wire is measured before, monitored when, and measured after the power button is pressed.

Gray wire is measured by the same before, during, and after that power button is pressed.

Purple wire is from a second (standby) power supply inside the PSU. It only powers a power controller and other functions that might create a wake-up. That voltage is always on and is stable. That voltage is why its power cord must be disconnected from a wall receptacle before making any changes,

Green wire goes to near zero to order a PSU power on. It should have dropped immediately in response to the front panel button press. Power controller is always working properly - immediately and always ordering the PSU on.

Gray wire is a PSU voltage monitor for key voltages. How that wire acts during power on is critical. It should rise within a second or two. When it does rise, does it do so quickly? Clearly, when voltage does not rise (no Power Good signal), a power controller will not let the CPU execute. You measured that. Question is why a voltage monitor (inside the PSU) does not say "Power is Good".

Hopefully voltage on each yellow, red, and orange wire are rising sufficiently fast. (confirm this for every one yellow wires). Reported voltage are well inside the good region. Question is why its voltage monitor sometimes does not see a good voltage.

I am troubled by a 1.5 volts on the 3.3 volt lines when power is suppose to be off. I am wondering if the 5VSB (purple wire) is leaking into circuits powered by 3.3 volts (orange wire). Try removing parts peripherals (ie NIC, GPU, Ram, etc) only when a power cord is disconnected from the wall. Then reconnect power to check for 'well less than 1 volt' on that orange wire.

This leakage may not cause an above voltage monitor problem. But it can eventually lead to another future failure.

Chances are a defect is inside the power supply. Your numbers does not report any bad voltages So I do not know why that monitor sometimes reports bad voltages.

Try this. With the power supply disconnected from motherboard and GPU, then use a paper clip to jumper the green wire to any black wire. Monitor a gray wire (Power Good) for a greater than 2.6 volts everytime. That would imply some voltage is not rising fast enough during power on. Or maybe a 3.3 volt leakage is causing problems.

I see two problems. It is not yet apparent if both problems are same.
 
westom said:
Those numbers are not taken as instructions requested. For example, a purple wire (+5VSB) is measured only with power off and power cord is connected.
Click to expand...

See power off measurements. My PSU does not use color coded wires, but purple would correspond to pin 9. With the power off and the cord connected, it measured 5.09V.

Green wire is measured before, monitored when, and measured after the power button is pressed.
Click to expand...

See measurements during power up. Pin 16 dropped from 4.62V to 99mV during a normal power up, and did virtually the same thing during failed power up. Voltages stabilized almost instantaneously on all pins, so there is little else to note other than before and after measurements.

Gray wire is measured by the same before, during, and after that power button is pressed.
Click to expand...

See measurements during power up. During a normal power on it rises from 19mV to 4.66V, and during a failed power on it only rises from 19mV to 145mV.

Purple wire is from a second (standby) power supply inside the PSU. It only powers a power controller and other functions that might create a wake-up. That voltage is always on and is stable. That voltage is why its power cord must be disconnected from a wall receptacle before making any changes,

Green wire goes to near zero to order a PSU power on. It should have dropped immediately in response to the front panel button press. Power controller is always working properly - immediately and always ordering the PSU on.

Gray wire is a PSU voltage monitor for key voltages. How that wire acts during power on is critical. It should rise within a second or two. When it does rise, does it do so quickly? Clearly, when voltage does not rise (no Power Good signal), a power controller will not let the CPU execute. You measured that. Question is why a voltage monitor (inside the PSU) does not say "Power is Good".
Click to expand...

Yes, it does rise quickly during a normal power on.

Hopefully voltage on each yellow, red, and orange wire are rising sufficiently fast. (confirm this for every one yellow wires). Reported voltage are well inside the good region. Question is why its voltage monitor sometimes does not see a good voltage.

I am troubled by a 1.5 volts on the 3.3 volt lines when power is suppose to be off. I am wondering if the 5VSB (purple wire) is leaking into circuits powered by 3.3 volts (orange wire). Try removing parts peripherals (ie NIC, GPU, Ram, etc) only when a power cord is disconnected from the wall. Then reconnect power to check for 'well less than 1 volt' on that orange wire.

This leakage may not cause an above voltage monitor problem. But it can eventually lead to another future failure.

Chances are a defect is inside the power supply. Your numbers does not report any bad voltages So I do not know why that monitor sometimes reports bad voltages.

Try this. With the power supply disconnected from motherboard and GPU, then use a paper clip to jumper the green wire to any black wire. Monitor a gray wire (Power Good) for a greater than 2.6 volts everytime. That would imply some voltage is not rising fast enough during power on. Or maybe a 3.3 volt leakage is causing problems.

I see two problems. It is not yet apparent if both problems are same.
Click to expand...

Thanks again for the help. I should have some time tomorrow to try the further tests you recommended.
 
This has actually been a long standing issue with computers for a long time. I've had several systems afflicted with this problem over the years.

The power supply has large filter caps and at power-up there is a big inrush of current to charge them but it takes time. The PSU voltages may not stabilize in time for power good to go high (logic high vs. low). The power good signal stuck at a low state will not allow the mobo to POST.

Powering off and back on helps since the caps now have a partial charge and will fully charge in time to assert power good and allow the mobo to POST.

The problem is also related to the load on the PSU at power-up and there may also be timing issues on power good between certain samples of PSUs and mobos.

I suggested to one mobo manufacturer years ago to use a circuit powered by the 5vSB (5v Stand By) to essentially do an automatic re-power after the initial power-up. The circuit would run a timer based on presence of 5vSB (5vSB is there as long as the PSU is plugged in). When it first powers up, the timer count is close to zero so it tells the ATX PSU to shut down and then powers it up again. It can do that since the circuit still has 5vSB to signal the PSU to power up again. The second power-up, the circuit sees the timer is now a higher count and does not trigger another re-power. In reality, I would think it's not really a timer per se, but a signal that takes a certain amount of time for 5vSB to be present before it goes high (charges it's own cap).

I have a couple of old ASUS P5Q mobos, and they do exactly that. If I disconnect their PSU from the wall, and then plug it back in and power up, it will do a short 2 second power-up, shut down, and then 2 seconds later it powers up and stays up (and POSTs correctly).

ASUS was not the manufacturer I suggested that to, I'm sure they figured it out on their own. :)

You have the choice to leave it like it is and just deal with it (it won't hurt anything). Or you could try a different power supply and/or mobo.

I think what I would do is get a cheap UPS battery backup. That would fix this problem and also protect the equipment from power drops & surges.

Edit: I said cheap UPS battery backup, but I would spend a few extra bucks and get a true sine-wave UPS. The cheaper stepped wave units may cause trouble with some computer PSUs.
I'm waiting for a Cyberpower PFC model to go on sale to get one for my server.
 
Last edited:
Spartacus said:
This has actually been a long standing issue with computers for a long time. I've had several systems afflicted with this problem over the years.

Powering off and back on helps since the caps now have a partial charge and will fully charge in time to assert power good and allow the mobo to POST.
Click to expand...

Thanks for the info and explanation.

Your description sounds pretty similar to what I have been experiencing, however many times when I try to power off and then back on, the problem persists. I have to turn the computer on, turn it off, then wait five minutes before turning it on again for it to POST normally. To me that demonstrates that there may be a more significant flaw in the PSU. Oddly this hasn't been the case over the last week, but was for several weeks prior.

I may try to contact Silverstone to see if they have any insight.
 
Epos7 said:
To me that demonstrates that there may be a more significant flaw in the PSU. Oddly this hasn't been the case over the last week, but was for several weeks prior.
Click to expand...
You have clearly seen what the problem is. Capacitors that take too long to charge inside a power supply do not affect the Power Good signal. Since that signal stays and is suppose to stay in not Power Good until those capacitors charge. This is normal. Larger capacitors that take even longer to charge would not create your issue. Your issue is that the Power Good circuits claim power stays bad even minutes after capacitors have fully charged.

Your motherboard is not letting the CPU work - as designed and intended - because the Power Good signal says some voltage is defective - constantly. But your meter says voltages are good.

Of course, when Power Good says voltages are bad, then check every yellow (12 volts) wire to confirm all have good voltages.

But slowly charging capacitors are not problematic. The reason Power Good (gray wire) can take as much as a second to rise above 2.6 volts is because all power supplies take some time to create good voltages. Your case is unique. When those voltages finally are good, the Power Good circuitry sometimes says some voltage is bad.

Stick to facts as made obvious by measured numbers.. Capacitors taking too long to charge do not cause Power Good to say some voltage is always bad. Capacitors taking too long to charge means Power Good takes longer to say all voltages are good. Your meter reading says voltages are constantly good but Power Good says a voltage is constantly bad.
 
westom said:
Your motherboard is not letting the CPU work - as designed and intended - because the Power Good signal says some voltage is defective - constantly. But your meter says voltages are good.
Click to expand...

Correct, but only in some instances where the computer has been disconnected from power. Therefore, something about being disconnected from power is sometimes causing the PWR_OK signal not to rise.

I should have some time today or tomorrow to try the further tests you recommended.
 
Lol..... everybody has an opinion I guess.

I think my college professor who designed switching power supplies would agree with the charging caps (failure to charge properly) being a very real issue in regards in power_good signal behavior. How could it not be? lol

Another issue is that some makers of low-end PSUs may simply tie power_good high which defeats the whole point of having that signal. In that case, power_good is high immediately so the CPU will start up before PSU voltages have stabilized. That may work most of the time, but sometimes it won't (resulting in no POST or errors).

Note that there are two sides of this issue.... the PSU must be operating correctly and correctly assert power_good. Likewise, the mobo must properly detect power_good and drop cpu_reset to allow POST. Minor variances and design problems on both sides happen. Sometimes a certain PSU will not work with a certain mobo. * I've seen that happen. *

Have a look at the ATX 12V Power Supply Design Guide for more info.

http://www.formfactors.org/developer/specs/atx12v_psdg_2_2_public_br2.pdf

One last thing.... a digital multimeter is of limited use in looking at some of these more complex problems. They are good for looking at nominal operating voltages, but are really of no use when you need to see signal timing, ripple, etc. You'd need a good oscilloscope for that. Giving somebody the task of a long list of measurements to take to troubleshoot this problem is kind of silly.

It comes down to trying a different PSU. And if that doesn't help, then try a different mobo.
 
Last edited:
Spartacus said:
One last thing.... a digital multimeter is of limited use in looking at some of these more complex problems. They are good for looking at nominal operating voltages, but are really of no use when you need to see signal timing, ripple, etc.
Click to expand...

One who did this stuff for 40 years and who designed power supplies says a meter can identify ripple voltage and other problems. To know why (and what numbers are relevant) requires additional knowledge.

Also known - big capacitors discharge quickly in seconds. This function, required in supply designs, is essential for human safety. Capacitors might be responsible for the OP's strange problem. But not big electrolytic capacitors.

Of concern is the 1.5 volts leaking into the 3.3 volt circuits. Obviously this can cause metastable states.

When voltages are fine, the power monitor says one voltage is defective. Do not get sidetracked by other speculation. That fact explains a CPU not permitted to execute the first time. That behavior is the source of all symptoms. No opinions apply. That fact clearly exists.
 
Ok, please educate me.....

How do you see ripple on a DVOM?

Knowing of course that looking at AC volts on a meter is not accurate for ripple.
 
westom said:
Please go elsewhere with your egotistical and irrelevant comments.
Click to expand...

Lol..... just as I thought, no real answers. :D

OP, sorry this turned into a pissing match.
It seemed to me like you were being sent on a wild goose chase though.

Try another high quality power supply. Maybe from a local supplier that will let you
return it if it doesn't fix the problem. Or borrow one from another computer if you can.
 
I appreciate all the help from everyone. Sometimes a lively debate is what it takes to get to the bottom of an issue :p

It does seem like the problem is the power supply, and while additional tests may tell me what exactly is wrong with the power supply, the easiest course of action is probably to just replace it. I've contacted Silverstone and started the RMA process. Since it's an SFX PSU, I doubt I could find one locally to try. I'll probably end up waiting a couple weeks to get the RMA back, but hopefully that will solve my issue.
 
Epos7 said:
It does seem like the problem is the power supply, and while additional tests may tell me what exactly is wrong with the power supply, the easiest course of action is probably to just replace it.
Click to expand...
Just remember what has already been determined. If the PSU is a problem, then two problems exist: power supply and potentially harmful leakage into 3.3 volts.

If reason for failure is only that leakage, then a PSU supply intermittently misbehaves because of that leakage. Either way, that leakage problem still remains.

Best is to check for 1.5 volts leaking into an orange wire (3.3 volts) when a new PSU is powered off. If that leakage remains, then a problem still exists. Then future strange failures may occur.

No lively debate existed. Another only attacked and denied rather than learn how a meter can measure more than simple stuff. Despite denials, the meter clearly said PSU voltages were always good.

We know this. A PSU monitor declared good voltages sometimes as bad and sometimes as good; when voltages were always good. Only a few minutes of labor would have defined that as two failures (PSU failure and ...) or one failure (a leakage elsewhere).
 
Thanks, when I receive the replacement power supply I will check pin 13 while powered off to see if the voltage leak disappears.

Removing components one-by-one as you suggested would have taken several hours being that this is an ITX build and it seemed wise to avoid that given that the PSU is almost surely at fault.

I have submitted the RMA form to Silverstone, hopefully the process is quick.
 
westom said:
Try this. With the power supply disconnected from motherboard and GPU, then use a paper clip to jumper the green wire to any black wire. Monitor a gray wire (Power Good) for a greater than 2.6 volts everytime. That would imply some voltage is not rising fast enough during power on. Or maybe a 3.3 volt leakage is causing problems.
Click to expand...

I took the power supply out of the computer to send back to Silverstone, but before I packaged it up I tried what you suggested above.

With nothing connected to the power supply, and with pins 15 and 16 shorted, I measured the voltage on all other pins with the power supply on and off.

With the power supply on I found 0V on all pins except 5.09V on pin 4 and 3.74V on pin 21.

With the power supply off I found 0V on all pins except 5.09V on pin 4 and 3.74V on pin 21. What I did notice is that several seconds after the power supply is switched off, the voltages on pins 4 and 21 begin to drop. They drop slowly for 5-10 seconds, until pin 21 hits about 1.9V, then instantly return to 3.74V and 5.09V respectively.
 
Epos7 said:
I took the power supply out of the computer to send back to Silverstone, but before I packaged it up I tried what you suggested above.

With nothing connected to the power supply, and with pins 15 and 16 shorted, I measured the voltage on all other pins with the power supply on and off.

With the power supply on I found 0V on all pins except 5.09V on pin 4 and 3.74V on pin 21.

With the power supply off I found 0V on all pins except 5.09V on pin 4 and 3.74V on pin 21. What I did notice is that several seconds after the power supply is switched off, the voltages on pins 4 and 21 begin to drop. They drop slowly for 5-10 seconds, until pin 21 hits about 1.9V, then instantly return to 3.74V and 5.09V respectively.
Click to expand...

Confusion is created by using pin #s rather than wire colors. Apparently you have confused pin 21 (red wire) with pin 9 (purple wire). Measuring pin 9 at 3.74 is a defect. Original instructions called for measuring that purple wire when power was not on. If I remember, you did not do that. So that defect was not reported.

This experiment shows why power on might be unstable when power is first applied. A power controller was starved for voltage at 3.74. Adjacent capacitors had enough residual charge to start normally a second time. That would be a defective Standby power supply inside the PSU.

Pin 16 (reported as pin 4) is a Green wire that also gets 5 volts from the same Standby supply. So the Standby supply provides sufficient voltage. But for some (internal) reason, that same 5 volts appears lower (defective) on pin 9 (purple wire)

First measurement (called for in the original instructions) on a purple wire (with nothing yet powered on) explains why power on is unstable a first time; works fine the second. No reason to suspect any other parts since earlier confusion (using pin #s rather than wire colors) probably explains an originally suspected and now unlikely motherboard leakage.
 
westom said:
Confusion is created by using pin #s rather than wire colors.
Click to expand...

As I mentioned before, the wires are not color coded on this PSU. All the wires are black, therefore pin numbers are the only way to identify each wire. Here is the pinout I used:

http://www.marnscda.com/Guide/ATX_PR1.jpg

All measurements were taken at the motherboard side of the cable.

Apparently you have confused pin 21 (red wire) with pin 9 (purple wire). Measuring pin 9 at 3.74 is a defect.
Click to expand...

Pin 9 is PWR_OK and measured 0V with power supply on and with power supply off, as stated. Pin 21 is +5V and measured 3.74V.

Original instructions called for measuring that purple wire when power was not on. If I remember, you did not do that. So that defect was not reported.
Click to expand...

Pin 9 (5VSB, typically color coded purple in power supplies with color coded wires) measured 5.09V when power is off, as stated in original measurements with power supply installed in computer. In new measurements, with power supply removed from computer and pins 15 (ground) and 16 (PS_ON) jumpered as instructed, pin 9 measured 0V with power supply on and with power supply off.

This experiment shows why power on might be unstable when power is first applied. A power controller was starved for voltage at 3.74. Adjacent capacitors had enough residual charge to start normally a second time. That would be a defective Standby power supply inside the PSU.

Pin 16 (reported as pin 4) is a Green wire that also gets 5 volts from the same Standby supply. So the Standby supply provides sufficient voltage. But for some (internal) reason, that same 5 volts appears lower (defective) on pin 9 (purple wire)
Click to expand...

Pin 16 (PS_ON, typically the green wire in power supplies with color coded wires) was jumpered to pin 15 (ground) as instructed. Pin 4 is +5V and measured 5.09V as stated in my previous post. Pin 9 (5VSB, typically the purple wire in power supplies with color coded wires) measured 0V, as stated in my previous post.

First measurement (called for in the original instructions) on a purple wire (with nothing yet powered on) explains why power on is unstable a first time; works fine the second. No reason to suspect any other parts since earlier confusion (using pin #s rather than wire colors) probably explains an originally suspected and now unlikely motherboard leakage.
Click to expand...
 
Last edited:
Epos7 said:
Pin 9 (5VSB, typically the purple wire in power supplies with color coded wires) measured 0V, as stated in my previous post.
Click to expand...
If that wire measures zero volts, then power could not turn on. However if that wire is actually the adjacent (other side) red wire, then it measures 5 volts only when powered on.

By swapping your pin # 4 with 16 and pin # 9 with 21, then everything makes sense. +5VSB pin 9 is outputting insufficient voltage that explains that 'does not start the first time but starts the second time' behavior.

A new supply should solve this problem. By swapping pins, no other defect is identified.
 
westom said:
If that wire measures zero volts, then power could not turn on. However if that wire is actually the adjacent (other side) red wire, then it measures 5 volts only when powered on.

By swapping your pin # 4 with 16 and pin # 9 with 21, then everything makes sense. +5VSB pin 9 is outputting insufficient voltage that explains that 'does not start the first time but starts the second time' behavior.

A new supply should solve this problem. By swapping pins, no other defect is identified.
Click to expand...

Ah yes you are correct. I forgot to account for the different orientation when taking the second set of measurements. That means I shorted pins 21 & 22 rather than 15 & 16.
 
Epos7 said:
Ah yes you are correct. I forgot to account for the different orientation when taking the second set of measurements. That means I shorted pins 21 & 22 rather than 15 & 16.
Click to expand...
Again, this is why/how we learn - by making mistakes.

Or as Tommy Smothers once said, "I'll never make that mistake again."
 
The repaired PSU arrived yesterday. As far as I can tell, the voltages measure OK.

If I jumper pins 15 and 16, then remove the jumper, the 12V rails taper off slowly such that they're still sitting at about 0.3V after a few minutes. I assume this is normal.

Installed PSU in computer and everything is running well.
 
Epos7 said:
If I jumper pins 15 and 16, then remove the jumper, the 12V rails taper off slowly such that they're still sitting at about 0.3V after a few minutes. I assume this is normal.
Click to expand...
PSUs for electronics can continue outputting power after AC power is lost. For example, essential when a UPS spends significant time with neither AC power nor battery power as that UPS switches from one to another. This 'time' decreases as the load increases.

That slowly decreasing power is also why a voltage monitor exists. Low voltage causes no hardware damage. But a computer may act 'demented' as voltages drops slightly below minimal values. A power controller may halt the CPU when voltages are high enough for some computer operations but are insufficient. When voltages slowly drop as you have observed.
 
You must log in or register to reply here.
Back
Top