[Novalug] final resolution: computer turned itself off without stored message

Jay Hart jhart@kevla.org
Tue Jan 29 20:12:35 EST 2008


I have had extremely great luck with PC Power & Cooling brand Power Supplies.
I wouldn't buy anything else.

http://www.pcpower.com/index.html

Jay


Note: I didn't know they are now part of the OCZ tech group.  I'd have to
research and see if their stuff is still as great as before.


> List,
>
> with thanks to all who tried to assist me with this problem, from
> actual help to helpful hints, this is what I think happened.
>
> When the power supply got too hot, the machine went into a 'power save'
> mode that you saw only if you happened to look at the screen while
> the computer did its thing. The reason for overheating was probably
> the power supply fan. It now has too much friction, so the motor could
> no longer turn it. It is not even possible to spin it up by sucking
> air through it with the shop vacuum. On dismantling the power supply,
> I also noticed a capacitor with a weird brown spot on it, so for all
> I know the power supply was about to fail due to the excessive heat.
>
> In the process of looking for the problem, I found dust all inside
> the machine, including the fans on the CPU, in its heat sink,
> and on the fan to the case (which sucks the air out of the case,
> in parallel to the power supply fan). So, I started worrying about the
> motherboard or the CPU overheating.
>
> On the suggestion of Jay Hart I installed various programs designed
> to keep track of what the hardware is doing. One is ls-sensors. This
> gives temperatures, voltages, and the speed of some fans. Another
> is gkrellm, which shows these data in a GUI. Since I'm more of a
> text window person than a GUIer I prefer the text-based sensors.
>
> After some diddling to turn on the various things to look at
> (which went really fast thanks to Jay's hand-holding), I can now
> execute 'sensors'. This gives the following:
>
> pereira@lithium:~$ sensors
> it8712-isa-0290
> Adapter: ISA adapter
> VCore 1:   +1.68 V  (min =  +4.08 V, max =  +4.08 V)   ALARM
> VCore 2:   +0.00 V  (min =  +4.08 V, max =  +4.08 V)   ALARM
> +3.3V:     +3.31 V  (min =  +4.08 V, max =  +4.08 V)   ALARM
> +5V:       +4.92 V  (min =  +6.85 V, max =  +6.85 V)   ALARM
> +12V:     +12.22 V  (min = +16.32 V, max = +16.32 V)   ALARM
> -12V:     -27.36 V  (min =  +3.93 V, max =  +3.93 V)   ALARM
> -5V:      -13.64 V  (min =  +4.03 V, max =  +4.03 V)   ALARM
> Stdby:     +5.00 V  (min =  +6.85 V, max =  +6.85 V)   ALARM
> VBat:      +4.08 V
> fan1:     3183 RPM  (min =    0 RPM, div = 2)
> fan2:        0 RPM  (min =    0 RPM, div = 2)
> fan3:        0 RPM  (min =    0 RPM, div = 2)
> M/B Temp:    +52 C  (low  =    -1 C, high =    -1 C)   sensor = thermistor
> CPU Temp:    +36 C  (low  =    -1 C, high =    -1 C)   sensor = thermistor
> Temp3:        -1 C  (low  =    -1 C, high =    -1 C)   sensor = disabled
>    ALARM
>
> The human-readable names of some of the signals had to be guessed
> at, and introduced by hand. For example, the motherboard has a VIA
> chip, and some of the modules that must be loaded in the kernel
> (which are found by executing 'sensors-define' or some similar command).
> You can touch the chip, and compare what you feel with the number.
> When the case was open this chip was maybe 42 C, now, with the case
> closed, it is substantially higher: apparently, in my case putting
> the side panel back onto the computer case messes up the air stream
> and the cooling. The chip is much cooler, 36 C. At first turn-on
> of the machine the temperatures are much lower (below 30 C).
>
> Fan1 is the one on the CPU, the fan on the case is either not
> accessible by 'sensors', or it's not turned on.
>
> I don't know what to do with the voltages. I always thought that you
> needed only +12 V and +5 V from the power supply. The MB makes the
> others. The more modern the CPU (mine is an AMD Athlon), the lower
> the voltage, so 1.68 V seems reasonable. The other chips apparently
> work on 3.3 V, which used to be a standard.
>
> I don't know yet what to do with the limits, and the ALARM, nor
> do I know how to make the 'sensors' put the temperature on a
> plot of temperature versus time, or in the logs (instead of
> every 10 minutes a 'MARK', which only tells you that the
> computer is still alive). All that needs more time to figure out.
>
> Another topic that came up on the list was the efficiency of
> computer power supplies. Instead of the one that came with the
> case (an ENLight EN8304846), which gives as its input specs
> 115 V, 6 A (and therefore either 690 W or 490 W, depending
> on the current being a peak or average value). The output is
> claimed to be 180 W for +5V, 160 W for +12 V, and another 20 W
> for three other supply values given (-5 V, -12 V, and another +5V).
> This is at most (they say) 300 W. So, the efficiency would be 61 %
> at most. The new power supply claims an efficiency of 70 %.
>  From the web I saw that well-designed switching power
> supplies (which these are) can be up to 95 % efficient.
>
> Next time I'll need a power supply, I'll buy an efficient
> design, not the one that happens to catch my eye.
>
> Nino
> _______________________________________________
> Novalug mailing list
> Novalug@calypso.tux.org
> http://calypso.tux.org/cgi-bin/mailman/listinfo/novalug
>





More information about the Novalug mailing list