I'm servicing a Gateway GM5260. It uses an Intel DP965LV motherboard and it has Windows XP installed. It also has this habit of freezing randomly. Then you are forced to hold in the power button to shut it off. Here is the interesting part. When you power it back on after it freezes, it will ALWAYS shut down about 1 second after the power button is pressed. If you restart it second time, usually it will power up fine. Occasionally, it will take a third try to keep the power on. If the computer goes through a normal shut down, the power up process always works the first time. I've swapped the RAM, video card, and power supply, and it made no difference. The only hardware issue I can find is that in the system bios, there is a value in the hardware monitor called, +1.5V and it hovers around 1.298V. I have inspected the caps and they don't look or test bad. The CPU and motherboard temps are normal. Has anyone experienced this type of problem before?
Your description sounds exactly like the usual bulging capacitor problem on the motherboard or in the power supply. You said that you checked (tested?) the motherboard caps, but did you check the power supply caps? The hardware monitor is more commonly called the "power good" line on an ATX power supply. It's suppose to be near +5V so your +1.5v indicates that the power supply is not happy.
Incidentally, what's happening is that when cold, the ESR (equivalent series resistance) of the electrolytic capacitors is rather high, resulting in lots of ripple, lousy regulation, bad breath, circuit malfunction, and other undesirable things. However, as the capacitor gets warmed up by the ripple current or nearby hot components, the ESR goes down allowing the sick capacitor to act like a good capacitor. The computah will probably work reasonably well, as long as it stays warm, but will revert to simulating a stubborn mule when it cools down.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Try watching what the ATX power good gray wire does on a working computah. There's small delay between turning on the power and when the line goes high. Normal operation is a logic high which means roughly +5VDC. +1.5V is definitely out of spec. I would venture a guess that it's really oscillation or pulsing on the power good line, which happens to average out to about +1.5V on a DVM.
Having the measured +1.5v be similar to the CPU voltage is coincidence. Recycling the above Wikipedia page: The ATX specification defines the Power-Good signal as a +5 volt (V) signal generated in the power supply when it has passed its internal self-tests and the outputs have stabilized. This normally takes between 0.1 and 0.5 seconds after the power supply is switched on. The signal is then sent to the motherboard, where it is received by the processor timer chip that controls the reset line to the processor.
Cheaper and/or lower quality power supplies do not follow the ATX specification of a separate monitoring circuit; they instead wire the power good output to one of the 5 V lines. This means the processor will never reset given bad power unless the 5 V line drops low enough to turn off the trigger, which could be too low for proper operation.
I guess the easiest test is to just replace the PS with a known good PS and see what happens. Checking the PS with a PS tester is also useful. I have a really old and crude one but am considering getting something like one of these:
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Just to make sure we're talking about the same thing, I said: "The only hardware issue I can find is that in the system bios, there is a value in the hardware monitor called, +1.5V and it hovers around 1.298V." From what I understand, it's supposed to be 1.5V, not 5V.
I did swap the power supply with a working unit and the same thing happened.
I used an ESR tester to test the caps on the motherboard and power supply. They were fine.
Finally, I noticed that during an anti-virus update, the computer froze up again. I uninstalled the anti-virus software (Avast) and it hasn't locked up since. A few days later I installed AVG, and it's still running fine. I also disabled the EIST service in the bios but I'm not sure if that mattered at all.
formatting link
What I never did figure out was why it took two power button pushes to restart the computer after a freeze up. It reminds me of what happens when you first apply AC to a computer (before even hitting the power button) the power will briefly come on and then shut off.
That would be the +1.5v DDR3 RAM power supply voltage: It should be very close to +1.5v.
I was referring to the voltage on the ATX power connector pin 8: as measured with a voltmeter.
Oops. So much for these theories.
I'm having a difficult time believing that any program can produce a power related boot failure BEFORE the OS is loaded. You mentioned: "When you power it back on after it freezes, it will ALWAYS shut down about 1 second after the power button is pressed. If you restart it second time, usually it will power up fine. Occasionally, it will take a third try to keep the power on." That's all before the OS or auntie-virus is loaded. About the only think I can conjure that might be involved is when the BIOS loads the microcode updates into the CPU. Weird.
Will you please find a scope or voltmeter, attach it to Pin 8 (power good) line of the ATX connector and see if there are any long delays, oscillations, or complaints on this line that might inspire the motherboard to shut down? Everything I read hear (except the auntie-virus stuff) points to a power related problem. If the +1.5v line is low, it might be defective RAM sucking excessive current from the Vdd power line. Run a RAM test and see if you get it to fail. (better)
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
ESR meters will not necessarily give a valid reading. SMPS need really good caps so an ESR reading of 'good' may not, in fact, be good enough. I was w orking on a PC power supply today only because it is used in a very special computer and the manufacture is phasing it out but our people intend to us e them a few more years. One of the caps I pulled read nothing on the meter but when I probed the board where it was, the reading was 'good'.
Consider solid polymers if available. Otherwise I go for the highest ripple current and longest hours rating. THEN price comes in to the picture. My t ime is way more expensive than any capacitor I put in.
'Lytics on a motherboard over 5 years old are suspect and probably need to go.
Most notably the fact that many of the caps are in parallel and the ESR readings should be much lower than out of circuit.
However, I see this present situation as perhaps two different problems which are:
Is the freezing problem really only software related?
Is the power up glitch ONLY after a freeze up a flaw or a feature? Even when cold in the morning, it starts up fine if the night before it powered down normally.
I'll go ahead and check out Jeff's suggestion:
"Will you please find a scope or voltmeter, attach it to Pin 8 (power good) line of the ATX connector and see if there are any long delays, oscillations, or complaints on this line that might inspire the motherboard to shut down? Everything I read hear (except the auntie-virus stuff) points to a power related problem. If the +1.5v line is low, it might be defective RAM sucking excessive current from the Vdd power line. Run a RAM test and see if you get it to fail."
Message to Jeff: What exactly should I see on my scope at Pin 8 and am I monitoring this constantly or only when I power up the machine? Also note that I did swap all four RAM sticks with new ones and the same thing happened (before I changed AV programs).
The PS_OK line should start at 0v. When you turn on the power, it should stay at zero for between 100 msec to 500 msec, when it should suddenly rise to something near +5v. The length of time it takes to rise, after the power is applied, is an indication of potential problems. I'm not certain what is considered normal or typical but I seem to recall about 300 msec. I trigger on the +5V to 0V PS_ON line used to turn on the PS.
The fun starts when it's 0 msec, which indicates a cheap junk power supply with no power good circuit. The line is simply tied to the +5V power supply output line. That will create a boot failure or startup hang because the CPU wants to see stable power supply voltages.
Longer times and strange oscillation are also problems. A PS with bulging caps on the +5V line (quite common) will show either long delays or 100KHz hash for a few msec until the caps warm up. I've also seen the line go high for a while, then drop, and go high again, which caused some odd startup failures. However, I didn't investigate the cause and just recycled the PS.
Table 13. PWR_OK Signal Characteristics Signal Type +5 V TTL compatible Logic level low < 0.4 V Logic level high 2.4 V and 5 V High-state output impedance 1 k? from output to common PWR_OK delay 100 ms < T3 < 500 ms PWR_OK risetime T4 = 10 ms AC loss to PWR_OK hold-up time T5 = 16 ms Power-down warning T6 = 1 ms There's more detail in Sect 3.3 (Pg 23).
Also, there's something else worth trying. The ATX power supply keeps power applied to some parts of the motherboard when the computah is turned off. Mostly, that's the ethernet and wireless cards which use WoL (wake on LAN) to turn the computer on remotely. You note that holding the power button down to shut off the computer causes the computer to shut down about 1 second after turning it back on. Try:
Hold the button down, wait, and then turn it back on.
Hold the button down, pull the power plug, wait about 20 seconds for everything to discharge, plug it back in, and turn it back on. If magically pulling the plug stops the shutdown on startup oddity, then methinks something might be odd with the WoL circuitry on the motherboard.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
I know from experience with this computer, that unplugging it and then plugging it back in again after a freeze up has no effect on the outcome. It will be interesting to see what happens if it is powered off without a freeze up. I would guess that it will still take two power up sequences to keep it on.
How am I supposed to see the 10ms pulse rise time with an old Philips PM
3217, nonstorage-type scope and then also determine the delay time? (By the way, I wouldn't mind having a newer scope if you have any suggestions.)
10 msec rise time requires only: 0.35 / 10 msec = 35 Hz bandwidth Unless your Philips PM3217 has a problem, you should be able to see that on any oscilloscope.
I have a fairly large collection of older oscilloscopes. Several are promised as bribes to various people. I'll throw together a list and email it to you. Shipping and packing is the major expense. The to be repaired pile: There are about twice as many scattered around the house and office.
My concern is that since it's a transient pulse, that it will not be on the screen long enough for me to see it unless you're saying that it might oscillate continuously.
If you trigger on the leading edge rising voltage and setup your scope for single trace, it will probably be visible. However a DSO (digital storage oscilloscope) will do better at capturing the pulse.
However, the rising edge is not as interesting as the time delay between when you apply power, and when the PWR_OK line goes high. At
50 or 100 msec/div, you should have no problems seeing if there are any oddities or excessive delays.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
I checked out the delay pulse time on pin 8. It seemed normal although it was hard to tell because a tenth of a second goes by rather quickly.
I did a test power off by pressing the power button while in the BIOS menu and it shut down immediately. There was not even the slightest delay compared to what happens when Windows freezes up. The next power up happened on the first try. I did the same thing when booting to the command prompt. Instant off on the power button press and then a restart on the first try. I wonder if the motherboard is keeping track when you get a bad shutdown and intentionally makes you press the power button twice.
0.1sec is a long time with an oscilloscope. Beg, buy, or borrow one. It helps. However, the power good line seems to be working if you didn't see any inordinately long delays. That leaves noise, which you're not going to see without an oscilloscope. As I previously suggested, use a sound card PC software oscilloscope. (No, you cannot run the PC scope software on the machine that you're trying to test).
I don't know about hitting the button twice, but many BIOS's will do a rather fast boot if you reboot, but take forever to test everything if you reboot after a failed boot. I guess(tm) the previous boot status is stored somewhere as long as AC power is applied. It's NOT in the power supply.
Incidentally, I recently fixed a Dell Optiplex 760 SFF machine that required pressing the power on/off button twice in order to get it to do anything. I replaced the power supply for other reasons (bulging caps) but that wasn't the cause. After some fumbling, it turned out to be bad contacts in the switch. However, that doesn't sound like anything that might produce the symptoms you're seeing.
When in doubt, replace everything.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Was I supposed to use one channel of the scope to monitor the power button line and then monitor the PS_ON line with the other channel, and then see the delay time between the two?
My customer picked up the unit quite a while ago and has had no complaints... so far.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.