Harddisk clicks -> System freezes/reboots
Robin.Laing at drdc-rddc.gc.ca
Tue Jan 23 17:24:15 UTC 2007
> Ric Moore:
>>>> Tim, back to ohm's law, wouldn't low voltage also make it run hot???
> Tim [answering about Ohm's law]:
>>> No. Voltage and current are inter-related. If a power supply is losing
>>> the ability to "supply," the current will go down. Correspondingly, the
>>> voltage will reduce at the same time (due to lack of supply).
>> In switching mode supplies (which is all there is on PC's these days),
>> the problem with low voltage is that the switching transistors tend to
>> stay on longer. At some point the cores can saturate, and in that case
>> low voltage = high current. It is one of the big drawbacks to switch
>> mode supplies.
> I was under the impression that Ric was asking about low voltage
> affecting the hard drive rather than the power supply.
> Specific to Ohm's law, voltage going low would mean the current would,
> too. But the other situation you refer to (reluctance, oscillator
> circuit design, etc.) is something else (not to do with Ohm's law).
> While current consumption may go up within the power supply circuitry,
> it wouldn't be *supplying* more current to the rest of the computer, at
> the same time as supplying less voltage.
> Oscillator design were never my strong point, they were badly taught and
> my interests were elsewhere (usually things *NOT* oscillating). ;-) I
> don't recall doing much LC theory with them, it was usually RC. As a
> matter of fact, I've got to resolve just that point in some equipment,
> here. A video effects mixer that's oscillating somewhere that it
> shouldn't, so the pedestal clamp is going nuts.
> So regarding Ric's question, if he means the hard drive getting hotter
> if the current went down. Probably not. But then if some of the
> circuitry latches up (gets stuck) when not properly powered, maybe it
> could get hot. But I'd hazard a guess that you might have a component
> burn out, at the most, on the drive, but the rest of the drive wouldn't
> get hot. Much of the heat generated by a drive is the air resistance
> against the platters as they spin. With the servo that shuttles the
> head back and forth adding to it.
I think with the power supply being low, the drive head will swing back
and forth between park and seek, thus can heat up the drive. Of course
in todays newer drives, heat can be an issue without this.
>> How's that for a design challenge?
> Agreed that there are some challenges involved, but it's only a power
> supply. Switch mode power supplies are an old design idea. It
> shouldn't be that hard to make a good one for someone who's fluent with
> them (i.e. the manufacturers who make thousands of them, yearly). How
> many more years of computer systems do we need before we can get one
> with a decent power supply? ;-) There's certainly less complexity in
> them than in the rest of the computer system.
If you demand it, you will get a good power supply. Of course that good
power supply can fail, as it did in my case. The issue for me was that
the warranty required me to ship it a certain way to the USA at my cost
which would have been over 50% of the cost of a new power supply. Thus
the warranty was useless. I will find the bad component myself and
In my case, the +5 rail was about .5 volts low and the drive was
clicking. Change the power supply and all is well.
I found an online power supply calculator that you put in the number of
drives, fans, processor type and other accessories. It would calculate
the size of power supply needed. In my case, the original supply was
~50 watts to low. The first new supply had a 100W buffer.
More information about the users