Is it possible to build a silent computer, without fans at all? What case should I buy?

[Ken Maltby]

not true

Very true, you don't even see him claim any actual experience
with any specific equipment. All his statements are generalities,
phrased to imply a broad experience and understanding. Where
are the links to the spec sheets for these PSUs? Where does he
even try to claim he's done any of the testing/experimentation that
would support his "observations"?

You, wisely, asked him to describe his testing procedure, and
what did you get back?

He told you already, he tried them and they died.

Where in his posts does he actually say that he ever had his
hands on even one such PSU? And "he tried them"? Really, all
of them? A significant sample? Makes & Models, please. It
would be nice to check his claims against the published data.

if you were correct then what would you be looking forward to?

You correctly asked for Makes&Models, I'm sure Kony is
burning up Google right now trying to find some.

Sounds
messed up to me. Unless you really think you might be wrong - which
makes what you wrote earlier a bit outrageous.


jh

Look, you can fall for this guy's claim to PC omnipotence, all
you want, no skin off my nose. At least you knew to ask a few
questions.

Luck;
Ken

 

[jameshanley39]

Very true, you don't even see him claim any actual experience
with any specific equipment. All his statements are generalities,
phrased to imply a broad experience and understanding. Where
are the links to the spec sheets for these PSUs? Where does he
even try to claim he's done any of the testing/experimentation that
would support his "observations"?

You, wisely, asked him to describe his testing procedure, and
what did you get back?


Where in his posts does he actually say that he ever had his
hands on even one such PSU? And "he tried them"? Really, all
of them? A significant sample? Makes & Models, please. It
would be nice to check his claims against the published data.

But I can't expect Kony, as a matter of course, to give / rattle off,
all useful relevant information without me or anybody asking him for
it. It'd be great, but It's a standard that is rarely, if ever, met.

It's quite typical, with the usenet tree structure, that somebody says
something, and it's not complete, and those with the energy for it,
question it, and it goes on until more information comes out.

You can't have one standard for Kony and another standard for
everybody else. I notice that Kony didn't back up with specifics,
regarding what he thought was rubbish. But the fellow that had
positive experience, didn't back up his experience with specifics
either - to put it mildly. Presumably, one would have to ask him too,
but first i'm interested in which Kony has tried, that he says die.

You correctly asked for Makes&Models, I'm sure Kony is
burning up Google right now trying to find some.



Look, you can fall for this guy's claim to PC omnipotence, all
you want, no skin off my nose. At least you knew to ask a few
questions.

I wasn't planning on asking him so I could check the specs (though I
applaud those that do check them or ask so they can check them). I
was asking, so as to add to the collective experience of what's top
and what's snot. Things can be rubbish or duds for reasons not even
hinted at in a spec. And I don't know my way around specs or
electrical issues to get a hint like that from reading it !!

 

[Rod Speed]

Sounds like more results from "Kony's Imaginary Testing Lab", to me.

Me too.

He aint called kony for nuffin.

 

[kony]

Not necessarily, I have a pretty good Hunter ceiling fan.
The passive heatsink could be hotter than an active heatsink,
at a point of measurement, but there is nothing to say that a
massive passive heatsink with the proper design couldn't
cool the part better than your standard CPU heatsink and
fan.

You are drifting into unrealistic theory. In practice, you
cannot achieve the same cooling with a massive passive
heatsink (exception in next paragraph), because of the
density of the (CPU in this case) heat. It would require
far too thick a copper base to get the heat away from the
core at the required rate, it is not just the total amount
of heat the 'sink can shed in C/W but the spot-temp at the
core-mating surface.

The exception would be a means to remove the heat faster,
like a heat pipe. Even so, we weren't talking about
different 'sinks, rather the same with a fan or not... as it
is with any parts in your system (besides those with a
water-block) having their 'sink as the constant and the
variable being chasssis-induced airflow (or not).

Now you are just being totally ignorant, and it is obvious
you have never encountered water cooling of any kind.
(Or ever tried to solder a copper pipe that has water in
it. ) The moving water is very effective at carrying away
heat. Waterblocks don't get hot.

I wrote "heated" and you'd be silly to ignore this vital
requirement, since it's pretty manditory to transfer that
thermal energy to the water. Heated doesn't have to mean
"hot" on a relative scale of perception based on a body temp
or threshold of pain.

In fact one way I test
home made waterblocks is to press it onto a hot electric
burner, holding it in my hand. They stay cool even when
the rest of the burner is glowing cherry red.

If you have quite high flow rate, cold enough water, or the
top is poorly interfaced to the bottom it could be possible,
but it is hotter than without heat. I have no idea why you
think this distinction is important, you seem to be trying
hard to ignore the important variables.

But I'm not guessing, I have been water cooling PCs and other
devices for a number of years now. And my temperature
measurements are as proper and "scientific" as any.

This is nonsense. It is beyond ridiculous to think
measurement of a heatsink casually fastened across muliple
FETS with the typical poor interface used by a motherboard
manufacturer could be even close to usable data.

The facts are right in front of you, but you deliberately
refuse to collect them. Same with capacitors, you don't
actually measure and compare both ways, instead dodging and
weaving around the inescapable truth.

Your parts without water blocks are running hotter. We
can't predict (especially without the critical data) if that
is "too" hot for your required lifespan or not but to
pretend there is no temp difference is to want the novelty
of a passive system so badly you deliberately put on
blinders.

I've measured capacitors that were 20C hotter in a passive
system, but if I provide details of board model, cap brand,
etc, will it be applicable to your system enough that you
wouldn't dismiss it? Doubtful, you need accurate data for
your particular system... but so far, only lots of random
excuses instead of just doing the manditory things any sane
system integrator would do, to accurately measure the spot
temps before concluding an experimental passive cooling
system did as you guessed it does. Scientific method -
requires more than a guess and defense, there's data
collection involved too.

 

[kony]

what makes/models have you tried?

Every single one of them runs hot. The remaining question
is how long you need it to work.

I googled
thermaltake fanless power supply overheat

I didn't get horror stories. One guy said it worked fine so far(2
days).

You seem to be entirely missing the point. 2 days is
nothing, it'll have to work for reasonably close to it's
MTBF rating, or being generous a few years at it's rated
power.

This is exactly what we'd expect to have happen on any
overheating PSU, it'll work fine for awhile but the stress
kills it prematurely. It doesn't just go pop in an hour
(usually) at a normal system load, but even then we don't
have any description of the system it's placed in, for all
we know that guy who had his work for 2 days, might've had
it overheat if the case had zero fans and the system was
stress tested at full load the whole time.

 

[kony]

Very true, you don't even see him claim any actual experience
with any specific equipment. All his statements are generalities,
phrased to imply a broad experience and understanding. Where
are the links to the spec sheets for these PSUs? Where does he
even try to claim he's done any of the testing/experimentation that
would support his "observations"?

Of course Ken, I don't even have a computer. LOL. I've
long been an advocate of taking temps, and demonstrated
doing so in the past. How many pages did I need to write,
per thread, just to get it to sink in that you need to stop
daydreaming and actually collect valid data about this
system?

You want to dismiss parts heating up, but don't even know
how hot they're getting, instead drifting into random
idealization instead of facts just about everyone knows
already, that these parts DO increase in temp and rely on
the (fan based) cooling subsystem for temp reduction.

The difference in temp is substantial enough that even your
bare finger can feel the difference, even without very
accurate data collection, if you had a mind to actually
compare both ways the difference would be so obvious that
you'd be embarrased to have dismissed it.

Are you seriously trying to argue around what the entire
engineering community considers basic fact?

We do know PSU caps fail on a regular basis, when subjected
to high temps. We likewise know these passive PSU run them
hotter, significantly so. I think you are capable of
googling for passive psu temp measurements, or even taking
your own but you don't want to focus on collecting evidence,
you have continually been the one with the *new* idea that
we can ignore the basic laws of physics because of a water
block on a different part. It's a bit incredible.

You have no validity to your argument until you compare an
optimally fan-cooled system to same parts with passive
(only) cooling, though by passive I mean using water-blocks
on the parts you mentioned. With the apples:apples
comparison isolating that variable, the parts without these
upgraded (waterblock) 'sinks will run hotter. It is
obvious, and demonstrated easily with any system.

What you have done is tried to skip the step of taking
accurate temp readings and comparing with, w/o fans, then
projecting how the elevated temp will effect lifespan
(projection only, since this generation of parts you are
cooling is significantly more current hungry than anything
old enough to draw any conclusions about longevity).

I've never claimed that no system can be passively cooled.
There are systems deliberately designed for lower heat,
lower current consumption that are ideal for this goal.
Then these ideal systems still have the requirement you want
to omit, actually determining the temps of any components
that have the potential for significant rise over ambient.

 

[kony]

Sounds like more results from "Kony's Imaginary Testing Lab",
to me.

I'm still waiting for your temp comparison of the system
this thread is about!

Nice try to change the subject, but the key issue is still
that you have no idea whether the parts in question are
running too hot. Since I don't have your specific system, I
can't very well predict the actual temp, only that they are
hotter, and it is potentially going to degrade lifespan more
than "most" will deem acceptible.

You went off on several tangents but never anything
relevant.

Temperature. Get some good readings with and without fans.
It's all just a silly fantasy without basic facts.

 

[jameshanley39]

Every single one of them runs hot. The remaining question
is how long you need it to work.

how hot is hot. You didn't check, ok.

Every one dies early?

You mention "a few years". Well, what's the normal lifetime of a power
supply? Probably not more than a few years anyway.

People buy new PSUs even without them dying.
Bare in mind that many people buy new computesr every few years
anyway. Either because their MBRD dies, or because it isn't up to date
enough for them.
And when doing that, new computers often use a new power supply
specification.
Around last year I started setting up P4s, they needed the 2x2 thing.
My old power supplies didn't have that. And I got a good deal on a
load of good quality new PSUs that had it.


It's very odd that you say that every single one of them runs hot.
If that is your experience, why do you keep buying them? Or do you get
them for free?
I doubt you fix peoples' computers where the people have fanless power
supplies.
If every single one of them dies, then this would be a serious problem
to look out for. If I found that, i'd have started a thread, giving
exact details.. seen if anybody else had similar experiences.

You seem to be entirely missing the point.

the point is that at least he provided some data. You provided none.

2 days is
nothing, it'll have to work for reasonably close to it's
MTBF rating, or being generous a few years at it's rated
power.

yeah but given what you said, for all we know, they could die in an
instant.
At least now we know somebody had them work for 2 days, it's an
improvement on your data!

<snip>

 

[Ken Maltby]

You are drifting into unrealistic theory. In practice, you
cannot achieve the same cooling with a massive passive
heatsink (exception in next paragraph), because of the
density of the (CPU in this case) heat. It would require
far too thick a copper base to get the heat away from the
core at the required rate, it is not just the total amount
of heat the 'sink can shed in C/W but the spot-temp at the
core-mating surface.

The exception would be a means to remove the heat faster,
like a heat pipe. Even so, we weren't talking about
different 'sinks, rather the same with a fan or not... as it
is with any parts in your system (besides those with a
water-block) having their 'sink as the constant and the
variable being chasssis-induced airflow (or not).

Your "Exception" is the rule for such "Passive CPU
coolers" and more and more for those with a fan as
well. Was it an accident that you snipped out your
MB "laying out in a room"challenge that I was replying
to in my paragraph you quote above. Your phony
challenge would only measure the cooling the passive
cooling heatsink can accomplish with or without fan
blown air, as you admit above. It says nothing
about whether a passive cooler could provide the
required level of cooling, which you have implied it
couldn't.

The only way your statement could possibly make the
slightest bit of sense, would be if you believe that a fan
could/would lower the MB temp. to a point lower than
the temp. of the water in the Watercooling system.
Which again goes to show that you have no experience
or understanding of watercooling, whatsoever. (Not
that this has prevented you discounting the effects of
watercooling the main heat sources within a system,
upon the thermal environment for the other components.)

I wrote "heated" and you'd be silly to ignore this vital
requirement, since it's pretty manditory to transfer that
thermal energy to the water. Heated doesn't have to mean
"hot" on a relative scale of perception based on a body temp
or threshold of pain.




If you have quite high flow rate, cold enough water, or the
top is poorly interfaced to the bottom it could be possible,
but it is hotter than without heat. I have no idea why you
think this distinction is important, you seem to be trying
hard to ignore the important variables.





This is nonsense. It is beyond ridiculous to think
measurement of a heatsink casually fastened across muliple
FETS with the typical poor interface used by a motherboard
manufacturer could be even close to usable data.

If you want to keep trying to make that point, you need to
explain how enough heat can make it through the FETs
packaging to be carried away by the small amount of moving
air that actually crosses the exposed surface, to prevent a
damaging rise in the temp of the parts, but that not enough
heat gets though to indicate higher temps in the part. Or to
be properly cooled by direct contact with a waterblock.

The facts are right in front of you, but you deliberately
refuse to collect them. Same with capacitors, you don't
actually measure and compare both ways, instead dodging and
weaving around the inescapable truth.

No, I can't compare failure rates or reduced lifespan of the
capacitors in my watercooled system, unless and until there
is a failure.

Actually that is my complaint with your observations. You
have obviously never had any watercooling experience. Yet
you keep claiming that what I observe is invalid, that there
are parts running hotter, even though I observe no such thing
(in fact quite the opposite). I have a watercooled "fanless"
system on which to make the comparison. You do not.
In fact I have systems running with the same MB that are
fan cooled as well. But you are not basing your statements
on observed data, scientific measurement, or the mysterious
"Others" results, just on how you think it must work.

Your parts without water blocks are running hotter. We
can't predict (especially without the critical data) if that
is "too" hot for your required lifespan or not but to
pretend there is no temp difference is to want the novelty
of a passive system so badly you deliberately put on
blinders.

As opposed to your "x-ray" vision and ability to measure
the temp. of parts in my watercooled system. The parts in
my watercooled system without waterblocks are no hotter
than those in my fan cooled systems. While I can make
that comparison, you can't; You don't have any water cooled
system to compare to. Not that you believe that should limit
your claims.

I've measured capacitors that were 20C hotter in a passive
system,

What "Passive" system, hotter than what? To a system with
Fans, I take it. If by "Passive" you mean you just removed or
disabled the fans, for some measurements, and just let it then
heat up in a closed box, then I might believe your results. If
I actually could believe you ever took any measurements.

BUT, one thing is sure, you weren't making a comparison
between a Watercooled system (Which YOU don't have)
and one of your fan cooled systems.

but if I provide details of board model, cap brand,
etc, will it be applicable to your system enough that you
wouldn't dismiss it? Doubtful, you need accurate data for
your particular system... but so far, only lots of random
excuses instead of just doing the manditory things any sane
system integrator would do, to accurately measure the spot
temps before concluding an experimental passive cooling
system did as you guessed it does. Scientific method -
requires more than a guess and defense, there's data
collection involved too.

Obviously you believe that applies to me but not to you.
And in any case you will just say my measurements are
meaningless. I have daily observation of two very similar
systems, one with fans and one water cooled, both run
without problems. I even run overnight video encoding
projects with both systems. Of the two systems the
watercooled one has the more thermally demanding
components.

When I can observe the conditions in both systems,
but you have no way of observing a watercooled
system, at all; who is doing the guessing?

Yes I could settle this with a set of thermal photos, but
that's an expense I won't undertake, just to refute a
blowhard like you. The conditions are as I observe, your
pronouncements can't effect the actual temps in my systems
so they mean little or nothing to me. It is too bad that there
may be some swayed to believe your current inaccurate
conclusions, based on your past generally useful postings.

Luck;
Ken

 

[Ken Maltby]

Of course Ken, I don't even have a computer. LOL. I've
long been an advocate of taking temps, and demonstrated
doing so in the past. How many pages did I need to write,
per thread, just to get it to sink in that you need to stop
daydreaming and actually collect valid data about this
system?

You want to dismiss parts heating up, but don't even know
how hot they're getting, instead drifting into random
idealization instead of facts just about everyone knows
already, that these parts DO increase in temp and rely on
the (fan based) cooling subsystem for temp reduction.

What I'm dismissing are your claims that parts on my
watercooled fanless system are any hotter than they are
on my other fan cooled system that uses the exact same
MB and a slightly slower speed CPU. I can easily dismiss
your claims by the simple observation of the two systems.

Parts operate at a certain temperature, based on the wattage
they dissipate. The factors that effect their operating temp.
relate to the voltage applied and current they draw. Yes
they "increase in temp." when they are turned on, they
increase until they reach their operating temp. (Just like a
toaster element, why don't you blow on your toaster so it
will last longer. Or install a fan so the element will not
prematurely fail.) Most parts are designed with packaging
and thermal characteristics that allow them to maintain their
operating temperature without damage, and reject any temp
buildup by simple black box radiation. There are some parts
that, by design, require additional cooling. With these factors
taken into account, there are no parts (that aren't fuses) which
are designed to thermally degrade at their operating temperatures,
within their operating lifetimes. Such parts would be, by
definition, defective.

The difference in temp is substantial enough that even your
bare finger can feel the difference, even without very
accurate data collection, if you had a mind to actually
compare both ways the difference would be so obvious that
you'd be embarrased to have dismissed it.

I, unlike you, can and have made such a very basic
comparison; and observe that where there is no
significant difference in the MB parts that have no active
cooling. Except that a few, such as those near the
watercooled northbridge are cooler on the watercooled
board.

Are you seriously trying to argue around what the entire
engineering community considers basic fact?

Just your claims, as you think they apply to the system I
have described and use.

We do know PSU caps fail on a regular basis, when subjected
to high temps.

External temps. that drastically exceed design specs., not
your mythical operating temp damage over time.

We likewise know these passive PSU run them
hotter, significantly so. I think you are capable of
googling for passive psu temp measurements,

No, lets say I'd like to see the ones you base your
claims on. Surely you can provide some in a day or
too?

or even taking
your own

Yes I can take my own, I have a modded PSU, but
I'm not the one making these claims, I'M not the one
saying "We likewise know these passive PSU run them
hotter, significantly so." YOU are. Not that this would
mean anything anyway if the manufacture designed them
to run hotter, and published a MTBF.

but you don't want to focus on collecting evidence,

I prefer to collect data. You haven't even collected any
specific PSU make and model, yet. You allow us no way
to compare the maker's published data to your claims.
Where did you get the information that "these passive PSU"
run significantly hotter? If "We know" this, maybe "We"
could provide the source of this information?

you have continually been the one with the *new* idea that
we can ignore the basic laws of physics because of a water
block on a different part. It's a bit incredible.

I have presented my observations of working systems
I operate. I made no claims about PSUs or MB parts,
other than those at my disposal and under my observation.
( Unlike your claims.) If you think that my observations
describe conditions that "ignore the basic laws of physics"
I think it more likely that your understanding of the physics
involved is somewhat lacking.

You have no validity to your argument until you compare an
optimally fan-cooled system to same parts with passive
(only) cooling, though by passive I mean using water-blocks
on the parts you mentioned. With the apples:apples
comparison isolating that variable, the parts without these
upgraded (waterblock) 'sinks will run hotter. It is
obvious, and demonstrated easily with any system.

What you have done is tried to skip the step of taking
accurate temp readings and comparing with, w/o fans, then
projecting how the elevated temp will effect lifespan
(projection only, since this generation of parts you are
cooling is significantly more current hungry than anything
old enough to draw any conclusions about longevity).

I've never claimed that no system can be passively cooled.
There are systems deliberately designed for lower heat,
lower current consumption that are ideal for this goal.
Then these ideal systems still have the requirement you want
to omit, actually determining the temps of any components
that have the potential for significant rise over ambient.

Oh, yeah, those "parts that typically fail sooner from heat
relatied stresses", I still haven't seen your listing of such parts
on a modern MB.

 

[k]

I went ahead and put the duct in since it was there. Kind of a pain
but it's not like I will be in there all that often. I'm pretty
impressed by the basic design of the case, it's easy to install stuff
into and can't beat the price.

Thanks Ashton, Ken and GT for your replies!

 

[kony]

Your "Exception" is the rule for such "Passive CPU
coolers" and more and more for those with a fan as
well. Was it an accident that you snipped out your
MB "laying out in a room"challenge that I was replying
to in my paragraph you quote above.

I'm snipping things out because it isn't my system, I'm not
burdened with continually pointing out what you have failed
to address, nor that none of your arguments replace the
obvious requirement of qualifying a cooling subsystem- to
accurately measure ALL parts that heat up to any significant
extent, but apparently argument is a replacement.

With lesser heat retained in the chassis due to water
cooling, you could easily have the same temps with a much
lower chassis airflow, but you wanted to just ignore some of
them so you can arbitrarily claim no fans is ok. OK in the
short term, perhaps - and I never claimed it wasn't viable
for a shorter lifespan.

These are not only issues relevant when one does away with
fans. It's also a regular factor in quiet system cooling
WITH fans, to assess how the changes in the cooling
subsystem effect ALL parts, not just the hottest that have
their own fan, waterblock, etc. ... but you don't want to
consider this, because the magic water block must do
miracles even to parts it doesn't cool.

The only way your statement could possibly make the
slightest bit of sense, would be if you believe that a fan
could/would lower the MB temp. to a point lower than
the temp. of the water in the Watercooling system.

No, it's also considering implementation into a system. The
water in the cooling system isn't necessarily at a higher,
or lower temp than the heated motherboard components (those
not cooled by water blocks). It doesn't necessarily matter
how the water temp relates to the ambient temp immeidately
around the heated parts. Without a fan, the parts
themselves are the heat source, they are creating this
higher ambient temp. With a fan removing heat from the
system, the ambient air temp is already lower, PLUS the
active airflow on the parts, both contribute to even lower
temps on these non-waterblocked parts, and the copper on the
board that is sinking away heat from some of them.

Which again goes to show that you have no experience
or understanding of watercooling, whatsoever. (Not
that this has prevented you discounting the effects of
watercooling the main heat sources within a system,
upon the thermal environment for the other components.)

You still don't get it. We could take the CPU and other
water cooled components out of the system ENTIRELY, have
zero heat produced by them, and remove the waterblocks too
of course, but rig the board so that it passes the same
amount of power to the remaining parts on the board, and an
external (outside the case and far enough away to not
significantly effect room ambient temp) load to replace the
load that was the now-removed parts. The remaining parts in
the system are creating heat, heat that is raising their
immediate ambient temp of board, epoxy casing, parts higher
up on the board if vertical, etc, without active airflow in
the chassis.

Where did you think this heat would go? It builds up,
unless your room fan is so strong it is effectively
replacing a chassis fan, but that's not really a passive
system then, because it still relies on external fan-forced
airflow.

If you want to keep trying to make that point, you need to
explain how enough heat can make it through the FETs
packaging to be carried away by the small amount of moving
air that actually crosses the exposed surface, to prevent a
damaging rise in the temp of the parts, but that not enough
heat gets though to indicate higher temps in the part. Or to
be properly cooled by direct contact with a waterblock.

The primary heat path of the FET is through the copper on
the board, which of course rises in temp based upon how much
airflow it has. The hotter it gets, the lower the thermal
delta and the lesser the rate of heatsinking.

Why you aren't getting an indication of higher temps is as
I'd already mentioned, you can't measure the base of a OEM
slapped-on heatsink draped across a bunch of different SMT
parts, interfaced with a variably effective thermal
interface (depending on who applied it, quality of interface
material present, etc) then all this on top of the
non-primary heat path, through the encapsulating epoxy.

Direct, *good* contact (DIY, not just accepting a factory
heatsink installation) with the sink or waterblock will
help, certainly some of the heat is being removed through
that epoxy, and it is good to keep it from getting too hot,
but nevertheless it is not the primary heatsinking path, it
is far lower efficiency. It would be a bit like putting a
single video card heatsink on the back of the video card
instead of the front.

No, I can't compare failure rates or reduced lifespan of the
capacitors in my watercooled system, unless and until there
is a failure.

Actually that is my complaint with your observations. You
have obviously never had any watercooling experience. Yet
you keep claiming that what I observe is invalid, that there
are parts running hotter, even though I observe no such thing
(in fact quite the opposite).

Then where did the heat go? The parts you dismiss aren't
producing less heat. The heat produced doesn't disappear,
it remains in the system for longer through a less effective
thermal path out of the chassis when there is no airflow,
resulting in rising chassis temp. If this is not the
situation then you have the case far more open than you have
described and the room fan blowing quite a bit into it. If
that is the situation and you are content with it, ok, but
how is it really a passive system if it has this room
airflow requirement?

What "Passive" system, hotter than what? To a system with
Fans, I take it. If by "Passive" you mean you just removed or
disabled the fans, for some measurements, and just let it then
heat up in a closed box, then I might believe your results. If
I actually could believe you ever took any measurements.

BUT, one thing is sure, you weren't making a comparison
between a Watercooled system (Which YOU don't have)
and one of your fan cooled systems.

Actually I was, you have somehow made another wild leap to
support your argument, because if I even start to supply
information you can't accept it- which is why I had already
suggested your measurements were necessary. But enough time
spent, if you can't resolve that heat doesn't vanish from
parts consuming current, that it is removed slower without
airflow, then all the direct measurements of capacitors or
chassis temp in the world won't matter because you reject
anything that doesn't agree with your preconceived notion.

If I had meant "just pull all fans", it would have been a
dissimilar situation and I wouldn't have bothered to mention
the caps being hotter. Perhaps I should have described the
scenario at length, but it would not be directly applicable
as distance to other parts (particularly the FETs) have a
direct bearing on cap temp in a fanless system, and your
board is not laid out the same... it is close enough to be a
factor, but it was only one example, but your parts do
create heat and do need the same types of measurements but
instead you continually just try to think through it with a
dismissal attitude instead of seeing there is more to it
than just 3 or 4 components having waterblocks.

I think we have both wasted enough time on this thread. So
I opt-out.

 

[kony]

how hot is hot. You didn't check, ok.

Every one dies early?

You mention "a few years". Well, what's the normal lifetime of a power
supply? Probably not more than a few years anyway.

This is a distinction, yes I meant long term use, not just a
few days of runtime. Just like you wouldn't be able to
predict the second your car might seize up if you redline it
continuously till it fails, I can't put a day, week, month
on when any given passive psu will fail, there are too many
variables to do this. I would expect most to fail within 3
years if ran at their rated max load, but any other system
is another variable among many.

We can instead see from all reviews that the internal temps
are higher, and it is established fact that components do
fail at an accelerated rate. These passive units do have
some unique 'sinking but they're not substantially different
when it comes down to discrete part specs.

A good PSU will last over a decade. I have quite a few that
have, AT PSU that run fine still in addition to early ATX.
What have we seen from other PSU running hotter? Failure in
months or a couple years, usually.

By the above (and below) I am considering one scenario only,
use of a passive PSU in a (either very close to, or
entirely) passively cooled case, no fan-forced airflow
through the PSU. Part of the reason for this is due to a
problem inherant with the passive psu, that most cases have
the rear exhaust fan which would result in air intake
through the back of the PSU, the heat from the PSU being
dumped into the case, then expelled out the back and this
again drawn up and back into the PSU for another cycle. If
someone where to mount only intake fans on the side and/or
front of the case, these passive PSU would fare (and effect
the system they're in) far more positively.

People buy new PSUs even without them dying.

Yes, that is one possibility, but does it mean the old one
was thrown away? Generally not if it was decent, it either
remains with the system as a whole or is reused. Certainly
there are exceptions but we have to at least consider the
typical situation, that a PSU is bought for a system and
stays with that system - then eventually it becomes the
first failure point (odds are), unless the motherboard had a
particular flaw like defective capacitors.

When considering an old 80486, it can't do the basic things
people expect - realtime video, audio, modern video
resolutions for typical 2D work with enough pixels that
everything doesn't look blocky in a windowed environment.

Now contrast that with what a Core2Duo and XXXX will be able
to do 10 years from now... it will take some less common
activity to make such a system undesirable for most common
uses, unless there are some revolutionary breakthroughs on
the next so-called "killer app", like maybe virtual reality,
but frankly I don't see it becoming more than a reason to
upgrade a video card in the next few years we'd hope a PSU
would last.

Bare in mind that many people buy new computesr every few years
anyway. Either because their MBRD dies, or because it isn't up to date
enough for them.

They don't generally throw away the old one, it still needs
a PSU. I wish I remembered where I saw it, but there is a
site online that did a survey about average system age, and
the average age is now close to 5 years... meaning a
substantial number of systems older than that. What was
available 4 years ago, in 2003?

nForce2 and Athlon XP,
http://www.tomshardware.com/2003/02/14/more_nforce2/
3GHz Pentium 4,
http://www.tomshardware.com/2003/02/03/pentium_4_with_dual_ddr/page20.html

Frankly I think the survey I saw was off, I'm suspecting
the average systems belong to people not so exposed to
online surveys and are now older than 4 years. As
importantly, I know a LOT of people who when their system
did fail, had no desire to do anything except get their
present system working again as cheaply as (reasonably)
possible, if it wasn't their PSU that failed they had value
in it and wanted to continue using it.

Regardless, surely you know someone who would gladly accept,
even pay money for a 2+GHz Pentium 4 system? It's going to
need a working PSU? There is no real justification to the
idea that it's ok if parts die, they should always last
until the last person who would want to use a system, gives
up on it. That's going to be a long, long time for anything
built today, it will be a part failure causing a typical
system to be thrown away if not just laziness of the owner
to not make it available to anyone who might want it.

Making parts with shorter lifespans ultimately results in
increased garbage, unnecessary addition to landfills,
manufacturing pollution. Even when an original owner doesnt
want a product anymore, it is better that it still work.

And when doing that, new computers often use a new power supply
specification.

So you'd just throw away all old parts? Seems a bit of a
waste, unless you were assuming the parts died- the opposite
of what I advocate, to make choices that result in a system
continuing to work and have a potential purpose to
"somebody". Systems fast enough to run WinXP without
severe pain for email, office, web browsing, are now 8 years
old. That's all a lot of people use their systems for.

Around last year I started setting up P4s, they needed the 2x2 thing.
My old power supplies didn't have that. And I got a good deal on a
load of good quality new PSUs that had it.

I was buying PSU with 2x2 connectors well over 4 years ago,
in the Pentium 3 era but I don't recall the first... but one
of the earlier might've been an Antec PP303XP. It only had
15A on the 12V rail, but this was before CPUs came near even
100W TDP, which with a roughly 92% board VRM subcircuit
efficiency, would be under 9A, plenty of current left for a
handfull of fans and a couple drives.

What has changed? Mostly 12V current, if your older PSU had
enough you didn't need a new PSU, just an inexpensive
adapter. Soon thereafter we saw 400W+ PSU with 18-20A on
12V rails, and so on.

It's very odd that you say that every single one of them runs hot.

Not so odd, what did you think the purpose of the fan was if
not to remove that heat? "Hot" is relative though, I didn't
mean you can use it to fry eggs (but actually, you probably
could come close in some of them).

If that is your experience, why do you keep buying them? Or do you get
them for free?

I get most of my non-preferred parts (for free) after they
fail, but with the newer breeds of PSU you can find online
reviews where the passive unit reviews do have particular
attention paid to internal temp, as it is a pretty obvious
factor when the PSU is so costly and (fortunately) derated
so it looks better than most of a similar wattage that had
fans.

I doubt you fix peoples' computers where the people have fanless power
supplies.
If every single one of them dies, then this would be a serious problem
to look out for. If I found that, i'd have started a thread, giving
exact details.. seen if anybody else had similar experiences.

If you have a hot skillet on the stove and I told you that
every single egg you put in it will fry, do I have to cook
every single egg you have before you believe it?

Cause and effect. You are thinking in terms of unknown
variables on a PC computer component, while I am considering
known facts about parts lifespan, in particular electrolytic
caps... especially because the solid alternatives are so
incredibly costly in high capacitance values, I keep hoping
one of these very expensive passsively cooled PSU will be
built befitting it's high price, as a case with a lot of
holes or screens in it, $5 worth of upgraded parts and a
couple extended heatpipe sinks (if that) can't account for
the added retail cost.

So if you were able to keep track of all owners of a given
passive PSU, yes it would be fair to say that on average,
they have a significantly shorter lifespan in number of
years, because they are not keeping the shortest lived parts
as cool, not substituting the substantially higher priced
parts that would fare better over the long term, only what
is required for immediately stable operation and hopefully
the duration of a warranty (but given the high price, it's
bound to be factoring in a potentially higher RMA rate
during the warranty period as well).

the point is that at least he provided some data. You provided none.

It's not necessarily my burden to hunt down information you
could find for yourself, unless it's something really
obscure but with any model PSU you might consider, you have
the applicable search terms and can seek a reviewer who has
taken measurements but as importantly, is a 3rd party
without a bias. Then comes experience and context, seeing
the non-defective cap failures from elevated temp.

You have no idea how many bad caps I've thrown away from
junk brought to me. Not "defective" caps, just not the
right part for the environment it was in so it's too hot,
from ambient condition and/or ESR too high. Maybe you have
seen a lot of failed caps? If so, can we agree that heat
kills them? If we can't agree on that, I suggest you head
over the badcaps.com and wander around for awhile, or read
the major cap manufacturers spec sheets and technical
guides.

There is nothing that changes the situation in a passive
PSU, unless they start building them with more appropriate
parts. If you have one you are considering, by all means do
look up the specs on the parts so you can see whether
appropriate changes were made.

yeah but given what you said, for all we know, they could die in an
instant.
At least now we know somebody had them work for 2 days, it's an
improvement on your data!

Not an instant, I expect that the majority will last a
couple years, but it's a large variable how positively or
negatively pressurized a case is from other fans (if
present), how heated that case is if there aren't any fans
in it, the room ambient temp, system load average as well as
peaks, hours of on-time. Just as every specimen of any
brand and model won't die at the same time, we can expect
some of the passive units to live quite a bit longer than
others, and never have any overheat-shutdown problems in the
interim that reviewers have caused.

Ultimately I am suggesting that you read some reviews if you
hadn't already, but also to consider that running a PSU for
a few hours is not any qualification of fitness for longer
term use. Resist trusting a reviewer's tendency to try to
judge fitness (especially considering I had never claimed
you couldn't run a system for a shorter period from one) and
just look at data, and ratings. A PSU has to live up to
them, that's the whole point of ratings, and if it does, how
long it will last.

If you really feel you would end up throwing away a power
supply after 4 years, maybe a passive unit is a reasonable
choice for you, if you are sure your system load on it is
low enough. It isnt really necessary though, any PSU can
have the fan running so slow it is not easy to hear if
higher temps are acceptible.

 

[Ken Maltby]

On 14 Feb 2007 16:19:21 -0800, "(e-mail address removed)"

To save the Internet bandwidth I will cut up and summarize
Kony's rambling circular arguments, that make it a 13KB,
post. His whole post is there for any willing to slog through it.

We can instead see from all reviews that the internal temps
are higher, and it is established fact that components do
fail at an accelerated rate.

Again, I'd like to read some of the specific reviews you
claim to have read. The ones you claim as a source for
your statements. Surely if "all reviews" are your quoted
source, you can provide ONE for us to check out?

What was
available 4 years ago, in 2003?

nForce2 and Athlon XP,
http://www.tomshardware.com/2003/02/14/more_nforce2/
3GHz Pentium 4,
http://www.tomshardware.com/2003/02/03/pentium_4_with_dual_ddr/page20.html

See you can post a link to a "review" (it has nothing to do with the
issues raised in this thread, but it is evidence that you know how to
do it.)

I get most of my non-preferred parts (for free) after they
fail, but with the newer breeds of PSU you can find online
reviews where the passive unit reviews do have particular
attention paid to internal temp, as it is a pretty obvious
factor when the PSU is so costly and (fortunately) derated
so it looks better than most of a similar wattage that had
fans.

We could find reviews, but we want to see the ones you keep
claiming as support for your statements. It is apparent the links
I posted provided you with some understanding of the "Derating"
concept. So can't you post some, I mean you must have kept
at least a few of the URLs, of these "all reviews".

So if you were able to keep track of all owners of a given
passive PSU, yes it would be fair to say that on average,
they have a significantly shorter lifespan in number of
years, because they are not keeping the shortest lived parts
as cool, not substituting the substantially higher priced
parts that would fare better over the long term, only what
is required for immediately stable operation and hopefully
the duration of a warranty (but given the high price, it's
bound to be factoring in a potentially higher RMA rate
during the warranty period as well).

Don't you get dizzy arguing in circles like that?

It's not necessarily my burden to hunt down information you
could find for yourself,

Unless, you are trying to make it prop up your arguments.

A PSU has to live up to
them, that's the whole point of ratings, and if it does, how
long it will last.

Hey, does that mean you finally agree that if a PSU maker
rates a PSU for 120,000hrs it should operate for at least
those rated hours - NO MATTER IF IT USES FANS or
NOT? (or runs hotter than any other PSU, or not? Not
that I'm saying a passively cooled PSU needs to run hotter,
much less so hot as to cause any reduced lifespan.)


I must of read too much of Kony's post, I need to go
make a cup of coffee.

Luck;
Ken